Meteora

METEOROLOGY

METEOROLOGY, in its original and etymological sense, included within its scope all appearances of the sky, astronomical as well as atmospherical, but the term is now restricted to the description and explanation of the phenomena of the atmosphere which may be conveniently grouped under weather and climate. These phenomena relate to the action of the forces on which the variations of pressure, temperature, humidity, and electricity of the atmosphere depend, but in an especial sense to the aerial movements which necessarily result from these variations.

In the more exact development of meteorology, the scientific investigation of climate long preceded that of weather. Humboldt’s work on Isothermal Lines, published in 1817, must be regarded as the first great contribution to meteorological science. The importance of this inquiry into the distribution of terrestrial temperature it is scarcely possible to overestimate, for, though the isothermals were necessarily to a considerable extent hypothetical, there cannot be a doubt that they presented a first sketch of the principal climates of the globe. Dove continued and extended the investigation, and in his great work On the Distribution of Heat on the Surface of the Globe, published in 1852, gave charts showing the mean temperature of the world for each month and for the year, together with charts of abnormal temperature. To this, more than to any other work, belongs the merit of having popularized the science of meteorology in the best sense, by enlisting in its service troops of observers in all parts of the civilized world.

In 1868 another series of important charts were published representing by isobaric lines the distribution of the mass of the earth’s atmosphere, and by arrows the prevailing winds over the globe for the months and the year. By these charts the movements of the atmosphere and the immediate causes of these movements were for the first time approximately stated, and some knowledge was thereby attained of some of the more difficult problems of meteorology. It was shown that the prevailing winds are the simple result of the relative distribution of the mass of the earth’s atmosphere, in other words, of the relative distribution of its pressure, the direction and force of the prevailing winds being simply the flow of the air from a region of higher towards a region of lower pressure, or from where there is a surplus to where there is a deficiency of air. It is on this broad and vital principle that meteorology rests, which is found to be of universal application throughout the science, in explanation, not only of prevailing winds, but of all winds, and of weather and weather changes generally. One of the more important uses of the principle is in its furnishing the key to the climates of the different regions of the earth; for climate is practically determined by the temperature and moisture of the air, and these in their turn are dependent on the prevailing winds, which are charged with the temperature and moisture of the regions they have traversed. The isobaric charts show further that the distribution of the mass of the earth’s atmosphere depends on the geographical distribution of land and water in their relations to the sun’s heat and to radiation towards the regions of space in different seasons.

In 1882 Loomis published a map showing the mean rainfall of the globe. This map and others that have been constructed for separate countries show conclusively that the rainfall of any region is determined by the prevailing winds considered in relation to regions from which they have come, and the physical configuration and temperature of the part of the earth’s surface over which they blow. The maximum rainfall is precipitated by winds which, having traversed a large breadth of ocean, come up against and blow over a mountainous ridge lying across their path, and the amount deposited is still further increased if the winds pass at the same time through regions the temperature of which constantly becomes colder. On the other hand, the rainfall is unusually small, or nil, when the prevailing winds have not previously traversed some extent of ocean, but have crossed a mountain ridge and advance at the same time into lower latitudes, or regions the temperature of which is markedly higher.

While the observational data for the determination of the geographical distribution of the prime elements of climate, viz., the pressure, temperature, moisture, and movements of the atmosphere and the rainfall were being slowly but surely collected, the great importance of the study of weather came gradually to be recognized. Additional impetus was given to this branch of study from its intimate bearings on the eminently practical question of storm warnings. Synchronous weather maps, showing the weather over a considerable portion of the earth’s surface, were constructed, and some advance was made in tracing the progress of storms from day to day. Unquestionably one of the first problems of meteorology is to ascertain the course storms usually follow and the causes by which that course is determined, so as to deduce from the meteorological phenomena observed, not only the certain approach of a storm, but also the particular course that storm will take. The method of practically conducting this large inquiry in the most effective manner was devised by the genius of Leverrier, and begun to be carried out in 1858 by the daily publication of the Bulletin International, to which a weather map was added in September 1863. This map showed graphically for the morning of the day of publication the atmospheric pressure, and the direction and force of the wind, together with tables of temperature, rainfall, cloud, and sea disturbance from a large number of places in all parts of Europe. From such weather maps forecasts of storms are framed and suitable warnings issued; but above all a body of information in a very handy form is being collected, the careful study and discussion of which is slowly but gradually leading to the issue of more exact and satisfactory forecasts of weather, and to a j uster knowledge of these great atmospheric movements which form the groundwork of the science.

The most cursory glance is sufficient to show that the ever-changing physical phenomena with which it is the business of meteorology to deal are all referable to the action of the sun, it being evident that if the sun were blotted out from the sky a cold lifeless uniformity would rapidly take possession of the whole surface of the globe. Meteorological phenomena naturally group themselves into two great classes,—those dependent on the revolution of the earth on its axis, and those dependent on its revolution round the sun taken in connexion with the inclination of its axis to the plane of its orbit. The science thus divides itself into two great divisions, the first comprising diurnal phenomena and the second annual phenome Diurnal March of Phenomena. Temperature.— Of the daily changes which take place in the atmosphere, the first place must be assigned to those which relate to temperature, seeing that on these all other changes are either directly or indirectly dependent. Observations of the temperature of the air are therefore of the first importance in meteorology. A perfectly accurate observation of the temperature of the air is unquestionably among the most difficult to make of all physical observations, the difficulty being to eliminate the effects of radiation of surrounding objects. The nearest approach yet made to the solution of this important problem of physical inquiry was made by Dr Joule in a communication to the Philosophical Society of Manchester (November 26, 1867, Proc., vol. vii. p. 35). But the manipulative skill and time demanded by the method there detailed render it quite unsuitable for general adoption anywhere in collecting the observational data required in the determination of this important element of climate. It is therefore necessary to fall on some method which, while it gives results that can only be regarded as approximate, secures the essential element of uniformity among the observations.

Fig. 1 represents Stevenson’s louvre-boarded box for the thermometers, which is now very widely used for temperature observations. The box is made of wood, and louvred all round so as to protect the thermometers inside from radiation, and at the same time secure as free a circulation of air as is consistent with a satisfactory protection from radiation. The box is painted white, both inside and outside, and screwed to four stout wooden posts, also painted white, firmly fixed in the ground. The posts are of such a length that when the thermometers are hung in position the bulbs of the minimum thermometer and hygrometer are exactly at the same height of 4 feet above the ground, the maximum thermometer being hung immediately above the minimum thermometer. This thermometer box is placed over a plot of grass, and in a free open space to which the sun’s rays have free access during as much of the day as surrounding conditions admit of. It will be observed that the thermometers are suspended on cross-laths in the centre of the box and face the door, which should always open to the north. It is not possible to overestimate the importance of seeing that uniformity of height above ground and method of protecting the thermometers is secured, since in no other way is it possible to obtain results from different places which shall be comparable with each other and thus supply satisfactory materials for the investigation and development of comparative climatology.

A desired uniformity is yet far from being attained among the meteorological systems of different countries. Thus in Russia the box for the protection of the thermometers is made of zinc, on the supposition that such a box follows more closely the changes of temperature of the air than a box of wood. Owing to these international diversities of observation, it is extremely desirable that steps were taken to ascertain, by Joule’s method of observing, the approximate errors peculiar to each sort of thermometer box, in order that the temperatures of different countries may be compared together in a more satisfactory manner than has yet been possible.

Interchanges of temperature among bodies take place by conduction, convection, and radiation. In meteorology the most important illustrations of conduction are the propagation downwards through the earth’s strata of the changes of the temperature of the surface as it is heated during the day and cooled during the night, and the propagation of the same changes of temperature through the lowest stratum of the air which rests on the surface. Since sand and light loose soils are much worse conductors of heat than clay and dense soils, it follows that loose soils and tracts of sand are subject during the day to higher temperature and during the night to lower temperature near the surface than dense soils, and that frosts and extreme temperatures do not penetrate so far into loose as into dense soils. It is on these differences that some of the more striking features of climates depend. As snow is one of the worst conductors of heat, owing to the quantity of air filling the interstices among the ice crystals, it protects the soil it covers by setting a limit to the depth to which the severe frosts of the surface penetrate, and by arresting the escape of the heat of the soil upwards to the air.

The communication of heat from one part of the earth to another by convection is seen on a grand scale in the winds and in the currents of the ocean. It is seen also in the ascending and descending currents of the atmosphere everywhere, which have their origin in the daily and unequal changes of temperature to which the surface of the earth is subject. The direct and beneficial effect which results from atmospheric and oceanic circulation is a more equable distribution of temperature over the globe, thus moderating the rigours of the polar regions and the heat of the tropics.

An interchange of heat is constantly going on among bodies exposed to each other, whatever be their temperature. This mode by which heat is communicated from one body to another is called radiation. Radiant heat proceeds in straight lines, diverging in all directions from the source, is only in a limited degree influenced by the air through which it passes, and is not diverted from the straight course by the wind. The intensity is proportional to the temperature of the source, and is greater according to the degree of inclination of the surface on which the rays fall.

If then a body be placed in the presence of other bodies, some colder and some warmer than itself, it will from this mutual interchange of temperature receive more heat from the warmer bodies than it radiates to them, and consequently becomes warmer; but it will receive less heat from the colder bodies than it radiates to them, and its temperature consequently falls. This is precisely the condition in which the earth is placed in space. When a part of the surface is turned towards the sun, that part of the surface receives more heat than is radiated from it; and the temperature consequently rises most in that region which for the time is perpendicular to the sun’s rays, and least round the annulus where the inclination of the surface is greatest. On the other hand, since the hemisphere turned from the sun radiates more heat than it receives from the cold regions of space, the temperature there falls. Owing to the essentially distinct conditions under which the earth is placed with respect to radiation, the subject falls naturally to be divided into two heads, solar radiation and terrestrial radiation.

Solar Radiation.— Of the sun’s rays which arrive at the earth’s surface, those which fall on the land and solid bodies generally are wholly absorbed by the thin surface layer exposed to the heating rays, the temperature of which consequently rises. Whilst the temperature of the surface increases, a wave of heat is propagated downwards through the soil. The intensity of the daily wave of temperature rapidly lessens with the depth at a rate depending on the conductivity of the soil, until at about 4 feet below the surface it ceases to be measurable. Part of the heat of the surface layer is conveyed upwards through the air by the convection currents which have their origin in the heating of the lowermost stratum of air in direct contact with the heated surface of the land.

Altogether different is the influence of the sun’s rays on water. In this case the sun’s heat is not all, indeed very far from all, arrested at the surface, but penetrates to a considerable depth. The depth to which the influence of the sun is felt has been shown by the observations made during the cruise of the “Challenger” to be, roughly speaking, about 500 feet below the surface of the sea. The rate at which, in perfectly clear water, this heat is distributed at different depths is a problem that has not yet been worked out. Since water is a bad conductor, the heat thus distributed does not, as takes place with respect to land, penetrate to still lower depths by conduction, but only by different densities prevailing at the same depths, whether these different densities be due to different temperatures or different degrees of salinity. Thus one of the more important distinctions between land and water surfaces in their bearings on climate is that nearly all the sun’s heat falling on land is arrested on the surface, whereas on water it is at once diffused downwards to a great depth. In examining temperatures of the sea taken at different depths, it is surprising to note the rapidity with which changes of temperature are felt at considerable depths, especially in cases when the temperature of the air rises rapidly, accompanied with strong sunshine.

In shallow water the sun’s heat raises the temperature much higher than that of deep water, this being obvious from the consideration that nearly the whole of the sun’s heat which falls on the surface is utilized in raising the temperature of the shallow layer of water; in other words, it is, so to speak, concentrated through a small depth of water instead of being diffused through a great depth.

Surface Temperature of the Sea.— The importance of a knowledge of this datum of meteorology will be at once recognized when it is kept in view that three-fourths of the earth’s surface is water, that the temperature of the air resting on this surface is in close relation to the temperature of the surface, and that the latter has, through the intervention of the winds, direct and important bearings on the temperature of the air over large portions of the land surfaces of the globe. During the years 1859-63 Captain Thomas, while engaged on the survey of the islands on the north-west of Scotland, made observations of the temperature of the surface of the sea every hour of the day at all seasons, and with sufficient frequency for the determination of the diurnal range of the temperature of the surface. The daily minimum, 0°*17 below the mean, occurred near 6 a. Μ.; the mean was reached about 11 A.Μ., the maximum, 0 o, 13 above the mean, between 3 and 4 p.m., and the mean again shortly before 2 a.m. Thus the daily oscillation of the temperature of the surface of the sea amounted on the north-west of Scotland only to 0° ’ 3. In lower latitudes the amount of the daily fluctuation is somewhat larger, but everywhere it is comparatively small, if care be taken to make the observations properly, or at a distance from land, where the influence of the heated or cooled land is not allowed to vitiate the results.

During the voyage of the “Challenger” a complete system of meteorological observations, including the temperature of the surface of the sea, was made every two hours as part of the scientific work of the cruise. These are now being discussed, and the writer of this article is, by permission of the Lords Commissioners of H.M. Treasury, allowed to use such of the results as have been already arrived at.

The diurnal march of the temperature of the surface of the North Atlantic has been determined from observations made on one hundred and twenty-six days from March to August 1873 and in April and May 1876, the mean latitude of all the points of observation being nearly 30 N., and the longitude 42° W. The following variations from the mean show the phases of this diurnal oscillation :— ,

[table]

Thus in mid Atlantic, about 30° N. lat., where the sun’s heat is strong, and at the time of the year when the sun is north of the equator, the diurnal fluctuation of the temperature of the surface is only 0° '80. It is highly probable that nowhere over the ocean does the mean daily fluctuation of the temperature of the surface quite amount to a degree. This small daily fluctuation is a prime factor in meteorology, particularly in discussions relating to atmospheric pressure and winds.

Temperature of Air over the Open Sea.— The following shows the daily march of the temperature of the air over the North Atlantic on a mean of the same one hundred and twenty-six days for which the temperature of the sea has been given:— [table]

The amplitude of the daily fluctuation of the air is thus 3° '21, or nearly four times greater than that of the sea over which it lies. During the same months the “Challenger” was lying near land on seventy-six days. The observations made on these days show a greater daily range of temperature of the air than occurred out in the open sea. The minimum, - 2° O5, occurred at 4 a.m., and the maximum, 2° '33, at noon, thus giving a daily range of 4°'38. The occurrence of the maximum so early as noon is doubtless occasioned by the greater strength of the sea breeze after this hour, this maintaining a lower temperature. Part of the increased range of the temperature of the air as compared with that of the sea was no doubt due to the higher temperature during the day and the lower during the night on the deck of the “Challenger” as compared with that of the air. But, after making allowance for this disturbing influence, it is certain that the temperature of the air has a considerably larger daily range than that of the sea on which it rests. The point is one of no small interest in atmospheric physics from the important bearings of the subject on the relations of the air and its aqueous vapour to solar and terrestrial radiation.

The hourly deviations from the mean daily temperature of the air at two places, one near the equator and the other in the north temperate zone, and both near the sea, viz., Batavia (6° 8' S. lat., 106° 48' E. long., mean temperature 78 β '7) and Rothesay (55° 50' N. lat., 5° 4' W. long., mean temperature 47° '3), are these :— [table]

The times of the four phases of the daily temperature at Batavia are—minimum about 5.50 a.m., mean 8.45 a.m., maximum 1.20 p.m., and mean 7.40 p.m.; while for Rothesay the same times are 4.30 a.m., 9.15 a.m., 3 p.m., and 8.20 p.m. At Batavia, where the days and nights are nearly equal during the year, there is little variation in these times through the months; but at Rothesay, where the days are much longer in summer than in winter, there is considerable variation in the times of occurrence of these phases. The following table shows the times of the phases for a number of selected places in the northern hemisphere for the two extreme months, January and July :— [table]

During the night in summer the temperature falls continuously from the effects of terrestrial radiation till the earliest dawn, when the daily rise in the temperature sets in owing to the heat reflected from the upper strata of the atmosphere, which have begun to be heated and lighted up by the rays of the morning sun. It will be observed that the time of the daily minimum temperature occurs earliest in high latitudes and latest in low latitudes. During winter, on the other hand, the minimum temperature takes place in several regions some time before dawn. At this season the two chief causes on which changes of temperature depend are the sun and the passage of cyclones and anticyclones; and it is probable that those cases where the minimum occurs markedly before the dawn are, where not occasioned by purely local disturbing causes, due to the mean diurnal times of occurrence of the changes of temperature which accompany the great atmospheric disturbances of cyclones and anticyclones.

In July the daily maximum temperature occurs generally from 2 to 4 p.m. At places, however, near the sea, which are within the immediate influence of the sea breeze, and in places at some distance from the sea, such as Calcutta, where the wind, being essentially a sea wind, attains its greatest daily velocity and the sky at the same time is much clouded, the maximum occurs nearly two hours earlier. In high situations, such as the St Bernard hospice, the highest daily temperature also occurs nearly two hours sooner than on the plains below. In the winter months the maximum is about an hour earlier than in the summer.

In investigating the daily curves of temperature, Sir David Brewster drew several interesting conclusions from them. By dividing the daily curve of temperature, deduced from the mean of the year, into four portions, at the points representing the two daily means and the two extremes, he showed that the four portions approximate to parabolas, in which the temperatures are the abscissae and the hours the ordinates. The correspondence between the observed and calculated results is so close that the difference did not in any case exceed a quarter of a degree Fahrenheit. This interesting result is true for places at which the horizon is open all round, so that no shadows of hills, trees, or buildings fall on the places where the thermometers are kept during the day. If a hill rises to the north of the place, by which the sun’s rays are never obstructed, it exercises little, if any, influence on the observations; but if one or more hills obstruct the rays of the sun after it has risen above the horizon, such obstruction affects the temperature while, and for some time after, the position in which the thermometer is placed is shaded from the sun.

Brewster further made the important remark that the mean of observations made at any pair of hours of the same name, such as 8 a.m. and 8 p.m., 9 a.m. and 9 p.m., &c., does not differ much from the mean temperature of the day. The pairs of hours which approximate closest to the true daily mean are 9 a.m. and 9 p.m., 10 a.m. and 10 p.M., 3 a.m. and 3 p.m., and 4 a.m. and 4 p.m. The mean of four hours at equal intervals from each other gives a result still closer to the true mean temperature.

In organizing any system of meteorological observation, by which it is intended to develop the climatology of the country, the determination of the hours of observation is a question of the first importance. If only two observations be made daily the best hours are 9 a.m. and 9 p.m., or 10 a.μ. and 10 p.m. ; and if there be four observations the best hours are 3 and 9 a. μ. and 3 and 9 p.m., or 4 and 10 a.m. and 4 and 10 P.M. If there be three observations the best hours are 9 a.m. and 3 and 9 p.m., or 10 a.m. and 4 and-10 p.m. ; but in these cases it is essential that the observations of a minimum thermometer be added to the temperature observations. These hours are further strongly recommended by the consideration that they are approximately coincident with the diurnal phases of atmospheric pressure, an exact knowledge of which lies at the root of nearly all climatological inquiries. The three equidistant hours which have been adopted in several countries, viz., 6 A.Μ. and 2 and 10 p.m., are only good as regards the temperature, not as regards atmospheric pressure. With respect to two daily observations, the hours 8 A. Μ. and 1 p.m., which have been adopted in some countries, are singularly unsuitable for the furnishing of the observational data required in the development of the climatologies of these countries; and, what is still more serious in a science where international co-operation is so imperatively demanded, these observations cannot be used with any satisfaction in such deeply important inquiries as the comparative climatologies of Europe.

The times of occurrence of the highest, lowest, and mean daily temperatures, and the amount of the daily range of temperature, are in a great degree influenced by the covering or want of covering of the earth’s surface on which the air rests. When the ground is covered with vegetation, the whole of the solar heat falls on the vegetable covering; and, as none falls immediately on the soil, its temperature does not rise so high as happens where there is no vegetable covering to shade the surface from the sun. The temperature of plants exposed to the sun is not so high as that of exposed soil in the vicinity. As regards forests, the four diurnal phases of temperature occur later than in the open country, and the maximum and minimum are less decided; and, since the maximum temperature of the air in forests falls short of the maximum in the open to a considerably greater extent than the minimum under trees is above the minimum in the open, it follows that the mean temperature of the air in forests is less than that of the open country adjoining. The reason of the difference is that the chilling effects of nocturnal radiation penetrate lower down among the trees than do the heating effects of solar radiation; and as the soil is not heated directly by the sun its temperature is lower, and consequently that of the air over it is also lower. A cleared space in a forest, sheltered by the surrounding trees, but open to the sun, has a warmer and moister atmosphere in spring and summer and very much moister in autumn than prevails in the open country adjoining, and has also the diurnal differences of range peculiar to a warmer and moister atmosphere.

One of the most important elements of climate is disclosed by the difference between the hour of lowest and the hour of highest mean temperature respectively, or, as it is usually expressed, by the daily range of temperature. We have seen that as regards the sea in the north-west of Scotland the difference is only 0° '3 and in the Atlantic about 30° N. lat. 0 o- 8, and that probably the diurnal range of temperature of the surface of the sea nowhere amounts to a degree. In the same part of the Atlantic the daily range of the temperature of the air resting on the ocean is 3 0, 2, and on the sea near land 4 o- 4. On advancing on the land, the daily range of temperature rapidly increases, and the rate of increase is greatly augmented when an inland position is arrived at to which any sea breezes that may prevail do not extend.

The true daily range of temperature is stated by observations made with maximum and minimum thermometers. Generally speaking, the amount of the range increases as the latitude is diminished, and as the distance from the sea is increased, but above all it increases in proportion to the dryness of the climate.

The differences of this vital element of climate are strikingly shown in the meteorology of India. In the Report for 1880 the following are the mean daily ranges of March of that year at a few places: at Goa 5 o- 4, Bombay 1Γ 5, 2, Kurrachee 23 o, 5, Jacobabad 37 o- 4, and Pachbudra (lat. 25° 55' N., long. 72° 18' E.) 41° 3. In the last case, undoubtedly one of the greatest mean daily ranges of temperature meteorology has yet recorded, the mean of the days was 103 o- 4 and of the nights 62 o- l. As March is altogether within the season of the north-east monsoon, the general drift of the wind over western India, where these are situated, is from the interior towards the sea, subject as regards Bombay and Goa to the influences of the land and the sea breeze. On the other hand, in June, when the south-west monsoon has fairly set in, the following are the mean daily ranges of temperature at the same places: at Goa 5 0, 6, Bombay 8 0, 2, Kurrachee 10°, Jacobabad 27° , 6, and Pachbudra 24 o, l. These show in a striking manner the powerful influence of the moister atmosphere spread over India by the south-west monsoon, under which the daily range of temperature falls at Kurrachee from 23 o- 5 to 10°, and in the excessively arid climate of Pachbudra from 41 o, 3 to 24 o, l. In these dry climates of the basin of the Indus, whilst the rainfall both in March and in June is practically nil, yet the relative humidity of the atmosphere is widely different. Thus the humidities for March and June respectively at 4 p.m., when the temperature is nearly the maximum for the day, are 48 and 77 for Kurrachee, 18 and 30 for Jacobabad, and 11 and 36 for Pachbudra. It is not so much the amount of cloud that determines the degree of fierceness of the sun’s heat in these climates as the relative humidity, or the dryness of the air, as pointed out by Strachey in 1866. Thus at Jacobabad less than half the amount of cloud appears in the sky in June as compared with March, but the relative humidities are 30 and 18, and the daily range of temperature 27° ’ 6 and 37 0- 4. If we except the dry arid wastes of Persia and Arabia, there is perhaps no other region of the globe where the daily range of temperature approaches that of the valley of the Indus. Thus in the dry climates of such places as Sacramento (California) in summer it amounts only to about 30°, at Madrid to 27°, and Jerusalem 24°. In central districts in the south of England it is about 20°; farther north it falls to 15°; and in the islands in the north, whose climate is strictly insular in its character, the summer daily range is only 10°. In Arctic regions, such as Spitzbergen and Boothia Felix, the range in winter varies from 0° O to l o O; in May, when the sun has reappeared and continues to rise and set, it rises to 14°; but in July, when the sun does not set, the range sinks to 10°.

But maximum and minimum thermometers not only show the mean daily range of temperature, they are also of great utility in giving observations for the determination of mean temperature. The mean temperature may be accepted as the mean of the twenty-four hourly observations of the day. If with such a system of observation daily readings of the maximum and minimum thermometer be compared, the value of the latter observations in questions of mean temperature may be arrived at. Double series of observations of this description have been made at many places. The following shows a comparison of the mean of maximum and minimum daily temperatures with means from observations made twenty-four times daily, the former exceeding the latter means in nearly all cases :—

[table]

These results show remarkable uniformity, and it may be inferred from them that mean temperatures deduced from maximum and minimum observations are about half a degree above the true mean temperature. In general climatological inquiries, observations with these thermometers have the strong recommendation of supplying from observations taken once a day the data for the determination of the mean temperature and mean daily range of localities ; to which falls to be added the further advantage of giving results more uniformly comparable for different places than could be afforded by observations made with a common thermometer at any single hour or pair of hours daily.

Daily Variation of the Humidity of the Air.— The gaseous envelope surrounding the earth is composed of two atmospheres, quite distinct from each other,—an atmosphere of dry air and an atmosphere of aqueous vapour. The dry air, which consists of oxygen and nitrogen, is always a gas, and its quantity remains constant; but the aqueous vapour does not continue permanently in the gaseous state, and the quantity present in the air is, by the ceaseless processes of evaporation and condensation, constantly changing. If the aqueous vapour remained permanently and unchanged in the atmosphere, or were not liable to be condensed into cloud or rain, the mixture would become as complete as that of the oxygen and nitrogen of the air. The equilibrium of the vapour atmosphere, however, is being constantly disturbed by every change of temperature, by every instance of condensation, and by the unceasing process of evaporation. Since dry air further materially obstructs the free diffusion of the aqueous vapour, it follows that the law of the independent pressure of the vapour and of the dry air of the atmosphere holds good only approximately. The aqueous vapour, however, constantly tends to approach this state. Since, then, the independent and equal diffusion of the dry air and the aqueous vapour is, owing to these disturbing causes, never reached, the important conclusion follows that the hygrometer can never indicate more than the local humidity of the place where it is observed. Hygrometric observations can therefore be regarded only as approximations to a true indication of the quantity of aqueous vapour in the atmosphere over the place of observation. It is, however, to be added that, while in certain cases the amount of vapour indicated is far from the truth, yet in averages, particularly long averages, a close approximation to the real amount is reached if the hygrometer be at all tolerably well exposed and carefully observed.

Aqueous vapour is constantly being added to the air from the surfaces of water, snow, and ice, from moist surfaces, and from plants. The rate of evaporation increases with an increase of temperature, because the capacity of the air for vapour is thereby increased. The atmosphere can contain only a certain definite amount of vapour, according to the temperature; when therefore the air has its full complement of vapour, or when, in other words, it is saturated, evaporation ceases. Thus the rate of evaporation is greatest when the air is driest or freest from vapour, and least when the air is nearest the point of saturation. Since currents of air remove the moister and substitute drier air over the evaporating surfaces, evaporation is much more rapid during wind than in calm weather. As air expands under a diminished pressure, its temperature consequently falls, and it continues to approach nearer to the point of saturation, or become moister; and, as it contracts under an increased pressure, its temperature rises and it recedes from the point of saturation or becomes drier. Hence ascending currents of air become moister with every addition to the ascent, and descending currents drier as they continue to descend. Thus as winds ascend the slopes of hills they become moister, but when they have crossed the summit and flow down the other side they become drier in proportion to the descent, and all the changes may be experienced from extreme dryness to saturation in the same mass of air, which all the time has practically had its amount of aqueous vapour neither added to nor diminished.

In an atmosphere of air and aqueous vapour perfectly mixed, the elastic force of each at the surface of the earth is the pressure of each. In this case the elastic force of aqueous vapour would be the pressure of the whole vapour in the atmosphere over the place of observation. This pressure is expressed in inches of mercury of the barometer. If we suppose the total barometric pressure to be 30Ό00 inches, and the elastic force of vapour to be 0’745 inch, the pressure or weight of the dry air, or air proper, would be 29 ’ 255 inches, and of the aqueous vapour 0’745 inch. From this it follows that the elastic force of vapour may be regarded as indicating the quantity of aqueous vapour in the air at the place of observation, or it may be designated the absolute humidity of the air.

The diurnal variation in the elastic force of vapour in the air is seen in its simplest form on the open sea. Grouping together all the hygrometric observations made on board the “Challenger” on the North Atlantic at a distance from land, from March to July 1873 (eighty-four days), we have for that time a mean elastic force of 0’659 inch, and the following diurnal variation :— [table]

Hence the minimum ( - Ό20 inch) occurs at the hour when the temperature of the surface of the sea and air resting over it falls to the daily minimum; it then rises to the mean a little after 9 a.m., and to the daily maximum ( + Ό20 inch) at 2 p.m., when the sea and air are also near the daily maximum, and falls to the mean shortly before 9 P.M.

Treating the observations made near land by the “Challenger” during the same months, the following is the diurnal variation disclosed :— [table]

The disturbance induced by proximity to land in the distribution of the aqueous vapour in the lower strata of the atmosphere is very striking. The maximum and minimum no longer follow the corresponding phases of the temperature of the surface of the sea and of the air. The disturbing agents are the sea and land breezes and their effects. Under the influence of the land breeze the time of the minimum humidity is delayed till about 6 A.M.; and under the influence of the sea breeze and its effects the amount of the aqueous vapour shows a secondary minimum from noon to 2 p.m. It is to be here noted that this midday minimum occurs at the hours of the day when the surface of the land is most highly heated, the ascending current of heated air rising from it therefore strongest, and the resulting breeze from the sea towards the land, also strongest. Now it does not admit of a doubt that the diminution in the amount of the aqueous vapour noted on board the “Challenger” near the shore points to an intermixture with the air forming the sea breeze of descending thin air-filaments or currents to supply the place of the masses of air removed by the ascending currents which rise from the heated surface of the land. At Batavia, on the north coast of Java, and at Bombay, the aqueous vapour is also subject to a secondary minimum during the warmest hours of the day.

During the summer months this secondary minimum is best marked at inland places such as Peking, Nertchinsk, Barnaul, Tiflis, and Ekaterinburg, but the time of its occurrence is about two hours later than it is over the North Atlantic. Over all these places at this season the ascending current from the heated land in the interior of Asia is very strong. On the other hand the lowering of the amount of aqueous vapour scarcely if at all appears as a feature in the summer climate of St Petersburg, and not at all in that of Sitka, where the sea breeze is equally not a constant feature of the climate of the district.

In the excessively dry, rainless, and hot climate of Allahabad, in April the diurnal minimum of the aqueous vapour occurs from 11 A. Μ. to 6 p.M., the time of absolute minimum being 2 and 3 p. Μ. During all other hours of the day the amount of the vapour is above the mean, a secondary minimum occurring from 1 to 4 A. Μ. At Allahabad, at this time, the absolute maximum vapour pressure occurs at 8 A.Μ. Quite similar to this is the diurnal distribution of the aqueous vapour in July at Lisbon and Coimbra, the minimum occurring from 10 A. Μ. to 8 p.M. At this time of the year the climate of this part of the peninsula is hot and dry and the rainfall insignificant in amount. As this region lies between the high atmospheric pressure so characteristic a feature of the meteorology of the Atlantic in summer and the comparatively low pressure over the continents southward and eastward, the winds are almost wholly north-westerly. In this connexion it is instructive to note that the time of maximum vapour pressure is from 4 to 7 A. Μ., when the velocity of the wind is near the minimum, and the chief minimum vapour pressure from noon to 4 p.m., when the velocity of the wind and ascending currents reach the daily maximum. These results show that the diminution in the vapour pressure during the hours when temperature is highest, which characterizes the climates of large tracts of the globe, is due to descending air-filaments or currents, which necessarily accompany the ascending currents that rise from the heated land.

At Geneva during the summer months the vapour curve exhibits two daily minima very strongly marked, the one shortly before sunrise and the other from 2 to 4 p.m., and two maxima, one from 8 to 11 a.m. and the other from 6 to 10 p.m. ; and with these the diurnal variations of cloud are in accordance. The peculiarly marked features of the vapour curve at Geneva are probably due to the size of the lake, which is large enough to give rise to a decided breeze during the day from the lake all round its shores and during the night to a breeze from the land all round upon the lake. On the setting in of the breeze, the mass of air composing it, having been for some time resting on the lake, is rather moist, and thus one of the daily maxima is brought about from 8 to 11 A. Μ. As the breeze continues the air supplying it is necessarily drawn from the higher strata of the atmosphere more copiously than in different situations; and, having thus acquired increased dryness in the descent, and having blown over the lake for too short a distance to materially influence its moisture, the air becomes constantly drier, till the minimum from 2 to 4 p.m. is reached. The lake breeze thereafter begins to diminish in force, and the air consequently becomes moister till the maximum vapour pressure of the day occurs when the lake breeze dies away and the land breeze has not yet sprung up. In the winter months, when these breezes do not prevail, the curve of diurnal vapour pressure shows only one maximum and minimum.

The relative humidity of the atmosphere must not be confounded with its vapour pressure or absolute humidity. The relative humidity, or, as it is more frequently called, the humidity, of the air is the degree of its approach to saturation. Complete saturation is represented by 100 and air absolutely free of vapour by 0, the latter state of things never occurring in the atmosphere, a humidity of 10 being of rare occurrence even in such arid regions as those of Arabia. The great significance of this element of climate is in its relations to the diathermancy of the air, and consequently to solar and terrestrial radiation. It is supposed that perfectly dry air would allow rays of heat to pass through it with at most only a very slight increase to its temperature therefrom. Let, however, a little aqueous vapour be added to it, a partial obstruction to the passage of radiant heat is offered, and the temperature of the mixture, or common air, is sensibly raised. Hence, other things being equal, the less the amount of vapour the more are the effects of radiation felt, or the greater the heat of the days and the cold of the nights. The mere amount of vapour in the air does not determine the degree of radiation, but it is the amount of vapour together with a certain temperature—in other words, the absolute and relative humidity of the air taken together— that determines the heating power of the sun and the degree of cold produced by terrestrial radiation.

The diurnal variation of the relative humidity is very different from that of the vapour pressure, and presents features of the simplest character. The following are the diurnal variations from the mean humidity 80 over the North Atlantic, from the “Challenger” observations in 1873 [table]

Thus the maximum humidity occurs from midnight to 4 A.M., or when the daily temperature is at the minimum, and the minimum humidity at 2 p.m., when the temperature is at the maximum, the curve of humidity being thus inverse to that of the temperature. With two slight modifications this is the diurnal humidity curve for all climates and seasons. In the calm which intervenes in the morning between the land and the sea breeze the humidity continues high, or even increases, though at the time the diurnal increase of temperature has already set in. The other modification is seen in the humidity curves for Nertchinsk and Barnaul during winter, these curves being not inverse but coincident with the daily curves of temperature. In the climates of Central Asia in winter, the amount of vapour is very small, and the increase to the relative humidity during the day is probably occasioned by the more active evaporation from the snow during the day and the stillness of the air favouring the accumulation of aqueous vapour near the surface of the earth.

Next to the winds, the aqueous vapour of the atmosphere, in the diverse ways in which in different localities it is distributed through the hours of the day, plays the most important part in giving to the different parts of the globe its infinitely diversified climates.

Dew.— Dew is deposited over the earth’s surface on comparatively clear and calm nights. As the cooling by terrestrial radiation continues, the temperature of objects on the surface is gradually lowered to the dew-point, and when this point is reached the aqueous vapour begins to be condensed into dew on their surfaces. The quantity deposited is in proportion to the degree of cold produced and the quantity of vapour in the air. Dew is not deposited in cloudy weather, because clouds obstruct the escape of heat by radiation, nor in windy weather, because wind continually renews the air in contact with the surface, thus preventing the temperature from falling sufficiently low. When the temperature is below 32', dew freezes as it is deposited, and hoar-frost is produced. The dew-point practically determines the minimum temperature.

of the night,—because if the temperature falls a little below the dew-point the liberation of heat as the vapour is condensed into dew speedily raises it, and if it rises higher the loss of heat by radiation speedily lowers it. This consideration suggests an important practical use of the hygrometer, it being evident that by ascertaining the dew-point the approach of frost or low temperature likely to injure vegetation may be foreseen and provided against.

Diurnal Oscillations of the Barometer.— The general character of the daily oscillations of atmospheric pressure is shown by the two curves of fig. 2. The solid line gives the mean oscillation for Bombay and the dotted line that for Vienna, these two curves being to a large extent typical of diurnal barometric oscillations in tropical and temperate regions as regards the two maxima and minima and the time of their occurrence.

A series of twelve maps of the globe were prepared for June, showing, for all stations whence observations have been obtained, the deviations at noon, 2 p.m., 4 p.m., &c., Greenwich mean time, from the mean daily pressure; and four lines were drawn indicating the positions of the two daily maxima and minima at these hours. For fully 30° north and south of the equator the lines of maxima and minima run north and south, but in higher latitudes these directions are changed, and the changes are chiefly conspicuous as regards the a.m. maximum and the p.m. minimum. Thus, for example, at 6 p.m. (G. Μ. T.) the line of p.m. minimum is for the latitude of London near 16° W. long.; in 30° N. lat. it is in 35° W. long., in which the line runs south as far as 30° S. lat.; its course thence turns south-westwards to near the Falkland Islands, 60° W. long. Hence in June the p.m. minimum occurs about three hours earlier in the Falkland Islands than to the south-west of Ireland, thus showing in a striking manner the influence of season on this phenomenon. In the middle and higher latitudes in summer, proximity to the sea delays the time of occurrence of the a.m. maximum and the P.M. minimum; whilst in continental situations the A.M. maximum occurs much earlier than in lower latitudes, and the p.m. minimum nearly as late as at places near the sea. In cases where the lines of maxima and minima cross a region such as southern and western Europe, whose surface is diversified by large tracts of land and sheets of water, the deflexions are of a remarkable character.

The retardation of the time of occurrence of the a.m. maximum is greatest in situations which, while eminently insular in character, are at the same time not far from an extensive tract of land. Of this Holland presents the best example in Europe; and there the a.m. maximum, which at Paris occurs at 8 A.Μ., does not occur at Utrecht till 9.30 A.Μ., at Amsterdam till 12.30 p.m., and at Helder till 2 p.M. There is thus as regards the same diurnal phenomenon in June a difference of six hours between Paris and Helder. Sicily and the south of Italy on the one hand and Madrid on the other present also the most striking contrasts. Again at Sifka (56° 50' N. lat., 135° W. long.), which has one of the most truly insular climates in the world, the a.μ. maximum is delayed to 2.30 p.m. ; whereas at Astoria, ten degrees to southward, it occurs at 9.30 A.Μ., and at Fort Churchill, in California, as early as 7 A. μ. There is thus as regards the same phenomenon a difference of 7 h 30 m between Sitka and Fort Churchill.

From hourly observations made in this month at the base, the top, and two intermediate points on Mount Washington (Ν'. H.) it was found that the time of occurrence of the a.μ. maximum at the base of the mountain, which is 2898 feet above the sea, was 8 a.m. ; at 4059 feet, 10 A.M.; at 5533 feet, 11 a.m. ; and at the top, 6285 feet, noon. Hence, as regards the time of occurrence, the influence of an isolated mountain like Mount Washington brings about a result similar to what is observed in insular situations. But the analogy is even closer. In insular climates the minimum in the early morning is very greatly in excess of that in the afternoon; and the same relation is observed on the top of Mount Washington, where the former is -0'020 inch and the latter -0'004 inch. Again in continental climates the minimum in the early morning is much the smaller of the two, and the same relation was observed at the base of the mountain, where the observed minima were respectively 0Ό06 inch and 0Ό20 inch. The differences presented by the daily curve of pressure at the top as compared with that at the base of the mountain have their explanation in the effects which follow the diurnal range of temperature. As the temperature is at the minimum at the time of least pressure in the morning, the atmosphere is more condensed in the stratum between the base and the top, and consequently the barometer at the top reads relatively lower. As the temperature continues to rise during the day, the stratum of air above the base of the mountain expands, thus placing more air above the barometer at the top, so that, while at the base pressure begins to fall at 8 a.μ., at the top it continues to rise till noon, simply from the mechanical upheaval of the air owing to the higher temperature. In the afternoon, when the minimum at the base falls to - 0Ό20 inch, it is only - 0'004 inch at the top, this relatively higher pressure at the top being due to expansion from temperature. The peculiar feature of the pressure curve at the top is essentially a temperature effect.

The diurnal oscillations of the barometer occur alike over the open sea and over the land surfaces of the globe. The atmosphere over the open sea, as already shown, rests on a floor or surface subject to a diurnal range of temperature so small as to render that temperature practically a constant both day and night. This consideration leads to the vital and all-important conclusion that the diurnal oscillations of the barometer are not caused by the heating and cooling of the earth’s surface by solar and terrestrial radiation and by the effects which follow these diurnal changes in the temperature of the surface, but that they are primarily caused by the direct and immediate heating by solar radiation, and cooling by nocturnal radiation to the cold regions of space, of the molecules of the air, and of its aqueous vapour. These changes of temperature are instantly communicated through the whole atmosphere from its lowermost stratum resting on the earth’s surface to the extreme limit of the atmosphere, which the flight of meteors proves to be not less than 500 miles. There are important modifications affecting the amplitude and times of occurrence of the four prominent phases of the phenomena observed over land surfaces, the temperature of which is being superheated during the day and cooled during the night; but it is particularly to be noted that the barometric oscillations themselves are independent of any changes of temperature of the floor on which the atmosphere rests.

Let us first look at the phenomena in the simplest form as found in the Pacific, or in the midst of the largest water surface of the globe. The following are the mean variations of pressure from observations made on board the “Challenger,” September 1 to 12, 1875, in mean latitude l 0 8' S. and long. 150° 40' W, the mean being 29*928 inches :— [table]

The most striking feature in these oscillations is the amplitude of the range from the a.m. maximum to the p.M. minimum, amounting to 0Ό87 inch, and the rapidity of the fall from 10 a.m. to 2 p.m. The same feature appears in all means deduced from observations made at least 12° on each side of the equator.

From October 12 to 22, 1875, in mean lat. 35° Γ S., long. 134° 35' W., the mean atmospheric pressure was 30*298 inches, and the difference between the A. Μ. maximum and the p.M. minimum was only 0*036 inch ; and from July 12 to 19, 1875, in mean lat. 36° 16' N. and long. 156° 11' W., the mean pressure was 30*328 inches, and the difference between the a.m. maximum and p.m. minimum was only 0*025 inch. Thus, with a mean pressure in the Pacific about lat. 35°-36° N. and S. much greater than near the equator, the oscillation is much less, being in the North Pacific less than a third of what occurs near the equator. Similarly, this oscillation is small (or even smaller) in the high-pressure areas in the North and South Atlantic as compared with the same oscillation near the equator.

It is well known that aqueous vapour absorbs the heat rays of the sun considerably more than does the dry air of the atmosphere; how much more physicists have not yet accurately determined. Consequently air heavily charged with aqueous vapour will be heated directly by the sun’s rays as they pass through it in a greater degree than comparatively dry air is. Now it is shown further on that the prevailing surface winds outflow in every direction from the areas of high mean pressure in the Atlantic and Pacific about lat. 36° N. and S. Since, notwithstanding, the pressure continues high, it necessarily follows that the high pressure is maintained by an inflow of upper currents, and as the slow descending movement of the air connects the inflowing upper currents with the outflowing prevailing winds of the surface, it follows that the air over high-pressure areas is very dry, and that it is driest where pressure is highest and the high-pressure area best defined. Hence over the best-defined anticyclonic regions the air will be least raised in temperature through all its height by the heat rays of the sun.

On the other hand, between these high-pressure areas of the great oceans there is a belt of comparatively low pressure towards which the north and south trades pour their* vapour unceasingly. The atmosphere of this belt of low pressure is thus highly saturated with aqueous vapour which rises in a vast ascending stream of moist air to the higher regions of the atmosphere. These equatorial regions thus present to the sun a highly saturated atmosphere reaching to a very great height. It is in these regions therefore that the atmosphere will be most highly heated by the sun’s heat rays as they pass through it. One of the most striking facts of meteorology is the suddenness with which this barometric oscillation increases in amplitude on entering on these parts of intertropical regions; and the rapidity with which its amplitude diminishes on advancing on the high-pressure regions of the horse latitudes is equally striking. The following are the mean oscillations in the middle regions of the four great oceans about lat. 36° from the a.m. maximum to the p.m. minimum about the time of the year in each case when the sun is highest in the heavens :—South Pacific, 0*036 inch;

North Pacific, 0*025 inch; South Atlantic, 0*024 inch; and North Atlantic, 0*014 inch. These amplitudes diminish as the ocean becomes more land-locked with continents, or as the anticyclonic region becomes better defined and currents of air are poured down more steadily from the higher regions of the atmosphere.

If the temperature of the whole of the earth’s atmosphere were raised, atmospheric pressure would be diminished, for the simple reason that the mass of the atmosphere would thereby be removed to a greater distance from the earth’s centre of gravity. Quite different results, however, would follow if the temperature of only a section of the earth’s atmosphere were simultaneously raised, such as the section comprised between long. 20° and 60° W. The immediate effect would be an increase of barometric pressure, owing to expansion from the higher temperature; and a subsequent effect would be the setting in of an ascending current more or less powerful, according to the differences between the temperature of the heated section and that of the air on each side. These are essentially the conditions under which the morning maximum and afternoon minimum of atmospheric pressure take place.

The earth makes a complete revolution round its axis in twenty-four hours, and in the same brief interval the double-crested and double-troughed atmospheric diurnal tide makes a complete circuit of the globe. The whole of the diurnal phenomenon of the atmospheric tides is therefore rapidly propagated over the surface of the earth from east to west, the movement being most rapid in equatorial regions, and there the amplitude of the oscillations is greater than in higher latitudes under similar atmospheric, astronomical, and geographical conditions. Owing to the rapidity of the diurnal heating of the atmosphere by the sun through its whole height, some time elapses before the higher expansive force called into play by the increase of temperature can counteract the vertical and lateral resistance it meets from the inertia and viscosity of the air. Till this resistance is overcome, the barometer continues to rise, not because the mass of atmosphere overhead is increased, but because a higher temperature has increased the tension or pressure. When the resistance has been overcome, an ascending current of the warm air sets in, the tension begins to be reduced, and the barometer falls and continues to fall till the afternoon minimum is reached. Thus the forenoon maximum and afternoon minimum are simply a temperature effect, the amplitude of the oscillation being determined by latitude, the quantity of aqueous vapour overhead, and the sun’s place in the sky.

All observations show that over the ocean, latitude for latitude, the amplitude of the oscillations is greater in an atmosphere highly charged with aqueous vapour and less in a dry atmosphere. It is also to be noted that in very elevated situations, particularly in tropical regions, the amplitude is greater proportionally to the whole pressure than at lower levels. This is what is to be expected from the law of radiant heat by which more of the heat rays of the sun is absorbed by the air, and particularly by its aqueous vapour, mass for mass, in the higher than in the lower strata.

When the daily maximum temperature is past, and the temperature has begun to fall, the air becomes more condensed in the lower strata, and pressure consequently at great heights is lowered. Owing to this lower pressure in the upper regions of the air, the ascending current which rises from the longitudes where at the time the afternoon pressure is low flows back to eastward, thus increasing the pressure over those longitudes where the temperature is now falling. This atmospheric quasi-tidal movement occasions the p.m. increase of pressure, which reaches the maximum from 9 p.m. to midnight, according to latitude and geographical position. This maximum is therefore caused by accessions to the mass of the atmosphere overhead, contributed by the ascending currents from the longitudes of the afternoon low pressure immediately to westward.

As midnight and the early hours of morning advance, these contributions become less and less and at length cease altogether, and pressure continues steadily to fall. But between the time when the increase of pressure from the overflow through the upper regions of the atmosphere ceases and the time when pressure increases from the heat rays, direct or indirect, of the returning sun, or during the hours of the night when the effects of nocturnal radiation are the maximum, pressure is still further reduced from another cause. Radiation towards the cold regions of space takes place, not only from the surface of the globe, but also directly from the molecules of the air and its aqueous vapour. The effect of this simultaneous cooling of the atmosphere through its whole height is necessarily a diminution of its tension. Since this takes place at a more rapid rate than can be compensated for by any mechanical or tidal movement of the atmosphere from the regions adjoining, owing to the inertia and viscosity of the air, pressure continues to fall to the morning minimum. This minimum is thus due, not to the removal of any of the mass of air overhead, as happens in the case of the afternoon minimum, but to a reduction of the tension or pressure of the air consequent upon a reduction in the temperature through radiation from the aerial molecules towards the cold regions of space. In the open ocean the morning minimum is largest in the equatorial regions, and it diminishes with latitude; but the rate of diminution with latitude through anticyclonic and other regions is generally less and more uniform than in the case of the afternoon minimum.

The amplitude and times of occurrence of the phases of the diurnal barometric tides are subject to great modifications over the land. The amplitude of the oscillation from the morning maximum to the afternoon minimum is greatest where the atmosphere is driest and the sky clearest, and least where the atmosphere is highly saturated and the sky more frequently and densely covered with clouds, being thus generally the reverse of what is observed to take place over the open sea. The meteorology of India affords the most striking illustrations of this remark. At Bombay in April during the dry atmosphere and clear skies of the north-east monsoon, the oscillation is 0*118 inch; but in July during the humid atmosphere and clouded skies of the south-west monsoon it falls to 0Ό67 inch. In the Punjab, where the air is drier, it is much greater, rising in exceptional years, such as 1852, to 0 ’ 187 inch. The much greater amplitude of this oscillation on land as compared with the open sea is entirely due to the heating of the earth. By this heating of the surface the lower strata of the air become also highly heated and the tension is increased; and, since the air does not expand freely, vertically and laterally, from its inertia and viscosity, the barometer rises. When, however, the resistance is overcome, the ascending current which sets in is stronger owing to its higher temperature. Since this higher temperature which has its origin in the superheated surface is in addition to the direct heating of the air by the heat rays of the sun as they pass through it, the morning maximum and the afternoon minimum over land are both more extreme than over the open sea. It follows that this oscillation is much larger over land, and largest in climates where insolation is strongest.

In places already referred to where the morning maximum is greatly retarded, such as Helder, Sitka, Valentia, and Falmouth, the afternoon minimum in the summer months is singularly small,—so small indeed that it does not fall so low as the mean pressure of the day. This peculiarity in the diurnal barometric tide is in all probability due to their insular position to the westward of a more or less extensive tract of land, by which a tidal overflow is propagated through the upper regions from the continental towards the insular situations. This tidal overflow receives its impulse from the ascending current from the land, which rises sooner and stronger from inland than from insular situations. On the other hand, on the open sea, and away from land in regions where the morning maximum and afternoon minimum are both small, the minimum always falls below the mean of the day, and the time of occurrence of the maximum is not retarded as is the case in insular situations. A map of deviations from the daily mean pressure of the morning minimum in summer shows, as regards the middle and higher latitudes, that it is greatest near the sea, and least in inland continental situations. Indeed in the interior of the Old-World continent the dip in the curve in the early morning is so small that the minimum does not fall below the daily mean pressure, but at most places remains considerably above it. The same relations are seen in north-western Europe, where the morning minimum is - 0Ό20 inch at Valentia and Falmouth, - 0Ό18 inch at Helder, and-0O12 inch at Amsterdam, whilst at Kew it is only - 0Ό02 inch. From its compact form and relations to the surrounding ocean, the Spanish Peninsula well illustrates the peculiarities of this phase of the pressure. The deviations from the daily mean pressure of the morning minimum are at Lisbon -0Ό22 inch and Coimbra - 0Ό11 inch, but at Madrid in the interior + 0Ό09 inch,—pressure in the last case, just as happens in the interior of Asia, not falling so low as the daily mean.

The larger minimum near the sea arises from the higher temperature there during the night as compared with more inland situations, from which results a tidal overflow through the upper regions from the sea towards the land, as the temperature of the latter falls lower than the sea during the night. The effect of this overflow is to reduce the pressure over those regions whence it proceeds and to increase it in those regions over which it advances. The shallowing of the morning minimum is greatest in the higher latitudes of continental climates and most complete at great elevations, where in some cases the minimum vanishes,—in other words, where the amount of aqueous vapour is small and the time is short during which no part of the atmosphere overhead is touched by the sun’s rays. Since the peculiarity is observable in the curves over nearly the whole continent, appearing even in the low latitudes of Calcutta and Madras, it might be suggested whether we have not evidence here of a vast tidal movement propagated through the higher regions towards that trough-like section of the atmosphere as it moves westwards over the continent where the temperature of the lower strata of the air is about the minimum of the day and pressure also about the minimum.

Reference has been made under Atmosphere to the smallness of the range from the a. μ. maximum to the p.m. minimum in the North Atlantic during summer. This phase in the diurnal distribution of pressure is represented in fig. 3, which shows for June the mean amount of the oscillation by lines of 10, 20, 40, 60, 80, and 100 thousandths of an inch, or 0Ό10 inch, 0Ό20 inch, &c. This abnormality begins in March, attains the maximum in June, and terminates in October. It is thus confined to the warmer months of the year, and, unlike most meteorological phenomena, is not cumulative, but follows the sun, so that its maximum occurs in June, and not in July as that of the temperature of the air, or in August as the temperature of the sea. The smallness of this range over the North Atlantic, which is less than occurs in any other ocean in the same latitudes, is to a large extent caused by the small dip in the diurnal curve of the afternoon minimum.

If the map of the distribution of pressure over the globe for July be examined (fig. 17) it is seen that this part of the Atlantic is occupied by a well-defined area of high mean pressure,—higher indeed than occurs at any season over any ocean; and it is shown below that out of this area the surface winds blow in all directions. But, since air is constantly being drained out of this region by the wind without diminishing the pressure, it follows of necessity that the high pressure must be maintained by accessions of air received from above through the upper currents. Now the regions whence such accessions can come are the upper currents which have their origin in the ascending currents that rise from the heated plains of Africa, Europe, the belt of calms, and the two Americas surrounding the North Atlantic. It is evident that the major portion of each day’s overflow of air from the continents through the upper regions of the air upon the Atlantic, whether this overflow takes place by convection currents or from a tidal movement similar to what has been already described, will take place during mid afternoon. In other words, the overflow will occur about the time of the afternoon minimum of the Atlantic, thus diminishing the dip of this minimum, and so producing the abnormally small range now under examination. It is in favour of this view that the abnormality follows the sun’s course and is not cumulative, and is felt also on both sides of the Atlantic, even although the weather on the east side is dry and all but rainless, and on the west moderately moist and characterized by a rather copious rainfall. It is also full of significance that the peculiarity is most strikingly seen in that part of the ocean of the globe which is closely hemmed in by large masses of land.

Influence of the Moon on Atmospheric Pressure.— Fifteen years’ hourly observations have been made at Batavia and discussed by the late Dr Bergsma in their relation to the lunar day, which was assumed in the calculations to commence with the time of the upper transit of the moon. The result of the inquiry is that atmospheric pressure at Batavia has a lunar tide quite as distinctly marked as the ordinary diurnal barometric tide, except that its amplitude is much less. The four phases are these :— [table]

The lunar tide has the important difference that its phases follow the moon’s apparent course much more' closely than the ordinary diurnal fluctuations of the barometer follow that of the sun. The two maxima occur about the 1st and 13th, and the two minima about the 7th and 19th, whereas these four daily phases of the diurnal barometric fluctuation occur with respect to the sun’s apparent course from one to six hours later. It is interesting to note that in the higher latitudes in inland situations during winter, or at times and in situations where the disturbing influences of temperature and humidity tend towards a minimum, the times of occurrence of the four phases of the daily oscillation of the barometer approximate to those of the daily lunar atmospheric tide.

Since a distinct lunar tide is traced to the attractive influence of the moon, it follows that the-attractive influence of the sun will enter as one of the several causes which determine the phases and amplitude of the diurnal barometric curve. It also follows from the much less attractive influence of the sun than that of the moon on the earth’s atmosphere that the effects of the sun’s attraction on the pressure will be wholly concealed by the much larger effects of the other forces concerned in determining the diurnal oscillation, except in the case or cases where the variation in the fluctuation is small at 1 and 7 a.m. and 1 and 7 p.m. Now at places north of lat. 45° N. the variation at 1 A.Μ. is small during the winter, and it is a singular fact that some years ago Rykatchew of St Petersburg drew the attention of meteorologists to the existence at these northern stations of a faintly marked third maximum; and it is further of importance to remark that, at many places where on the mean of years the third maximum is scarcely or not at all marked, it appears in the mean of some of the separate years. Thus, though it does not appear in the mean of the twenty years ending 1873 at Greenwich for January, it appears in nine of the individual years. It is highly probable that this maximum, which may be named. Rykatchew’s maximum from its discoverer, is due to the attractive influence of the sun, its amplitude and time of occurrence being in accordance with such a supposition.

Diurnal Variation of the Force of the Wind.— During the three and a half years’ cruise of the “Challenger,” ending with May 1876, observations of the force and direction of the wind were made on 1202 days, at least twelve times each day,—650 of the days being on the open sea and 552 near land. The observations of force were made on Beaufort’s scale 0-12, being the scale of windforce observed at sea. The mean diurnal force of the wind on the open sea and near land respectively is shown in fig. 4, where the figures on the left are Beaufort’s scale, and those on the right the equivalents in miles per hour. The solid line shows at the different hours of the day the mean force on the open sea, and the dotted line the mean force near land.

As regards the open sea it is seen that the diurnal variation is exceedingly small, there being two apparent slight maxima, about midday and midnight respectively. On examining, however, the separate means for the North and South Atlantic, North and South Pacific, and the Southern Ocean, there is no uniform agreement observable among their curves, the slight variations which are met with being different in each case. It follows therefore that the force of the winds on the open sea is subject to no distinct and uniform diurnal variation. The difference between the hour of least and greatest mean force is less than a mile per hour.

Quite different is it, however, with the winds encountered by the “Challenger” near land, the force of the wind there giving a curve as pronouncedly marked as the ordinary diurnal curve of temperature. The minimum occurs at 2 to 4 a.m. and the maximum from noon to 4 p.M., the absolute highest being at 2 p.m. The curves constructed for each of the five oceans from the observations near land give one and the same result, or a curve closely accordant with the curve of diurnal temperature. The differences between the hours of least and greatest force are as follows :—Southern Ocean 6⅜ miles, South Pacific 4∣ miles, South Atlantic 3∣ miles, and North Atlantic and North Pacific 3 miles per hour.

In the case of each ocean the velocity of the wind on the open sea is considerably in excess of that near land, but in no case does the maximum velocity near land, attained about midday, reach the velocity of the wind on the open sea. The 650 daily observations on the open sea give a mean hourly velocity of 17⅜ miles, whereas the 552 near land give a velocity of only 12∣ miles per hour. The difference is greatest at 4 a.m., when it amounts to upwards of 6 miles an hour, but is diminished by the rising temperature till at 2 p.M. it is less than 3 miles an hour.

At Mauritius, which is situated within the south-east trades, the minimum velocity of the wind is 9 - 7 miles per hour, occurring from 2 to 3 A.Μ., from which it rises to the maximum 18 - 5 miles from 1 to 2 p. Μ., the influence of the sun being thus to double the wind’s velocity. At Batavia, situated in a region where the mean barometric gradient is much smaller, the differences are still more decided. From 1 to 6 A. Μ. 85 per cent, of the whole of the observations are calms, whereas from noon to 2 p. Μ. only 1 per cent, are calms. In all months, the minimum velocity occurs in the early morning, when the temperature is lowest, and the maximum from 1 to 3 p.Μ., when the temperature is highest, the mean minimum and maximum velocities being to each other as 1 to 21. At Coimbra the mean maximum hourly velocity is five times greater than the minimum hourly velocity in summer, whereas in winter it is only about a half more. At Valentia, in the southwest of Ireland, one of the windiest situations in western Europe, the three summer months of 1878 gave a mean hourly velocity of 13 - 3 miles per hour, the minimum oscillating from 10 to 11 miles an hour from 9 p.m. to 6 A.Μ., and the maximum exceeding 16 miles an hour from 11 A. Μ. to 5 p.m. The absolute lowest hourly mean was 10 miles at 11 p.μ., and the highest 18 miles at 1 p.m., the velocity about midday being thus nearly double that of the night. Many observations might be added to these, including those published by Hann, Koppen, Hamberg, and others, which go to establish the fact that the curves of the diurnal variation of the velocity of the wind generally conform to the diurnal curves of temperature. The curves are most strongly marked during the hottest months; and the maximum velocity occurs at 1 p.m. or shortly thereafter, being thus before the time of occurrence of the maximum temperature of the day, and the minimum in the early morning, or about the time when the temperature falls to the lowest. The rule also holds good with all winds, whatever be their direction. The exceptions to this rule are so few and of such a kind that they are probâbly to be attributed to causes more or less of a local character.

Hann has shown, for a number of places in northern Europe, that with a clear sky the velocity is doubled from the minimum to the maximum, with a sky half covered the velocity is three-fourths greater, and with a sky wholly covered the velocity is only a half more. On the other hand at the strictly inland situation of Vienna, with a clear sky the velocity is double, and with a sky half covered it is two-thirds greater, but with a covered sky the diurnal variation in the wind’s velocity becomes irregular and faintly marked. Hann has also examined the winds at Vienna, and found that winds of a velocity not exceeding 30 kilometres an hour show a mean diurnal increase from 11 kilometres at 6 A. Μ. to 16’8 at 1 p.m., but that winds of velocity exceeding 30 kilometres an hour exhibit only a faintly marked and irregular increase of velocity during the day.

In offering an explanation of this remarkable fact regarding the diurnal variation in the velocity of the wind in all climates, it is to be remarked that the minimum velocity occurs when terrestrial radiation and its effects are greatest, but the increase of the velocity closely follows the sun, and the maximum is reached nearer the time the sun crosses the meridian than perhaps any of the other maxima or minima of meteorology which are dependent on the sun’s diurnal course. It is also to be noted that the winds over the open sea are practically uninfluenced by solar and terrestrial radiation, for there the diurnal curve of variation in the force of the wind is all but a straight line. On nearing land, however, the wind’s force exhibits a diurnal curve of variation as distinctly marked as, and bearing a close resemblance to, the analogous curve of temperature; while on the land itself these features become still more decidedly pronounced. Lastly, the amount of the diurnal variation of the temperature of the surface of the sea is less than a degree, whereas over all land surfaces the diurnal variation of the temperature is large, even where the ground is covered by vegetation, and enormously large over sandy wastes.

From this it follows that, so far as concerns any direct influence on the air itself, solar and terrestrial radiation exercise no influence on the diurnal increase of the velocity of the air with the increase of its temperature,—or, if any influence at all, such influence must be altogether insignificant, as is conclusively shown by the wind observations of the “Challenger” over each of the five great oceans of the globe. The same observations show that on nearing land the wind is everywhere greatly reduced in force. The retardation is greatest during the hours when the daily temperature is at the minimum; and it is particularly to be noted that, though the temperature rises considerably, no marked increase in the velocity sets in till about 9 a.μ., when the temperature has begun to rise above the daily mean. From this time the increase is rapid (see fig. 4); the maximum is reached shortly after the period of strongest insolation; and the velocity falls a little (but only a little) during the next three to five hours, according to season, latitude, and position, and falls again to near the minimum shortly after the hour when the temperature is at the mean. Even at the maximum, the velocity near land falls considerably short of the velocity which is steadily maintained over the open sea by night as well as by day.

The period of the day when the wind’s velocity is increased is practically limited to the hours when the temperature is above the daily mean, and the influence of this higher temperature is to counteract to some extent the retardation of the wind’s velocity resulting from friction and from the viscosity of the air. The increase in the diurnal velocity of the wind is in all probability due to the superheating of the surface of the ground and to the consequent ascensional movement of the air, tending to counteract the effect of friction and of viscosity between the lowermost stratum of the air and the ground. It is of importance in this connexion to keep in view the fact that in cloudy weather a temperature much higher than might have been supposed is often radiated from the clouds down upon the earth’s surface, which accounts for the phenomenon of the diurnal variation in the wind’s velocity occurring frequently also in cloudy weather. On the other hand, during the night, when terrestrial radiation is proceeding, the temperature of the surface falls greatly, and instead of an ascensional movement in the lower stratum of the air there is rather a tendency towards a descensional movement (if the wind be light there is an actual movement) of the lowest air stratum down the slopes of the country; and since the friction between the wind and the surface of the earth is thereby increased the diurnal velocity of the wind falls to the minimum during these hours (see also p. 156).

Among the most marked exceptions to the general rule of the diurnal distribution of wind force may be cited the bitterly cold furious blasts of wind encountered in narrow valleys in such mountainous regions as the Alps during clear and comparatively calm nights. These are simply the out-rush of the cold air poured into the upper basins of the valleys by the descensional currents from the slopes which the chilling effects of terrestrial radiation set in motion. On the other hand, the air of the valleys becomes heated and expands during the day, thus giving rise to a warm wind blowing up the valleys, which, on account of the vapour it carries with it from the lower levels, frequently covers the higher slopes and tops of the mountains with cloud and drizzling rain.

Diurnal Variation in the Direction of the Wind.— In all climates near seas and other large sheets of water, where the distribution of atmospheric pressure is tolerably equable, or the barometric gradient small, and the sun heat moderately strong, land and sea breezes are of daily occurrence. In such places a breeze from the sea gradually sets in in the morning, which gradually rises to a stiff breeze during the heat of the day and again towards evening sinks to a calm. Soon after this a breeze sets in from the land, blows strongly seaward during the night, and dies away in the morning, giving place to the sea breeze as before. These breezes are occasioned by the surface of the land being heated in a much higher degree than that of the sea during the day; the air over the land being thereby made lighter ascends, and its place is supplied by the cooler air of the sea breeze drawn landward, and partly also by descending currents, as shown by the humidity observations of the “Challenger,” which indicate increasing dryness when the sea breeze is strongest. Again during the night the temperature of the land and of the air over it falls below that of the sea, and the air of the land thus becoming heavier and denser flows over the sea as a land breeze. As the best-marked and most frequently occurring cases of the sea breeze begin some distance out at sea and gradually approach the land, it is very probable that, as suggested by Blanford, the ascending heated air flows seaward as an upper current, and that the increased barometric gradient thus caused largely accounts for these breezes.

Sea and land breezes are thus determined by the relative positions of the land and its coasts, subject to a further modification arising from the rotation of the earth. Thus on the coast of the Gulf of Lyons the sea breeze from the south veers to south-west and dies away as a west wind, while the land breeze from the north gradually veers to north-east and dies away as an east wind. On the coast of Algeria, on the other hand, the sea breeze veers from north to north-east and dies away in the east, whereas the south land breeze veers to south-west and dies away in the west. Sea breezes also occur in such unsettled climates as that of Scotland, when the weather conditions are favourable. These conditions are presented when an anticyclone overspreads the country, with its accompanying fine settled weather, small variation in the distribution of atmospheric pressure, clear skies, and consequently strong sunshine. Under these conditions the following are the veerings of the wind off the coast of Berwickshire. In the morning the wind is north-west till about 10 A.Μ., when it veers to north, falling all the time till finally it sinks to a calm. A little before noon it springs up from north-east or east, veers to south-east from 2 to 3 p. Μ., where it continues till 7 p. μ. , about which time it veers to south and then south-west, diminishing in force and finally sinking to a calm. About sunset it springs up from west, veering to north-west during the night, where it continues till the following morning. The wind thus virtually makes the round of the compass, is strongest from northwest and south-east and weakest at north-east and south-west, being thus strongest when its course is perpendicular to the line of coast.

The observations made by the “Challenger” in the region of the north-east trades in 1873 show a small diurnal variation in the direction of the wind, the variation being from E. 47° 5' N. at 2 to 6 A. μ. to E. 56° N. at 10 A. μ. to 2 p.m., the variation being thus 8° 55' towards north during the hottest hours of the day. At Mauritius observatory, which is several miles from the sea, the daily period in the direction of the wind is from E. 22° 15' S. at 4 a. μ. , being the most southerly point, to E. 7° S. at 1 p. μ. , and thence back to E. 22° 15' S. at 4 a. Μ. The diurnal variation is 15° 15', and thus the influence of the sun impresses on the wind at this observatory a more truly easterly character.

At the Austrian naval station at Pola, near the head of the Adriatic, the daily variation in the direction of the wind is well-marked. Starting from a point east of south at 5 A.Μ., it gradually veers round to westward, the most westerly point, almost due west, being reached at 5 to 6 p.m., after which it gradually shifts back to its starting point in the morning. Here we have evidently a diurnal wind-system different from that of the land and the sea breeze. Pola is situated near the south-western extremity of the peninsula of Istria, and the direction in the early morning of east by south is the direction the wind would take if a small anticyclone overspread the peninsula; and the direction from the west in mid afternoon is the direction the wind would have at Pola if the peninsula were occupied by a small cyclone with the lowest pressure in the centre. Now the influence of solar radiation is to form, through the ascending current from the heated land, a diminution of pressure over the land,—in other words, what is essentially a cyclone. On the other hand, during the night the influence of terrestrial radiation is to generate, through the cooling of the land and the air resting above it, a relatively higher atmospheric pressure in the interior of the peninsula with its characteristic system of out-blowing winds.

At Coimbra, in July 1878, the diurnal variation of the wind’s direction was from W. 49° 37' N. at 2 to 6 A. Μ. to W. 33° 15' N. at 4 to 6 p.M., the amount of the variation being thus 16° 22' in the direction of west. At Valentia, in the south-west of Ireland, during the summer months of 1878 the diurnal variation of the wind’s direction was from W. by S. at 7 to 9 A. Μ. to S.W. by W. at 5 to 7 P.M. The variation was thus from a point nearly south to a point nearly south-west, or through nearly 45° in the direction of west. On the other hand, at Aberdeen during the same months of 1878, the diurnal variation of the wind’s direction was from S.W. at 6 to 7 a. Μ. to S. by E. at 12 to 4 p.M., the variation being thus 56° from south-west in the direction of east through south. Attention is here drawn to the exactly opposite ways in which the diurnal veering or shifting of the wind takes places at Valentia and Aberdeen, but particularly to the important circumstance that in each case the diurnal changes in the wind’s direction which actually occur are precisely those that would take place on the supposition that during the hottest hours of the day an ascensional movement of the air sets in from the heated lands of the British Islands, and that an in-draught takes place all round, which with the descending currents makes good the loss caused by the up-draught. Thus then both the diurnal increase in the wind’s velocity and the change in its direction which observation shows to take place during the hottest hours of the day are traced to the same cause, viz., the heating of the surface by the sun, the heating of the lowest stratum of air resting on the surface, and the ascensional movements which are the necessary result.

It is instructive to note that at Nukuss, at some distance to the south of the Sea of Aral, where the summer direction of the wind is northerly, the north component is at the daily maximum at 4 p. Μ., having shifted into this direction from north-east, where it is at 9 a. μ. Much or nearly everything remains to be done in working out this problem in its practical details as one of the important elements of climatology, with the view of arriving at some definite knowledge of the influence of physical configuration and different vegetable coverings of the surface on radiation and on the velocity and direction of the wind.

Diurnal Variation in the Amount of Cloud.— Mists and fogs are visible vapours floating in the air near the surface of the earth, and clouds are visible vapours at a considerable height. These forms of visible vapour are all produced by whatever lowers the temperature of the air below the dew point,—such as radiation from the molecules of the atmosphere towards the cold regions of space, the simple expansion of the air of ascending currents, the mixing of cold air with air that is warm and moist, and the cooling of the air in contact with the surface of the earth when its temperature has been lowered by nocturnal radiation.

The forms of clouds are endless. Since clouds are subject to certain distinct modifications from the same causes which produce other atmospheric phenomena, the face of the sky may be regarded as indicating the operation of these causes, just as the face of man indicates his mental and physical sûtes. Hence the importance of the study of clouds, and hence the necessity of a nomenclature of clouds as the basis of accurate and comparable observations. An adequate nomenclature of clouds is still a desideratum. Luke Howard’s classification, which continues to hold its ground as a provisional nomenclature, was proposed by him in 1803, and by it clouds are considered as divided into seven kinds. Of these, three are simple forms, the cirrus, the cumulus, and the stratus ; and four intermediate or compound, the drro-cumulus, the drro-stratus, the cumulo-stratus, and the cumulo-cirro-stratus, nimbus, or rain cloud.

The cirrus cloud consists of wavy, parallel, or divergent filaments, which may increase in any or all directions. It is the cloud of the least density, the greatest elevation, and the greatest variety of figure. It is probable that the particles composing it are minute crystals of ice or snowflakes. The cirrus is intimately connected with the great movements of the atmosphere; and it is solely from the movements of the cirrus that we have any direct knowledge of the upper currents of the atmosphere. In recent years much has been done, particularly by Professor Hildebrandsson of Upsala and Clement Ley, in investigating the relations of this cloud to storms and other changes of weather.

The cumulus is the name applied to those convex or conical heaps of clouds which increase upwards from a horizontal base. They are generally of a very dense structure, are formed in the lower regions of the atmosphere, and are carried along by the aerial current next the earth. They form the tops of the ascending currents which rise from the heated ground, and have a diurnal period so well marked that they are often named the “cloud of the day.” The form of stratus comprehends those mists and fogs which in the calm evening of a warm summer day make their appearance in the bottom of valleys and over low-lying grounds, and sometimes spread upwards over the surrounding country like an inundation; they have an equally well marked daily period, and are frequently called the “cloud of night.” The cirro-cumulus is made up of small roundish masses, lying near each other, and quite separated by intervals of sky. It may be considered as formed from the cirrus by the fibres of that cloud breaking, as it were, and collapsing into roundish masses, thus destroying the texture but retaining the arrangement of that cloud. This singularly beautiful cloud is commonly known as a mackerel sky, and is of most frequent occurrence during dry warm summer weather. The cirro-stratus consists of horizontal masses thinned towards the circumference, bent downwards or undulating, and either separate or in groups. Since this cloud has great extent and continuity of substance, but little perpendicular depth or thickness, it is the cloud which most frequently fulfils the conditions for the phenomena of coronae, solar and lunar halos, parhelia or mock suns, and paraselenae or mock moons. The cumulo-stratus is formed by the cirro-stratus blending with the cumulus, or spreading underneath it as a horizontal layer of vapour. The cumulo-cirro-stratus, or nimbus, is the well-known rain-cloud, which consists of a cloud or system of clouds from which rain is falling. At a considerable height a sheet of cirro-stratus cloud is extended, under which cumulus clouds drift from windward; these rapidly increasing unite and appear to form one continuous grey mass from which the rain falls. The breaking up of the lower grey mass indicates that the rain will soon cease. When a rain-cloud is seen at a distance, cirri appear to shoot out from its top in all directions; and it is observed that the more copious the rainfall the greater is the display of cirri. The cirrus, cirro-cumulus, cirro-stratus, cumulo-stratus, and nimbus are connected more or less closely with the great atmospheric movements of the cyclone and anticyclone. In what follows here only the amount of sky covered will be taken into account, and not the species of cloud covering.

The diurnal variation in the amount of cloud in the sky on the open sea is very small. The following are the means of two hundred and seventy-seven days’ observations on board the “Challenger,” stated in percentages of sky covered :— [table]

Two maxima are here indicated, the one about or shortly after sunrise and the other in the early part of the afternoon; and two minima, the one at noon and the other from sunset to midnight. The difference between the extremes is only 6 per cent, of the sky.

At Batavia the daily maximum is from 6 to 11 P.M., and the minimum from 8 to 11 A Μ., the extremes being 52 per cent, at 9 A.Μ. and 69 per cent, at 7 p.m., —a difference of 17 per cent. Of four daily observations at Mauritius, the maximum is 50 per cent, at 1 P.M. and the minimum 38 per cent, at 6 A. Μ. At Coimbra, observations of clouds have been made live times daily, and six years’ results give the maximum 63 per cent, at 9 p.m. and the minimum 52 per cent, at 9 A. Μ. At this place, during July and August, the greatest amount of cloud occurs at 6 p.m., and in these months the rainfall at Coimbra is very small. The minimum is more pronounced at 9 a. μ. than at any other period; in winter this phase occurs about four hours later. At the continental situation of Vienna, during the warm months of the year the maximum is at 2 p.M., with a secondary maximum about 6 A.Μ., and the minimum from 10 p.m. to 2 a.m. ; but during the cold months the maximum is at 6 A.Μ. and the minimum during the evening and night. In the Rocky Mountains, the chief maximum, 57 per cent., is at 3 p.m., with a secondary one 30 per cent, at 5 a.μ. ; and the chief minimum 20 at 3 A. Μ. and a secondary one 29 at 11 p.m. At Helsingfors the maximum of cloud occurs from 10 a. Μ. to 2 p.M., and the minimum from 10 p.m. to 2 a. μ.

Much yet remains to be done with regard to the determination of the diurnal variation of cloud, but from the above one or two deductions of a general character may be drawn. A maximum occurs in the morning and continues till shortly after the sun has risen, and this maximum is more decidedly pronounced over the open sea than over land. Its appearance is without a doubt due to the general cooling of the atmosphere through its whole height by terrestrial radiation, and its disappearance to the heating of the air, which commences about sunrise. Then follows one of the diurnal minima, which continues till midday, or a little later; in other words, it continues till, owing to the diurnal heating of the air by the sun, the ascending current has fairly set in. The period of this ascending current marks the second maximum, which during the warmer months is larger than the morning maximum over land. The cumulus is the characteristic cloud of this maximum. These clouds are merely the summits of the ascending currents which rise from the heated land, where the aqueous vapour is condensed in cloud by the expansion which takes place with increase of height.

These cumulus clouds throw a not unimportant light on the behaviour of the ascending currents which rise from the surface when heated by the sun,—inasmuch as they point to the fact that the current ascending from the surface is broken up and thereafter grouped into separate well-defined ascending currents, which are marked out and overtopped by these cumuli; and further it is probable, from their well-defined position, that the air composing the ascending currents is not only warmer but also more humid than the air in the clear interspaces at the same heights. It may also be regarded as highly probable that it is down through these clear interspaces that the descending air-filaments shape their course in their way to take the place of the air-molecules that ascend from the heated surface of the earth.

The secondary minimum occurs from about sunset onwards during the time of the p.m. maximum of atmospheric pressure. In a highly saturated atmosphere, which is so characteristic a feature of many tropical climates at certain seasons, this time of the day is remarkable for the amount of cloud; and it is during those seasons and hours that heat-lightning, or lightning without thunder, attains its annual and diurnal maximum period, which is from six to eight hours later than that of thunderstorms. The morning maximum, shortly before and after sunrise, has two quite distinct and characteristic clouds accompanying it. One of these is the cumulo-stratus, which is a consequence of the cooling of the atmosphere through all its height by nocturnal radiation. As the colouring of the cloudlets is often singularly fine, it has been a favourite theme with poets of all ages and climes. The other, which results from the cooling of the surface of the earth by terrestrial radiation and thence of the lower stratum of the atmosphere, is quite different, being a low creeping mist, appearing first in low-lying situations, and gradually extending upwards as the temperature falls. Fog is a well-known form of this cloud, having its diurnal maximum in the morning and early part of the day.

Diurnal Variation in the Amount of the Rainfall.— From the sixteen years during which hourly observations of the rainfall were made at Batavia Dr Bergsma has given a table showing the diurnal variation, of which the following shows how much per cent, of the total daily amount fell every two hours :— [table]

The diurnal curve of rainfall is thus very distinctly marked at Batavia. The minimum is from 6 to 10 A. Μ. and the maximum from 2 to 6 p.m., —10'7 per cent, falling during the four hours ending 10 a. Μ., but 25 - 7 per cent, in the four hours ending 6 p.m.

The observations were arranged and averaged by Dr Bergsma with the view of seeing how far the phases of the moon influence the. rainfall. The results for the eight phases of the moon, beginning with new moon, showing the mean amount of rain in twenty-four hours during the seventeen years ending 1880, are these 0'243, 0'236, 0 193, 0'181, 0'212, 0'183, 0Ί89, and 0'203, and daily mean 0'205 inch. The influence of the moon’s phases on the rainfall at Batavia is thus quite decided from these seventeen years; for, while the mean daily rainfall is 0'205 inch, it rises at full moon to 0'243 inch, from which time it gradually falls to 0Ί81 inch at the third octant, rises to 0'212 inch at the fourth octant, falls again to 0183 inch at the fifth octant, and finally rises to the maximum at the time of new moon.

At Coimbra, where the rainfall has been observed every two hours for the six years ending 1881, the means show a minimum of 3Ό8 inches from midnight to 2 A.Μ., a maximum of 4Ό3 inches from 2 to 4 A. μ., a second minimum of 3'20 inches from 10 A.Μ. to noon, and a second maximum from 2 to 4 p.m. These four phases of the rainfall are pretty nearly accordant with the four phases of the barometric pressure, the maximum periods being near the times of minimum pressure, and the minimum periods near the times of maximum pressure. The rainfall at Philadelphia shows a decided maximum at 6 p.M. and minimum at 3 A. Μ. At Vienna during the summer half of the year there are three maxima and three minima in the curve of the daily rainfall, the chief maximum, which is nearly double of each of the other two, occurring from 2 to 4 p.m., and the chief minimum from 3 to 6 a. μ. At this place the number of hours of rain has been recorded, showing two maximum periods, the first from 2 to 8 a. μ. and the second from 4 to 7 p.m. Since the time of maximum amount indicates a number of hours for the fall under the mean of the day, it follows that the showers from 2 to 4 p.m. are often very heavy. At Prague the chief maximum is from 2 to 6 p. Μ. and the minimum from 1 to 6 A. Μ. At Zechen the maximum is from 2 to 7 p.m., with a secondary maximum from 5 to 8 A.Μ., and the minimum from midnight to 4 a. μ. At Bern the maximum fall is from 7 P. Μ. to midnight. Twenty-one years’ observations at Calcutta show the maximum occurrence of rainfall during the rainy season from June to October to be from 11 A. μ. to 6 p.Μ., the absolute maximum being from 2 to 3 p.m. and the minimum from 9 p.m. to 1 a. μ.; and during the hot dry season from March to May the maximum is from 5 to 9 p. Μ., and the minimum from midnight to 9 A. Μ.

The data already collected show the general occurrence of a diurnal maximum from about 11 a.m. to 6 p.m., and this feature of the curve is particularly well seen in the rainfall of continental climates during the summer half of the year. A marked diminution of the rainfall is very generally observed from about sunset to midnight, when the diurnal amount of cloud in many climates falls to the minimum, and the evening maximum of atmospheric pressure takes place. The time of the morning minimum pressure from about 2 to 6 a. μ. is, curiously, strongly marked in some places with an increase, while in others it is as strongly marked with a diminution, in the rainfall. The “Challenger” observations show that the occurrence of rain on the open sea is inversely as the temperature, 684 days’ observations giving 96 cases from 9 to 4 p.m., and 135 from midnight to 2 a.m., these being the minimum and maximum periods.

Diurnal Variation of Thunderstorms.— It has just been remarked that a prominent feature of the diurnal rainfall in continental climates during the summer months is the increased rainfall from about 11 a.m. to 6 p.m., and the Vienna observations point to unusually heavy falls occurring at these hours. This is caused by the rains which accompany the thunderstorms of these regions. The following table gives for the hours of the day the times of occurrence of thunderstorms at Ekaterinburg in the Ural Mountains during the fourteen years ending 1872 :— [table]

Hence the thunderstorms at this place have a diurnal period as strongly marked as any other meteorological phenomena, and in this respect Ekaterinburg is fairly representative of extratropical continental climates during summer. For the fourteen years no thunder occurred during the six months from October to March. The mean of the six hottest months shows the maximum to take place from 3 to 4 P.M. and the minimum from 4 to 5 a.m., these being the times of occurrence of the two minima of pressure. At this season, however, the morning minimum pressure is but faintly marked in such climates as those of Siberia. During the twelve hours from 9 a.m. to 9 p.m., when the temperature is above the daily mean, 717 of the whole number occurred, thus leaving only 139 for the twelve hours when the temperature is below the daily mean. The great majority of the thunderstorms occur during the part of the day when the ascensional movement of the air from the heated ground takes place, and they attain the maximum when the temperature and this upward movement are also at the maximum. Owing to the westerly winds from the Atlantic which prevail over Europe and western Siberia during summer, the maximum rainfall of the year occurs over this extensive region in this season; and the importance and significance of the inquiry into this element of climate lie in the fact that the greater portion of the summer rains is discharged over these regions by the thunderstorm. The “Challenger” observations on the open sea show the maximum occurrence of thunderstorms to be from 10 p.m. to 8 a.μ., 22 being observed during these ten hours and 10 during the other fourteen hours of the day,—a result which suggests that over the ocean terrestrial radiation is more powerful than solar radiation in causing vertical disturbances in the equilibrium of the atmosphere.

Atmospheric vapour and ascending currents thus play an important part in the history of these thunderstorms. Where the climate is dry and rainless, like that of Jerusalem in summer, thunder is altogether unknown. On the other hand, where during a particular season an anticyclone with its vast descending current -in the centre remains over a region, as happens over the centre of the old continent during the winter, over that region thunder is equally unknown during that season. Further, in such places as Lisbon and Coimbra, where the summer rainfall is small and its occurrence infrequent, thunderstorms become less frequent, and the hours of their occurrence are later in the day than they are before and after the dry season.

The thunderstorms at Mauritius call for special notice. There are two maxima in the diurnal curve, the larger from noon to 4 p. Μ. and the smaller from 3 to 6 A. Μ., which are near the times of the barometric minima; and two minima, from 9 p. Μ. to 1 A. μ. and from 8 to 10 a. μ., these being near the times of the barometric maxima. But the important point as regards the thunderstorms of Mauritius is that for twelve years none were recorded in June and July, one only in August, one in September, and three in October. The annual period of the thunderstorms of this island extends from near the end of October to the middle of May, or during the time of the greatest rainfall, practically none occurring during the rest of the year. But rain continues to fall during the four months of no thunder, the mean monthly rainfall being then about 2 inches, falling, however, in September to 1 ’ 37 inches. During these four months, therefore, there is in the air the aqueous vapour, and, these being dry months, there is the condition of ascending currents. There appears, however, to be then wanting another element which seems essential to the. electrical manifestations of the thunderstorm, viz., the conditions which give masses of descending cold air along with the ascending current of warm moist air. During the months when thunder is of no unusual occurrence the high pressure of Asia repeatedly advances, as Dr Meldrum has pointed out, close on Mauritius; and so frequently is this the case that he considers the belt of calms between the two trade winds to stretch in a slanting direction from Madagascar to Ceylon. As long, as this state of things occurs with more or less frequency, the conditions of a descending cold current of large volume are provided, and thunderstorms occur. But during June, July, August, and September, when atmospheric pressure is low in Central Asia, and there is an unbroken increase of pressure from Asia southwards to Mauritius, and while Mauritius remain^ in the heart of the south-east trades, the conditions of descending cold currents of any considerable volume are not present, and thunder is there unknown at that season.

Now in situations which afford the three conditions of aqueous vapour, ascending currents, and descending cold currents, whilst the diurnal and annual periods are quite distinctly marked, the phenomena are more uniformly distributed through the hours of the day and months of the year than elsewhere. Pola and Fiume, at the head of the Adriatic, being shut in and encompassed by lofty Alps, are illustrations. At Fiume the greater maximum occurs from 11 a.μ. to 4 p.m. and the smaller from 2 to 4 a. μ., and the minima from 10 p.m. to 1 A. Μ. and 5 to 9 a. Μ. While during the twelve hours the temperature is above the mean of the day from May to September the number of the thunderstorms here was 245 for the nine years ending 1879, the number during the twelve hours the temperature is under the mean was 185. The comparatively large number during the colder hours of the night is no doubt due to the warm moist atmosphere of this confined sea and the close proximity of the Alps.

There is still another set of conditions favouring the development of thunderstorms in certain climates which the observations made at Stykkisholm in the north-west of Iceland illustrate. During the fourteen years ending 1879 there occurred here twenty-three thunderstorms, but there was only one in the six warm months from April to September; in other words, the thunderstorms of this climate are essentially winter phenomena. Further, of the twenty-three hours in which they occurred, only three were at a time of the day when the sun was above the horizon, viz., twice in March and once in September; in other words, the thunderstorms of Stykkisholm are nocturnal phenomena. It is instructive to observe that in the north and north-west of Scotland thunder occurs most frequently during the night and in winter, whereas in central, southern, and eastern districts it occurs most frequently during the day and in summer,—the thunderstorms in the former case approximating in type to those of Iceland and in the latter to those of Ekaterinburg. A little reflexion shows that in northwestern Europe it is during winter and during night that warm moist ascending and cold dry descending currents are most frequently brought into close proximity during the great Atlantic storms of the season; and it is at the changes of wind, humidity, and temperature accompanying the passage of the centres of the cyclones that the thunder peals are heard. On the other hand, in the east and south of Scotland it is during the hot months of the year that these ascending columns of warm moist air and descending columns of cold dry air are most frequently brought together, and there, accordingly, thunder with the heavy rains which accompany it is of most frequent occurrence from 11 A. Μ. to 5 P.M. from May to September These essentially different types of thunderstorms have been classed by Mohn as heat thunderstorms and cyclonic thunderstorms.

Given an initial difference of electric potential, it is easy to understand from the effects which follow the sudden extraordinary condensations of the aqueous vapour that take place how the most violent thunderstorms are produced. The difficulty is to account for the production of the initial difference of electric potential,—how, for example, in the same great aerial current of the south-west monsoon, this difference of potential is produced in the molecules of aqueous vapour at Calcutta but not in the aqueous vapour at Mauritius. It is to the physicist that meteorologists still look for the explanation.

Diurnal Period in the Occurrence of the Whirlwind, Waterspout, Dust Storm, and Tornado.— Whirlwinds, waterspouts, dust storms, and tornadoes are essentially the same, differing from each other only in their dimensions, their intensity, or the degree in which the moisture is condensed into visible vapour, while the hailstorm and the rainstorm are simply the manner and degree of the precipitation accompanying them. In several important respects they differ widely and radically from cyclones (see Atmosphere, vol. iii. p. 33). The largest tornadoes are of so decidedly smaller dimensions when compared with the smallest cyclones as to admit of no shading of the one into the other. Cyclones occur at all hours of the day and night, whereas whirlwinds and tornadoes show a diurnal period as distinctly marked as any in meteorology. Finally, cyclones take place under conditions which involve unequal atmospheric pressures or densities at the same heights of the atmosphere, due to inequalities in the geographical distribution of temperature and humidity; but whirlwinds occur where for the time the air is unusually warm or moist, and where consequently temperature and humidity diminish with height at an abnormally rapid rate. Cyclones are thus phenomena resulting from a disturbance of the equilibrium of the atmosphere considered horizontally, but whirlwinds and tornadoes have their origin in a vertical disturbance of atmospheric equilibrium.

Among the most remarkable of the tornado-swept regions of the globe are certain portions of the United States; and to the examination of these the meteorological service of the States has given special attention by a systematic, careful, and minute observation of the attendant phenomena and the destructive effects. The tornadoes of the last eighty-seven years, numbering about six hundred, have been classed under the different States where they are reported to have occurred, and fig. 5 shows this relative distribution over the States. The areas of greatest frequency are at long distances from each other. That part of the great basin lying west of the Mississippi, including the States of Iowa, Missouri, Kansas, and Nebraska, is the region in which tornadoes are most frequent. Tornadoes occur at all seasons, being most frequent, however, from April to September, and least frequent in December and January.

The hour of occurrence of one hundred and sixty-two of the tornadoes is given in the official report as follows :— [table]

Thus the diurnal period of tornadoes is analogous to the period for temperature, wind velocity, and thunderstorms. The atmospheric conditions which appear invariably to precede the formation of the tornado are violent contrasts of temperature and humidity immediately to the north and south of the path to be traversed by the storm. It is highly interesting to observe from fig. 5 that the region of most frequent occurrence of tornadoes is the region where a large number of the cyclones of the United States appear to originate (and the same region Loomis has shown to be remarkable for violent contrasts of temperature occurring within limited spaces and times), and that, as appears in the regions of the Alleghanies, they decrease in frequency with height.

Fig. 6 shows the waterspout in different aspects. A black cloud covers the sky, from which a projection is let down in the form of an inverted cone, as at A, which continues to increase and extend downwards. The sea immediately beneath is soon thrown into violent agitation, showing that the whirling movement which began in the clouds has extended to the sea, and is doubtless continuous throughout, though the portion of the column from A downwards is not yet made apparent by the condensation of its contained vapour into cloud. As the whirling movement of the column becomes more intensely developed, the increased rapidity of the gyrations brings about increased rarefaction of the air within, with the inevitable result of increased condensation of the vapour into cloud downward. The protrusion of the cloud and its extension downwards are thus not due to the descent of vapour from the clouds, but to the visible condensation of the vapour of the spirally ascending air-currents arising from an increasing rarefaction due solely to the accelerated rate of the gyrations, the condensation being analogous to that of the cloud seen in exhausting an air-pump.

Under each of the columns of fig. 6 the surface of the sea is seen to be more or less heaped up, as well as in violent agitation, showing that atmospheric pressure immediately under the gyrating columns is less than it is all round. On land, when the tornado passes directly over a dwelling house or other closed building, it often happens that the whole building, walls and roof, is thrown outward with great violence, the wreckage presenting the appearance of a sudden explosion, proving that atmospheric pressure outside the building was instantaneously and largely reduced, and the building shattered to fragments by the expansion of the air within. It is in this way that the tornado does some of its most dreadful work.

The wind of the tornado reaches a velocity probably never equalled in cyclones. During the Ohio tornado of February 4, 1842, large buildings were lifted entire from their foundations, carried several rods through the air, and then dashed to pieces, some of the fragments being carried distances of 7 and 8 miles; and large oaks nearly 7 feet in girth were snapped across like reeds. This tornado swept on its course at the rate of 34 miles an hour, and at one place did its fearful work in the brief space of a minute. The tornado which passed over Mount Carmel (Illinois), June 4, 1877, swept off the spire, vane, and gilded ball of the Methodist church, and carried it bodily 15 miles to north-eastward. The velocity of the ascending currents which kept this heavy object suspended in the air for 15 or 20 miles must have been very great.

Of the tornadoes the progressive courses of which were recorded, 310 advanced towards N.E., 38 towards S.E., 16 towards E.N.E., 14 towards E., 7 towards N.N.E., 5 towards E.S.E., and 3 towards S.S.E. The course is thus always towards some easterly direction, the great majority being towards the north-east. The velocity of their onward movement varied from 12 to 60 miles an hour, the average being 30 miles an hour. The time occupied in passing a particular spot varied from 10 seconds to half an hour, the mean time béing nearly six minutes and a half. The width of the path of destruction marked with debris and other relics of the violence of the tornado varied from 40 to 10,000 feet, the average being 362 yards. The direction of the whirling movement of the tornado was invariably from right to left, or the opposite of the movement of the hands of a watch, resembling in this respect the vorticose movement of cyclones in the northern hemisphere. The passage of the tornado cloud is often described as accompanied with remarkable noises, which observers variously characterize as terrible, deafening, a terrific crash, the roar of a thousand trains of cars, or the uproarious din of innumerable pieces of machinery.

The usual position of the gyrating columns of cloud is vertical; but occasionally a curving form or slanting direction is assumed. It is probable that to these latter forms many stationary or slowly moving dangerous squalls are to be referred, which spring up with unexpected suddenness in lakes and arms of the sea in mountainous regions.

The dust storm of India, Arabia, and Africa is a well-marked type of the whirlwind. Previous to the outbreak of a dust storm the air is unusually calm and sultry, just as happens in the case of the tornado. The simplest form of the dust storm is that of a tall aerial column of sand moving onwards, and drawing into itself, as it whirls round in its course, dust and other light bodies within the sweep of the strong air-currents which blow along the surface of the ground and converge vorticosely round the base of the column. A form commonly seen is shown in fig. 7, which represents several dust columns grouped together, each whirling independently round its own axis with incurving air-currents at the base, while the whole group of columns is borne bodily forward, and presents striking aspects as the forms and relative positions of the columns are changed. The importance of the observations made on dust storms as leading to a correct understanding of the whirlwind consists in this that it affords conclusive evidence that there is a strong inflow of the air along the surface of the ground all round vorticosely towards the base of the whirlwind, and that these same inflowing air-currents afterwards ascend through the air along the central axis of the whirlwind, carrying with them the evidence of their ascent in the visible solid particles of dust, sand, and other light objects they whirl up with them in their ascending course.

Owing to the extreme dryness of the air-currents involved in the dust storm, the rarefaction generated by the rapidity of the gyrations is insufficient to produce condensation of the aqueous vapour in the interior of the column. Quite different, however, is it with waterspouts and tornadoes, where, in the great majority of cases, the air near the surface before being drawn into the ascending vortex is of a high temperature and near the point of saturation. From the extreme rarefaction to which these air-currents are subjected, owing to their sudden ascent in a rapidly gyrating column, excessive condensation follows, with an aqueous precipitation at times so astonishing that it can only be fittingly described as an aerial torrent of solid water, or an aerial avalanche of hail and ice.

Certain tracts of the ocean included within what may be called permanent anticyclones, or where atmospheric pressure is higher than all round, are characterized by an absence or comparative absence of rain. These regions are also remarkable for clear skies and strong sun heat. Similarly small anticyclonic areas occurring between or in the vicinity of cyclones are characterized by dry air and clear skies, and it is under these conditions that the strongest sun heat is felt. When, as repeatedly happens in the warmer months of the year, anticyclones remain practically stationary for some time, the lowermost strata of the air become abnormally heated,—thus bringing about a vertical disturbance of the equilibrium of the atmosphere out of which whirlwinds originate. It is under these conditions that white squalls or fair-weather whirlwinds occur, the originating cause of this special form being the great dryness of the air due to its place in the anticyclone, and the abnormally rapid diminution of temperature and humidity with height owing to the strong insolation through the clear dry atmosphere. The clouds accompanying the white squalls are at a great height, but the commotion and boiling of the sea under them and following them as they drift onwards show that the squalls are true whirlwinds, the vapour column of the waterspout not being formed solely on account of the extreme dryness of the air which ascends the columns. The white squall accompanies fine weather, and its appearance is sudden, its duration brief, and its destructive power at times so dreadful that it has been known to strip a ship of every sail and mast in a few seconds, and leave it rolling a helpless log amidst the tremendous sea which follows it. In sailing through such regions a close lookout should be made, particularly when the weather looks singularly fine, the skies beautifully clear, the air calm or nearly so, and the temperature and moisture of the air on board the vessel noticeably high.

Diurnal Period of Hail.— The hail here referred to is round, hard, and compact, and in the form of clear or granular ice, the hailstones sometimes being found when broken across to be composed of alternate layers of these two states of ice. The following figures show the number of times it has occurred during the different hours of the day at Coimbra during the last six years :— [table]

A diurnal period is thus well-marked at Coimbra, where forty-eight out of the fifty-nine cases have occurred from 10 A. μ. to 4 P.M. This period is essentially the same as those calculated for a large number of places in representative climates, care having been taken to limit the inquiry to the particular hail described above. The important point to be noticed in the diurnal period of hail is that the time of maximum is about two hours earlier than the maximum period of thunderstorms. The maximum period for the thunderstorm is when the ascending current from the heated land is at its greatest force for the day; but the maximum period for hail is some time before the ascending current has fully established itself, or at that time of the day when the vertical disturbance of the atmosphere is greatest,—in other words, when atmospheric temperature and vapour fall with height at a much greater rate than the normal. In the higher latitudes hail falls almost exclusively during the warmer months of the year. In regions where the summer climate is practically rainless no hail falls; and where the rainfall is small and at distant intervals few cases of hail occur. Thus at Coimbra, where little rain falls in summer, hail was recorded as having fallen only once in the six years during the four dry hot months from June to September.

All hail is probably connected immediately with whirlwinds, more or less developed; and it is when the hailstorm is one of the phenomena attendant on the tornado or on a great thunderstorm that it assumes its most destructive form. The theory of the formation of hail has been stated by Ferrel in his Meteorological Researches for the Use of the Coast Pilot, part ii. p. 85. The vapour carried aloft by the gyrations of the tornado is below a certain height condensed into cloud and rain, but above that height into snow. Let the raindrops formed below be carried up into the snow region by the powerful ascending currents of the tornado and be kept suspended there a little while, and they become frozen into hail. If now these be thrown quite outside the gyrations of the tornado, they fall to the earth as a shower of compact homogeneous hailstones of clear ice of ordinary size. If, however, they are caught in the descent and carried in toward the vortex by the inflowing currents on all sides, they are again rapidly carried aloft into the freezing region. A number of such revolutions of ascent and descent may be made before they fall to the earth. While high up in the snow region, the hailstones receive a coating of snow; but, while traversing the region lower down where rain yet unfrozen is carried up, they receive a coating of solid ice. Thus alternate coatings of snow and ice are received, and the number of each sort indicates the number of revolutions described before the hailstones fell to the ground. When the nucleus is composed of compact snow, as is generally the case, the hailstone had its origin high up in the snow region as a small ball of snow, or soft hail (Graupel in German and grésil in French); but when it is composed of clear ice throughout it was formed in the rain region, carried up into the snow region and there frozen, and immediately afterwards fell to the ground.

Monthly, Annual, and Irregularly Recurring Phenomena. The Temperature of the Sea.— Figs. 8 and 9, representing the distribution of the temperature of the surface water of the ocean for the two extreme months February and August, are reproduced chiefly from The Wind and Current Charts for Pacific, Atlantic, and Indian Oceans, published by the British Admiralty in 1872.

In February (fig. 8) the temperature of the surface of the sea falls to the annual minimum over the northern hemisphere, and rises to the maximum in the southern hemisphere. The course of the isothermals more closely follows the latitudes in the Pacific, Indian, and South Atlantic Oceans; but the divergence from the latitudes is great and striking over the North Atlantic. The wider and more open the ocean the more does the distribution of the temperature approach the normal; and the more confined the ocean the greater is the divergence from the normal. The key to the anomalous distribution of the temperature of the ocean is furnished by the charts of the distribution of atmospheric pressure and the prevailing winds of the globe. So far as observation has gone it would appear that the surface currents are practically altogether caused by the prevailing winds over the respective oceans, subject to such deflexions in their courses as are occasioned by the land.

In the southern hemisphere the currents on the west side of the Indian Ocean flow southwards along the east coast of Africa, and, since the currents here pass from lower to higher latitudes, the temperature along the whole extent of this coast is raised considerably above the normal. On the other hand, since the currents on the west coast of Africa flow from south to north—in other words, from higher to lower latitudes—the ocean currents which impinge on this coast have a temperature much under the normal. The winds and currents on the coasts of South America are precisely analogous to those of Africa, and the distribution of the temperature of the sea is also similar. The temperature of the ocean on the east coast of that continent is for the same latitudes everywhere higher than on the west coast. Even in the smaller continent of Australia the same law holds good.

In the northern hemisphere a different distribution of the temperature of the sea is seen at this season. In the Atlantic the temperature is very much higher on the west of Europe than on the east of America. On the east of America from Wilmington to Boston occur the most rapid transitions in the mean temperature of the ocean anywhere on the globe, the temperature falling in that short distance from 70° to 30°, whereas on the eastern side of the Atlantic these isothermals pass Cape Verd Islands and Spitzbergen respectively. In the winter months the prevailing winds of the east side of North America are north-westerly, whilst in the central and eastern portion of the Atlantic they are south-westerly, thus pouring along the east coast of America the icy currents of the Arctic regions, but over the central Atlantic and along the western shores of Europe the warm waters of southern climates. The easterly and south-easterly winds of Scandinavia in winter lower the isothermals along these coasts. A striking feature of the winter isothermals of the Atlantic is the singularly high temperature along the centre stretching from Spitzbergen towards the south-west and extending in a modified degree as far south as the West Indies. In the Pacific this feature of the mid-ocean temperature is much less pronounced, and the excess of temperature on the west of America over what occurs in the same latitudes of eastern Asia is not so great as the difference observable between the two sides of the Atlantic.

The highest mean temperature in February (85°) occurs in the Indian Ocean to the south-west of Sumatra, and there is a patch the temperature of which is 84° to the north of Madagascar. The highest means in the Atlantic are 82° in the north-east angle of the Gulf of Guinea, and 81° off the north-east coast of Brazil. In the Pacific the highest are 83° to the north of the Fiji Islands and 81° near the Marshall Islands.

In August (fig. 9) the southern half of the Red Sea shows a mean temperature of 90°, being the highest mean recorded for the ocean anywhere at any season. Patches showing a summer mean of 85° occur in the Chinese Sea to the east of Tonquin, in the Bay of Bengal to the east of southern India, about Socotra, and to the west of Central America. But the most extensive regions of high temperature are in the west of the Pacific between long. 165° E. and the Philippines· northward nearly to Japan and southward to New Guinea, and the Gulf of Mexico and the adjoining part of the Atlantic as far east as long. 57° W. A patch of remarkably low temperature occurs in the Pacific a little to the west of Galapagos, where the mean is only 70°, being 10° lower than what occurs anywhere else near the equator at this season.

The influence of currents is strongly expressed in the temperature of all the oceans. In the south or Asia the monsoons are S.W., S., and S.E. Under the impulse of these monsoonal winds an extensive surface drift of the waters of the equatorial regions is earned northwards towards southern Asia, and consequently very high temperatures characterize these seas in summer. It is instructive to note the effect on the temperature of the sea resulting from the region of high atmospheric pressure in the North Atlantic at this season. Out of this anticyclonic region the winds blow in all directions, giving rise to surface currents flowing in the same directions. Thus to the west of Africa the winds and currents are from north to south; and hence the temperature of this part of the ocean is abnormally reduced. On the other hand, on the west side of this high pressure area, the prevailing winds and currents are from south to north, and it will be seen that the temperature of the whole of the region swept by the southerly winds is abnormally raised. Ou the north side of the area, the winds and currents are westerly as far as about long. 35° W., and over that space the isothermals follow the parallels of latitude. Farther to eastward and northward the prevailing winds become south-westerly, thus propelling northwards along the western shores of Europe, by oceanic surface drifts, the warmer waters of southern latitudes. Meanwhile the currents of cold water and ice drifts from the Arctic regions keep the temperature off America to the north of Newfoundland at a figure considerably lower than is observed in any other region in the same latitudes. In August similar relations exist as in January between the east and west coasts respectively of South Africa, South America, and Australia, all of which are readily explained by the charts of mean atmospheric pressure and the resulting prevalent winds.

One of the most striking facts of ocean temperature is that the temperature of the Southern Ocean from about 50° to 60° S. lat. is practically the same in January and August, a circumstance due chiefly to the magnificent icebergs of that ocean.

The. Temperature of the Land.— In regions where the rainfall is distributed through all the months of the year, and where snow covers the ground for only a small part of the year, the mean temperature of the soil nearly equals that of the air. But when the year is divided into wet and dry seasons, and when snow lies during a considerable portion of the year, the mean annual temperature of the soil may be above or below that of the air. The greatest difference between the temperature of the soil and that of the air occurs where the surface of the ground is covered during several months with snow. Snow is a bad conductor of heat, and thus obstructs the free propagation of the cold produced by radiation downwards into the soil, and the escape of heat from the soil into the air. In this way, over a considerable portion of the Russian empire, the temperature of the soil is considerably in excess of that of the air. Thus at a place 120 miles south of Archangel the temperature of the soil is 10° higher than that of the air; and at Semipalatinsk it is 9° higher.

The daily changes of temperature only affect the soil to depths of about 4 feet. The precise depth varies with the degree of the sun-heat and with the nature of the soil. Similarly the heat of summer and the cold of winter give rise to a larger annual wave of heat propagated downwards, the amplitude of which diminishes with the depth till it ceases to be perceptible. Principal Forbes showed from observations on the Calton Hill, Edinburgh, that the annual variation is not appreciable lower than 40 feet below the surface, and that under 25 feet the change of temperature through the year is small. The depth at which the annual variation ceases, or where the temperature remains constant, is a variable depending on the conductivity and specific heat of the soil or rock, but particularly on the difference between the summer and winter temperatures. The rate at which the annual wave of temperature is propagated downwards is so slow that at Edinburgh, at a depth of 24 feet, the highest annual temperature does not occur till January 4, and the lowest till about July 13, thus reversing the seasons at this depth. At Greenwich, at a depth of 25 j feet, these phases of the annual temperature occur on November 30 and June 1.

Professor Everett in the Report of the British Association for 1879 has summarized, the results of the observations of underground temperature. The temperature of the surface of the ground is not sensibly influenced by the flow of heat from below upwards, but is determined by atmospheric and astronomical conditions. The temperature gradient is defined as the rate of increase of the temperature downwards, and it may be taken as averaging one degree Fahrenheit for every 50 or 60 feet, the exact rate in particular cases being very variable. Thus the temperature gradient of the soil is about five times steeper than the temperature gradient of the air. The temperature gradient is steepest beneath gorges and least steep beneath ridges; and hence the underground annual isothermals are flatter than the uneven surfaces above them. This is the case even with the uppermost isothermal of the soil, and the flattening increases as we pass downwards until at a considerable depth they become horizontal. Where the surface of the ground aιid the isothermal surfaces beneath it are horizontal, the flow of heat is vertical, and the same quantity of heat flows across all sections which lie in the same vertical. In this case the flow across a horizontal area of unit size is equal to the product of the temperature gradient by the conductivity, if the latter term be used in an extended sense, so that it includes convection by the percolation of water, as well as conduction proper; and hence, in comparing different strata in the same vertical, the gradient varies in the inverse ratio of the conductivity.

Since the effects of the cold generated by nocturnal radiation mostly accumulate on the surface of the earth, but the effects of solar radiation are spread to some height by ascending currents from the heated ground, it might be expected that the annual temperature of the surface layer of the soil would be lower than that of the air resting over them. Observations prove that such is the case. Springs which have their sources at greater depths than that to which the annual variation penetrates have a constant temperature throughout the year, and if they do come from a depth considerably greater than this they may be regarded as giving a very close approximation to the mean annual temperature of the place. The temperature of cellars is also very near the mean annual temperature of the locality; at any rate this temperature may be secured for cellars anywhere.

Distribution of Temperature in the Atmosphere.— Of the larger problems of meteorology, the distribution of temperature in the atmosphere over the land surfaces of the globe was the first that received an approximate solution (by Humboldt). But as regards the ocean, which comprises three-fourths of the earth’s surface, the question of the monthly and annual distribution of temperature in the atmosphere over it can scarcely yet be said to have been seriously looked at. The isothermals of the temperature of the atmosphere which cross the oceans continue still to be drawn essentially from observations made on the islands and along the coasts of these oceans. The first step towards the solution of this vital problem in climatology and other branches of meteorology is the construction of charts of mean monthly temperature of the surface water of the sea over all parts of the ocean from which observations for the purpose are available. In prosecuting this line of inquiry, excellent work has been done by the Meteorological Office as regards parts of the Atlantic between the tropics and the ocean to the south of Africa, and also by the Dutch, French, and German meteorologists. With such charts it would not be difficult, by a careful comparison during the same intervals of time between the temperature of the surface of the sea and that of the air resting over it, to construct monthly charts of the temperature of the atmosphere over the oceans of the globe.

In this connexion the whole of the observations of the temperatures of the air and sea made on board the “Challenger” have been examined, and sorted into one hundred and seventy-four groups according to geographical position, and the differences entered on a chart of the route of the expedition. In the Southern Ocean between latitudes 45° and 60° the temperature of the sea was lower than that of the air, the mean difference being l o, 4. The temperature of the air is here higher owing to the prevailing W.N.W. winds, and that of the sea lower owing to the numerous icebergs. To south of lat. 60° S. the sea was nearly 2°O warmer than the air, the result in this case being due to the open sea, which keeps up a higher surface temperature, and to an increased prevalence in these higher latitudes of southerly winds, thus lowering the temperature of the air.

The period during which the temperature of the sea exceeded that of the air was from June 1874 to March 1875, or during that part of the cruise from Sydney to New Zealand, and through the East India Islands to Hong Kong and thence to the Admiralty Islands. During the whole of this time, except when passing the north of Australia, the sea was much warmer than the air, the general excess being from 2 β to 3 β , rising even near Tongatabu to upwards of 4°. The climate of the southern part of this extensive region at the seasons visited has a large rainfall, much cloud, and consequently a comparatively small evaporation and sunshine. In June, when the “Challenger” passed the north of Australia, the climate was very dry, the sunshine strong, and the evaporation large, and there the sea was slightly colder than the air. In the Atlantic between lat. 20’ N. and 20° S. the sea was everywhere warmer, the mean excess being about a degree; and in the Pacific between lat. 30° N. and 30° S. the sea was also warmer, the mean <xcess being a degree and a half.

On the other hand, in the Atlantic from lat. 40° to 20’ Ν’, the sea was, on the mean, half a degree colder than the air. This region is remarkable for the high pressure which overspreads it, for the winds and currents which flow out in all directions, for its clear skies, strong sunshine, and consequently large evaporation, by which the temperature of the surface of the sea is lowered, and that of the air resting on it, being open to the heating influence of the sun, is raised. Similarly in the North Pacific from lat. 40’ to 30’ the temperature of the surface of the sea was half a degree lower than that of the air.

These remarks apply only to the observations made strictly on the open sea. Near land very great differences were observed which varied with season. Thus at Hong Kong during the latter half of November 1874 the sea was 3’7 warmer than the air, the low temperature of the air at this season being caused by the lower temperature of the land and the northerly winds which then prevail; on the other hand, at Valparaiso in November and December of the following year the sea was 5° - 8 colder than the air during the three weeks the “Challenger” was there, the difference being due to the cold oceanic current which sweeps northwards past that coast, and the rapid increase in the temperature of the air at that time of the year. These results will help us in gaining some knowledge of the temperature of the air over the oceans of the globe in February and August, taken in connexion with a careful examination of the sea temperature of these months represented in figs. 8 and 9.

The distribution of temperature over the surface of the globe is shown by figs. 10 and 11, which represent the temperature of the two extreme months January and July for the eleven years 1870 to 1880. The region of highest temperature, which may be regarded as comprised between the north and south isothermals of 80°, forms an irregularly shaped zone, lying in tropical and partly in subtropical countries. On each side of this warm zone the temperature diminishes towards the poles, and the lines showing successively the gradual lowering of the temperature are, roughly speaking, arranged parallel to the equator, thus showing in an unmistakable manner the predominating influence of the sun as the source of terrestrial heat. While this decrease of temperature corresponds in a general way to what may be called the solar climate, there are great deviations brought about by disturbing causes.

Among these disturbing causes the unequal distribution of land and water holds a prominent place. In January the earth presents to the perpendicular rays of the sun the most uniform surface, or the largest water surface, and in July the most diversified surface, or the greatest extent of land. Hence the zone of the earth’s surface comprised between the isothermals of 80° is less irregular, and also spreads over an area more restricted, in January than in July. In July the areas enclosed by the isothermals of £0° and 90° are much larger in the Old World than in the New, it being the former which presents the larger land surface to the perpendicular rays of the sun; and in January, the summer of the southern hemisphere, the most extensive area of high temperature occurs in Africa and the least in Australia, the high-temperature area of South America being intermediate. In contrast to this the belt of temperature exceeding 80 β is of least breadth where it crosses the Pacific and Atlantic Oceans, the absolute minimum breadth being in July in the Pacific, the largest ocean, where the disturbing influence of the land is least.

During the cold montl-s of the year, when the sun’s heat is least and the effects of terrestrial radiation attain the maximum, the greatest cold is over the largest land surfaces which slant most to the sun. Thus the lowest mean temperature that occurs anywhere or at any season on the globe is — 55° , 8 at Werchojansk (lat. 67° 34' N., long. 133° 5Γ E.) in north-eastern Siberia. In Arctic America the lowest isothermal is - 40° Ό. During the winter the ocean everywhere maintains a higher temperature in all regions open to its influence, as is seen, not only in the higher latitudes to which the isothermals push their way as they cross the Atlantic and Pacific, but also in their irregular courses over and near the Mediterranean, Black, Caspian, and Baltic Seas, Hudson’s Bay, the mouth of the St Lawrence, the American lakes, and all other large sheets of salt and fresh water. The disturbing influence of sheets of water on the temperature in all seasons is very strikingly shown when the isothermals are drawn for every degree, these marking out the prominent features of local climates, a knowledge of which is of so great importance to the agriculturist, the horticulturist, and the invalid. Figs. 12 and 13 represent charts of temperature of this description for the British Islands for 1870-1880 from the Jour, of Scot. Meteor. Soc., vol. vi. In the winter of the southern hemisphere the depressing influence of the land on the temperature is but slightly felt, owing to the small extent of the land surfaces and the comparatively low latitudes to which they extend southwards. In the south east of Australia and in the basin of the La Plata, a lower temperature prevails in the interior.

Another prominent disturbing cause operating on the mean temperature is to be found in the seasonal areas of low and high mean pressure in their connexion with the prevailing winds. Of these the most marked is the system of IbW pressure about Iceland during the winter months (see fig. 14). Since this region of low pressure gives to western Europe its prevailing south-west and south winds, and to North America its north-west winds in winter, it is plain that the temperature of western Europe is thereby abnormally raised by the simple fact of its prevailing winds coming from the ocean and from lower latitudes, and that the temperature of North America is abnormally lowered by its prevailing winds coming from the Arctic regions and from land. The opposite action of these two winds, which are part and parcel of the same atmospheric disturbance about Iceland, is shown from the fact that, while the mean temperature of the south coast of Hudson’s Bay in January is — 20°, in the same latitude in the Atlantic to the west of Scotland it is as high as 44°, or 64° higher. A similar though less striking result accompanies the low-pressure area in the north of the Pacific in winter.

Another area of low mean pressure which powerfully affects the temperature is the low barometer which overspreads the interior of Asia during the summer months (see fig. 17). Since from this disposition of the pressure the prevailing winds of Europe and western Asia are northwest and west, and over eastern Asia south-east and east, it follows that the temperature is abnormally raised on the eastern side and depressed on the western side of the continent by the direction from which they severally receive their prevailing winds. This is well shown by the course of the summer isothermals of 80°, 70°, 60°, and 50° across the Old Continent.

Since the strongest insolation occurs where the air is driest, the hottest summer climates are met with in those tropical and subtropical regions where no rain falls. The most extensive of the rainless regions during the summer months is perhaps that which extends from the Punjab westwards through Persia, Arabia, and North Africa to Spain. This is the region where the hottest climates of the globe are to be encountered. Similarly no rain falls at this time of the year in lower California and the States adjoining, and this feature of the climate, taken in connexion with the relatively low temperature of the coast due to the winds and ocean currents from the north which sweep past it, results in sharp contrasts of temperature within short distances such as have no parallel in any other climate.

Of the areas of seasonal high mean pressure, the high barometer of Central Asia in winter stands out in characteristic prominence (see fig. 14). Now, since the prevailing winds which necessarily form a part of this feature are south and south-west over Russia and western Siberia, the temperature of these inland regions is considerably higher than would otherwise be the case. On the other hand, since the prevailing winds are north-west in eastern Asia, the temperature of these regions is thereby abnormally depressed. It is this consideration chiefly which explains how it is that, while the mean January temperature in latitude 60° and longitude 120° E. is -30°, in the same latitude but in longitude 43° E. the mean temperature is 10°, or 40° higher, even though both regions are equally continental in their character.

The high mean pressure in the summer in the Atlantic between Africa and the United States has with its system of winds the most decided influence in bringing about the abnormal distribution of the temperature of that and adjoining regions. Since on its west side the prevailing winds are necessarily southerly, the temperature of that region is abnormally raised, and, on the other hand, since on its east side the winds are northerly, the temperature of the region is abnormally depressed. The result of these two opposite winds is seen in the slanting direction of the isothermal of 80° across the Atlantic, which slanting direction is continued far into the interior of North America for the reasons already stated.

These important bearings of cyclonic and anticyclonic areas on temperature and climate may be thus summarized. The temperature is abnormally raised on the east sides of cyclonic areas, and abnormally depressed on the west sides; but, on the other hand, temperature is abnormally raised on the west sides of anticyclonic areas and depressed on their east sides. In the southern hemisphere these directions are reversed.

Another set of influences, powerfully affecting the temperature, come into play where the surface of the land rises above the sea into elevated plateaus, lofty peaks, or mountain ranges. Thus it has been observed on Ben Nevis and other mountains that the wind during the day in summer exhibits an ascensional tendency due to the circumstance that the temperature of the surface of the mountain is heated in a much greater degree than the air strata at the same levels all around it. An ascensional current consequently rises from the mountain, which is maintained at a steadily stronger rate than at lower levels, because the drain from the updraught is easily supplied from the free surrounding atmosphere. It is the strong insolation at high elevations in the summer months which explains the excessively high day-temperatures encountered in the Rocky Mountains; and from the same conditions, viz., the rarity and purity of the atmosphere, by which terrestrial radiation is but little checked, come the low temperatures of the nights of these climates in the same season. From this cause it follows that the elevated lands in the interior of continents tend to reduce mean atmospheric pressure in summer to a greater extent than would otherwise be the case. In winter, on the other hand, the temperature of elevated regions in the interior of continents is very much colder than that of the surrounding atmosphere at the same heights, because in such regions the air is exceedingly dry and rare, and consequently radiation to the cold regions of space but little checked. Hence down the slopes of these high lands there are poured in all directions descending currents of very cold air, which intensify the rigours of the winters experienced on the low lands round their base, where accordingly the lowest mean winter temperatures occur. These elevated lands thus materially add to the high atmospheric pressure of the interior of continents during the cold months of the year.

But it is ocean streams and ocean currents which produce the greatest abnormalities in the distribution of the temperature of the air, and a glance at figs. 10 and 11 will show that it is in the North Atlantic where this cause is most strikingly seen. The increase thus accruing to the winter temperature is greatest about the north of Norway. It is also very great in the British Islands; thus, if no more heat were received than is due to their position on the globe in respect of latitude, the mean winter temperature of Shetland would be 3° and that of London 17°. But mainly owing to the heat given out by the Gulf Stream and other warm currents of the Atlantic their mean winter temperatures are respectively about 39° ’ 5 and 39°, Shetland being thus benefited 36°'5 and London 22°. The chart of the winter temperature of the British Islands well illustrates the influence of the surrounding ocean in maintaining a higher temperature. It will be seen that the south-west of Ireland is 7° warmer than the east coast of England in the same latitudes. The strong drift current from near Behring’s Strait southward along the coast of America has a powerful influence, particularly in lowering the summer temperature of that coast,—thus bringing about, in conjunction with the dry rainless climate of the interior, what are perhaps the most violently contrasted climates, within narrow limits, as regards their temperature. The deflexions of the isothermals near the Baltic, Mediterranean, Black, and Caspian Seas and the freshwater lakes of America all point to the disturbing influence of these sheets of water on the temperature.

The height and direction of mountain ranges is an important element in determining climate. If the ranges are perpendicular to the prevailing winds and of a considerable height, they drain the winds of much of their moisture, thus causing to places to leeward colder winters and hotter summers, by partially removing their protecting screen of vapour, and exposing them more completely to solar and terrestrial radiation. Of this Norway and Sweden and the British Islands form excellent illustrations. It is this that makes the most important distinctions among climates in regions near each other, as respects both animal and vegetable life. With regard to the decrease of temperature with height, very much yet remains to be done before an approximation to the law of decrease can be stated. During the five months observations were made on Ben Nevis in the summer of 1881 the difference between the mean temperature at sea-level adjoining and at the top of the Ben, 4406 feet above the sea, was 15° , 7, which shows a mean decrease of l o Fahr, for every 280 feet of elevation. The actual differences from day to day varied from l o ∙4 to 23 o- 2. As Ben Nevis forms a peak, and is in the very middle of the strong winds from the Atlantic, it is highly probable that this rate of decrease is a close approximation to the true decrease of the temperature of the air during the summer months in that part of the British Islands. When observations are made on elevated plateaus of some extent, the rate of decrease deduced from the observations will be less than the true rate in the free atmosphere in summer and greater in winter. The rate is thus a variable quantity, varying with latitude, situation, dampness or dryness of the air, calm or windy weather, and particularly with the season of the year. One degree Fahrenheit for every 300 is the rate of decrease generally assumed.

Amount of Aqueous Vapour.— It is scarcely possible to overestimate the importance of a knowledge of the horizontal and vertical distribution in the atmosphere of its aqueous vapour, for it may be truly said that it forms one of the prime factors in all the larger problems of atmospheric physics. A first rough approximation to the geographical distribution of the vapour of the atmosphere was published by Mohn in 1875 in his Grundzüge der Meteorologie, p. 84, in which vapour-pressure curves are drawn fur the globe for January and July. These leave much still to be done, not only in a further discussion of observations already made, but also in improvement of the methods of observation and in the tables for their reduction. The chief point of interest in Mohn’s vapour curves is their striking resemblance to the isothermals of the same months, and they also suggest that this line of inquiry is yet destined to make large contributions to our knowledge of the unceasing changes which occur in the pressure, temperature, cloud, rain, and movements of the atmosphere.

Still less is known of the vertical distribution of aqueous vapour. It decreases, like temperature, with the height, and if the statement generally made be at all correct, that half of the whole vapour of the atmosphere is contained in the lowest 6000 feet, and that at 20,000 feet high there is only about a tenth of what is at the earth’s surface, the rate of decrease with height proceeds at a greatly more rapid rate than is consistent with the supposition that it forms an independent vapour atmosphere existing under its own pressure. The establishment of an increased number of high-level stations, and a more systematic inquiry than has yet been attempted into the upper currents of the atmosphere, are much needed in the further development of this branch of meteorology. In carrying out the inquiry, invaluable assistance will be obtained from observations of the diurnal range of the barometer and from well-devised methods of observing the effects of solar radiation at the earth’s surface.

Amount of Cloud.— In Scotland, which lies completely within the region swept by the south-westerly winds from the Atlantic, and presents a well-defined mountain range lying across the track of these winds, the clouds have a distinct annual period. In the west, at places quite open to these westerly breezes, the amounts of cloud in spring, summer, autumn, and winter are respectively 67, 69, 71, and 74, and the annual mean 70. In the east, in such districts as East and Mid Lothian, which have extensive ranges of hills between them and the Atlantic, the proportions are 59, 63, 62, and 60, and the annual mean 61. Thus about a tenth more of the sky is covered with cloud at the western as compared with the eastern situations, and the distribution of cloud differs materially in western and eastern climates. In the west winter is the cloudiest season, but in the east it is summer, and these are respectively the months when most rain falls in the several climates. Everywhere spring is the season when the sky is clearest. In England, owing to the protection afforded by Ireland and Wales to the west and the comparative absence of ranges of hills, the amount of cloud is less than in Scotland, and it is more equally distributed over the country. The minimum amount occurs in spring, and the maximum in winter and autumn.

Some of the best illustrations of the seasonal variation in the distribution of cloud are afforded by the Old Continent. These variations are the simple consequence of the systems of wind caused by the high winter and low summer pressures of that continent. In eastern Siberia the prevailing winds in winter are N.W. or continental, and in summer S.E. or oceanic; and accordingly at Ajan, Nertchinsk, and Blagoweshtchensk the mean amounts of cloud in these two seasons are 18 and 44. On the other hand, in western Siberia and eastern Europe the prevailing winds in winter are S.W., or from lower to higher latitudes, and in summer N.W., or from higher to lower latitudes. Kazan may be taken as fairly representing this extensive region, and there the amounts of cloud for the four seasons beginning with winter are 71, 48, 44, and 62. As the N.W. winds of summer rise over the Ural mountains in their course, condensation of the aqueous vapour is increased, and hence over this region the cloud in winter and summer is nearly the same, the mean amounts at Bogoslpvsk, Ekaterinburg, and Zlatoust being respectively 53 and 52. At Tiflis and Kutais, situated on the high ground which lies between the Black Sea and the south of the Caspian Sea, the means for winter and summer are 53 and 55. On the eastern coast of the Black Sea the westerly winds of summer are accompanied with the annual maximum cloud, the winter and summer amounts at Redut-Kale being 59 and 69. In Central Siberia, to which the S.W. winds of winter do not extend, and to the north of latitude 55°, the amount of cloud is much diminished, and the cloudiness of summer is nearly the same as that of winter.

In India, in all regions which lie open to the summer monsoon, the minimum amount of cloud occurs during the winter and the maximum in summer,—the mean amounts being 19 and 74 at Calcutta, 16 and 86 at Bombay, 48 and 71 at Colombo, and 25 and 90 at Rangoon. At Trincomalee, on the east coast of Ceylon, and thus exposed to the rains of the N.E. monsoon of winter, and largely protected from the rains of the S.W. monsoon of summer, the amounts of cloud in these seasons are 52 and 59. At Darjiling (6912 feet) and Chakrata (7022 feet high), both on the Himalayas, whither the summer monsoon penetrates, the mean amounts are respectively 53 and 86, and 43 and 73. At Leh, in Kashmir, the amounts are 59 and 51, the excess being thus in winter. In the Punjab and to westwards, or those regions in southern Asia to which the summer monsoon does not extend, the cloud in winter is everywhere greater than in summer. Thus the amounts are 24 and 18 at Mooltan, 38 and 25 at Peshawar, 27 and 19 at Jacobabad, and at Quetta, in Baluchistan, 5500 feet high, 42 and 14. Similar relations as to cloud obtain in Australia and the other continents where high pressures rule in the interior during the cold months and low pressures during the warm months of the year. The maximum cloud occurs with winds from the sea and winds advancing into the colder regions of higher latitudes, and the minimum with winds which have traversed an extensive track of land and winds advancing into the warmer regions of lower latitudes. As the subject, however, is essentially one with rainfall, it is not necessary to prosecute it further.

The other atmospheric movements on which the amount of cloud depends are the ascending and descending currents of the atmosphere,—the ascending currents with clouded skies occurring in the belt of calms and over cyclonic areas and regions, and the descending currents with comparatively clear skies over anticyclonic regions. The region of maximum vapour and densest cloud-screen on the globe is the equatorial belt of calms between the trades, which has an annual movement northward and southward with the sun as already explained. To ascensional movements is to be ascribed part of the cloudiness of the southern and eastern sides of the winter cyclonic regions of the North Atlantic and North Pacific, and of the cyclonic regions of low summer pressure in the interior of Asia and other continents. On the other hand the comparatively small amount of cloud in the anticyclonic regions of the Atlantic and Pacific Oceans, and in the high-pressure regions of the interior of Asia and other continents during the cold months of the year, is due to the vast down-currents which occupy the centres of the anticyclones, and which become relatively drier as they descend owing to the increasing pressure to which the air is subjected.

Distribution of Atmospheric Pressure.— The importance of a knowledge of the distribution of atmospheric pressure, or of the mass of the atmosphere, over the globe in its varying amounts from month to month is self-evident. Observations teach us that winds are simply the movements of the atmosphere that set in from where there is a surplus towards where there is a deficiency of air; and observations also teach that isobaric maps (i.e., maps showing the relative distribution of mean pressure) and maps showing the prevailing winds are in accordance with each other. Since prevailing winds to a large extent determine the temperature and rainfall of the regions they traverse, isobaric maps may be considered as furnishing the key to the more important questions of meteoro logical inquiry. At the time of the first publication of isobaric maps of the globe in 1868, it was impossible to do more than present the subject in its broad general features, owing to the scantiness and quality of the materials then existing. But since then meteorological stations have been largely multiplied in all parts of the civilized world, and the general adoption of the issue of storm warnings has necessitated the use of more accurate barometers and uniform methods of observing. Since there is thus now the means of a more exact representation of this fundamental datum of meteorology, we have prepared a new set of isobaric maps, showing the distribution of the earth’s atmosphere and the prevailing winds for January (fig. 14), July (fig. 17), and the year. They have been constructed from mean values calculated for the same eleven years (1870-80 inclusive) as the isothermal maps figs. 10 to 13, pressure of 30 Ό inches and upwards being represented by solid lines, and of 29 - 9 inches and under by dotted lines, while the arrows show the directions of the prevailing winds at the localities indicated by the respective arrow-points.

Mean Atmospheric Pressure in January (fig. 14).—In this month, when the influence of the sun on the northern hemisphere falls to the minimum, the greatest pressures are massed over the continents of that hemisphere, and the least pressures over the northern parts of the Atlantic and Pacific Oceans, over the Antarctic Ocean and southern hemisphere generally. In the southern hemisphere there are three patches where pressure rises to 30 inches, viz., in the Atlantic between South America and Africa, south of the Indian Ocean, and in the Pacific between Australia and South America.

In the northern hemisphere, on the other hand, pressure rises in Central Asia to upwards of 30 ’ 5 inches, the mean pressure for January being at least 30 - 4 inches at Peking, Semipalatinsk, and Yenisei, and fully 30 5 inches at Irkutsk and Nertchinsk, in the upper basin of the Amur. This is the region where the normal atmospheric pressure attains to a maximum which is much higher than is reached in any other region or at any other time of the year. It will be observed that this region of highest pressure occupies a position near the centre of the largest continent. The area of high barometer is continued westward through Europe, through the horse latitudes of the Atlantic to Carolina, and thence through the United States to California, whence it crosses the Pacific to Asia. This belt of high pressure thus completely encircles the globe, broadening as it passes the land and contracting as it crosses the ocean. Its greatest breadth is over Asia and its least over the Pacific, or where land and ocean attain respectively their maximum dimensions.

Pressures greatly under the average cover the northern portions of the Pacific and Atlantic and also the greater part of the Arctic regions. In the north of the Pacific the normal pressure falls to about 29’6 inches between Kamchatka and Alaska. In the north of the Atlantic, however, a still lower mean pressure obtains over a narrow belt stretching from Iceland to the south of Greenland, the normal at Stykkisholm in the north-west of Iceland being 29 , 385 inches, and at Ivigtut in Greenland 29’361 inches. This low average for Ivigtut is the lowest normal known to occur anywhere and at any season in the northern hemisphere, and it is significant that the place is immediately to the north of that part of the Atlantic where a considerable number of the storms which sweep over Europe have their origin, and where not a few of the storms which cross the Atlantic from America develop intensity.

It has been seen that the highest mean pressure occurs near the centre of the largest extent of land; but as regards the two oceans the lowest pressure is met with in the northern division of the Atlantic, which is the lesser ocean. An inspection of fig. 14 shows, however, that the low-pressure area of the Atlantic is bounded to southward by systems of much higher pressures than are to be found in the Pacific. The result of this arrangement is that much stronger winds blow northward over the Atlantic and round upon Iceland; and, as these more quickly advance into colder latitudes, there is thus a greater and more frequent concentration of vapour and lowering of the barometer in the north of the Atlantic. The heavy rainfall of north-western Europe may be referred to as confirming this view.

A belt of low pressure passes through the equatorial regions quite round the globe. This marks the well-known region of calms towards which on either hand the trade winds blow. In the Atlantic it lies quite north of the equator even in January, when the sun’s course is farthest to southward, and it lies nearly parallel with the equator. On the other hand, in the Indian Ocean the position of the line of lowest pressure is to the south of the equator and not parallel with it, but taking a slanting course from near the north of Madagascar towards Sumatra, thence towards the low pressure which prevails at this season in Australia; its course is then a little to northwards, and crosses the Pacific to the central regions of South America. Its path is thus a devious one, being north of the equator only in the eastern part of the Pacific and in the Atlantic, but elsewhere to the south of it, being drawn farthest southward when under the influence of the regions of low pressure which now occupy central Australia, central and southern Africa, and central Soutlι America. In this trough of barometric depression nearly all the tropical storms of the Indian Ocean have their origin.

There are several important modifications of the isobaric lines as originally published. In 1868 the region of lowest pressure in the northern hemisphere in winter was represented as extending from Iceland to north-eastward; now the area of lowest pressure is seen to extend from Iceland south-westward to Greenland. In connexion with this point Captain Hoffmeyer discussed the weather of the North Atlantic during several winter months, and published the results in 1878, which conclusively showed that the meteorology of Greenland and Iceland exerts on the distribution of atmospheric pressure a powerful influence not before properly recognized, resulting in the mean minimum of pressure being localized distinctly to the south-west of Iceland, and that in addition to this minimum there are two subordinate minima, one in Davis Straits and the other in the Arctic Ocean midway between Jan Mayen and the Lofoten Isles. The investigation further established the fact that, when any particular one of these three minima plays an important part, the other two either do not appear at all or occupy quite a subordinate place, and that according as one or other of these minima of pressure predominates so is the character of the weather, as regards mildness or severity, of the winter of northwestern Europe and regions surrounding the North Atlantic. As regards the British Islands, the displacement of the minima to westward of the position shown in fig. 14 means milder winter weather, whereas a position more in the direction of the north of Norway means severer winter weather.

Another change implying important consequences is seen in the United States, where, instead of one, two distinct centres of maximum pressure occur, or rather the high pressure of the western and central States is separated from that of the southeastern States by a region of lower pressure occupying the region of the Mississippi States. Professor Loomis first drew attention to this peculiarity in 1879 in an inquiry into the distribution of pressure over the United States, and established the fact that there are two distinct areas of high pressure, the larger having its centre in Utah, and the less overspreading the greater portion of the southeastern and southern States, and that these two areas of high pressure are clearly separated from each other by a broad extensive region of lower pressure stretching in a south-western direction from the region of the great lakes to western Texas. The reason assigned by Professor Loomis is undoubtedly correct, that the relatively low normal pressure of the Mississippi States is due to the fact that the path usually taken by the barometric minima of

American storms in the earlier part of their course is from Texas to the lakes. Since, on the other hand, the centres of comparatively few storms, with their low barometer readings, cross the southern and south-eastern States, the normal winter pressure is higher there than it is along the Mississippi.

Another important modification occurs in India, where the isobar of 30 inches is deflected to the south-east toward Madras and thence towards the north-east to near Akyab in Arakan. This remarkable deflexion well shows the important influence exerted on the course of the isobar by large well-defined sheets of water and extensive tracts of land. The distribution of pressure here indicated, by which south of lat. 22° the normal pressure is considerably higher in the east than in the west of India, has, through the agency of the winds resulting from it, the most intimate and vital bearings on the distribution of the winter rains and temperature over considerable portions of India; and the same relations hold, but in a degree still more striking, in the meteorology of Ceylon.

The remarkable effect in interrupting or changing the course of the isobars is particularly λvell illustrated by the lines in the region of the Aral, Caspian, and Black Seas. As the point is of no small importance in meteorology, and is best illustrated by the Mediterranean and the countries to the north of it, we subjoin a map of this part of Europe (fig. 15), showing the normal pressure with greater distinctness and fulness than could be shown on fig. 14, the isobars being drawn for every half-tenth of an inch.

Here we see two distinct areas of high pressure, the one in Hungary and the other in the Peninsula, where the normal pressure exceeds 30’20 inches. The latter is the larger of the two, and may bo regarded as the prolongation of the region of high pressure which characterizes the Atlantic immediately to the south-west at this season. The high-pressure area included within the isobar 30’15 inches is of peculiar interest. In the Peninsula it covers a pretty broad area, but to the north-east it contracts to a narrow neck between the Bay of Biscay and the Gulf of Lyons, and again expands to north-eastward covering the distance from Carlsruhe to Modena, its prolongation eastward being there somewhat suddenly interrupted. At some distance to the eastward the second region of high pressure is met with, which is properly a part of the high pressure that overspreads the interior of the Old Continent in the winter months, its western limit being the isobar of 30’15 inches, which passes round by Pinsk, Cracow, Vienna, Laibach and the upper southern slopes of the basin of the Danube, Sebastopol, and thence southward in the direction of Cyprus.

The position of the latter of these regions of high pressure is approximately midway between the south coasts of Asia Minor and the Baltic. In other words, its position occupies the interior of this part of the Old Continent; and it is instructive to note that the position of the Black Sea and the Greek archipelago in the south portion of this region pushes the isobar of 30 Ί5 inches a good deal to northward. The position of the region of high pressure in the Peninsula, France, and Switzerland is also decidedly inland. It does not, however, exactly occupy the middle space of the land lying between the Mediterranean and the North Sea, owing no doubt to the circumstance that the very steep barometric gradient from France to Iceland greatly lowers the pressure over the whole of the northern half of France. It follows that the abnormally high pressure which so remarkably characterizes the interior of the Old Continent during the cold months of winter is represented, though in a greatly reduced form, westwards through the central districts of that continent.

These two regions of high pressure are separated from each other by a large area of comparatively low pressure overspreading the greater portion of the Mediterranean Sea,—marked off in fig. 15 by the isobar of 30 , 10 inches, within which pressure is everywhere less than 30’10 inches. This region includes an area of still lower pressure within the isobar of 30 05 inches, bounded by Sicily, Corfu, Athens, and Crete. Hence the singularly low pressure which characterizes the northern part of the Atlantic at this season has its analogue in the south of Europe, which is unquestionably due to the higher temperature and larger humidity of the climates of southern Europe which they owe to the Mediterranean.

It is deserving of special notice that, while the increase of the normal pressure of January from Genoa to Geneva is 0Ό81 inch, it is only 0Ό21 inch from Trieste to Riva, and that to the north of the Adriatic as far as latitude 50° pressure is considerably lower than obtains to the west and east of that region. An examination of the daily weather maps of Europe shows that not unfrequently the storms of north-western Europe on advancing as far to eastward as Denmark seem to connect themselves in some degree with Mediterranean storms prevailing at the time through a north and south prolongation of a system of low pressures. The comparative frequency with which this occurs is probably occasioned by the general drift to eastward of the atmosphere of Europe, considered as a whole, taken in connexion with the high mountainous ridge which bounds the Adriatic on its eastern side, from which it follows that the air overspreading the deep basin of the Adriatic is often highly saturated with vapour, and this highly saturated air is drawn northwards through central Europe when north-western storms of Europe with low barometric depression centres pass across Denmark and the Baltic. Thus the low normal pressure to the north of the Adriatic, separating the two regions of high pressure to the east and west of it, is in some respects analogous to the low normal pressure of the Mississippi valley, which separates the higher normal pressures of the Rocky Mountains and of the southeastern of the United States.

The influence of land and water respectively in the cold season of the year is well shown in fig. 16, which represents for every 0 020 inch the normal pressure over the British Islands in January, drawn from means calculated for two hundred and ninety-five stations.

It is in the winter months that the isobars of the British Islands crowd most closely together, and in accordance therewith strong winds are then most prevalent. The crowding of the isobars reaches the maximum in January, forming what is probably the steepest mean monthly barometric gradient that occurs at any season anywhere on the globe. The point, however, to which attention is here drawn is the remarkable influence of St George’s Channel and the Irish Sea in diminishing the pressures as they cross these seas, and of the land in increasing the pressure, which is seen in the curves occupying approximately the central districts of Great Britain from the Isle of Wight to Cape Wrath. This shows on a comparatively small scale the influence of the land in raising the normal pressure, and of the sea in lowering it, during the cold months of the year, justes is seen on the grand scale in central Siberia and the north of the Atlantic.

Mean Atmospheric Pressure in July (fig. 17).—In this month the physical conditions are the reverse of what obtains in January, the effects of the influence of the sun on the temperature and humidity of the atmosphere rising to the maximum in the northern and falling to the minimum in the southern hemisphere. With the solar conditions reversed, a comparison of figs. 14 and 17 shows that the distribution of atmospheric pressure in July is, considered in a broad sense, the reverse of what takes place in January.

In the southern hemisphere atmospheric pressure during the winter season is above the general average of 30 inches between lat. 10° and 40° S. This belt of high pressure encircles the globe, and embraces four regions where pressure rises considerably above this general high average. These regions are in South Africa, about lat. 20°, where it rises to a little above 30’20 inches; in Australia, where it rises on the Murray river very nearly to 30’20 inches; in South America, where in the basin of the La Plata, about lat. 30°, it rises to 30’13 inches; and in the ocean to westwards, where it reaches 30 02 inches. The point to be noted with respect to the position of these centres of high pressure at this season is that they occur over surfaces between latitudes 20° arid 36°. As compared with January, pressure in July over nearly the whole of this broad belt of the southern hemisphere is about two-tenths of an inch higher, which is the simple result of season. A comparison of January and July shows that this large accession to the pressure of the southern hemisphere is accompanied by an extraordinary diminution of pressure over the continents of the northern hemisphere.

Now, just as the greatest excess of pressure during the winter of the northern hemisphere occurs in the continent of Asia, so the greatest diminution of pressure in the summer months takes place in the same continent. The position, however, of these two extremes is far from being in the same region or even near each other. In the Old Continent the maximum occurs in the valley of the upper Amur, where, at Nertchinsk, the normal pressure in January is about 30’500 inches; whereas the lowest normal pressure in July is 29’412 inches, and occurs, so far as observation enables us to locate it, at Jacobabad on the west side of the basin of the Indus. The difference of these two normals is 1Ί88 inch; and over no inconsiderable portion of central Asia the normal pressure of July is an inch less than that of January. In other words, the influence of the sun in summer as exerted on the temperature and aqueous vapour of the atmosphere and atmospheric movements resulting therefrom is so powerful as to remove a thirtieth part of the whole mass of the air from this extensive region.

The large extension in recent years of good meteorological stations over the Russian and Indian empires enables us to lay down with much greater precision than formerly the lines of pressure. Of the changes indicated by the new isobars, the most important perhaps is the position of the region of minimum pressure in Asia, which is now seen to occupy the basin of the Indus, and thence stretches over a somewhat broad region to westward nearly as far as the head of the Persian Gulf. The point is of no small importance in atmospheric physics, inasmuch as it places the region of least normal pressure in July as close geographically to the region where at the time terrestrial temperature is highest as the region of highest normal pressure in January is situated with respect to the region where in that month terrestrial temperature is lowest in Asia.

The July isobars of India are of singular interest, and imply consequences of the utmost practical advantage to the empire. From Cutch southward the normal pressure is everywhere higher, and considerably so, along the whole of the west than it is in the east in the same latitudes, the difference being approximately half a tenth of an inch. This is represented on the map by the slanting of the isobars from north-west to south-east as they cross this part of India; and it is to be noted that the east and west coasts of Ceylon show the same manner of distribution of the pressure. The consequence of this peculiarity in the distribution of the pressure is that the summer monsoon blows more directly from the ocean over western and southern India than would have been the case if the isobars had lain due east and west, and thus probably precipitates in its course a more abundant rainfall over this part of the empire. But a more important consequence follows from the geographical distribution of the pressure over the valley of the Ganges. If the normal pressure there had diminished in the manner it does over India to the south of the Gangetic valley, the winds λvould have been south-westerly and the summer climate practically rainless. This, however, is not the case, but the normal pressure diminishes westwards along the valley of the Ganges, as the following mean July pressures will show:—Calcutta, 29’576 inches; Patna, 29’535 inches; Lucknow, 29’522 inches; Roorkee, 29’505 inches; and in crossing westward into the Punjab pressure falls still lower—to 29’439inches at Mooltan and 29’412 inches at Jacobabad. Indeed pressure in July is 0’220 inches lower at Jacobabad than at Sibsagar on the Brahmaputra, nearly in the same latitude. It necessarily follows from this distribution of the pressure that the summer monsoon, which blows northward over the Bay of Bengal, is deflected into an E.S.E. wind which fills the whole valley of the Ganges, distributing on its way a most generous rainfall over that magnificent region.

The influence of the land in lowering the pressure in summer is well illustrated by the course of the isobars over western Siberia and Russia, where pressure is seen to fall relatively lowest along the middle line of the Old Continent. In this connexion it is interesting to note the course of the isobar of 29’90 inches over that part of Europe where the breadth of the land is considerably increased—between the Baltic and Constantinople. In contradistinction to this the influence of the Aral, Caspian, and Black Seas in maintaining a higher pressure appears in the remarkable prolongation eastward of the isobars of higher pressure over the region of these seas, being in striking contrast to the lower pressures which prevail to the north and south.

The lowering of the normal pressure is very decided in the inland regions of Spain, North Italy, and Scandinavia. The effect is most strongly seen in Spain, the largest and compactest of these regions. Thus, while the normal pressure diminishes between Lisbon and Barcelona from 30’086 to 30’048 inches, the sea-level pressure at Madrid falls nearly to 30’000, and the pressure at Saragossa and Valladolid is nearly as low. This lowering of the pressure over the interior influences materially its summer climate. As remarkable an illustration of the principle as can be pointed to anywhere is seen in the north of Italy; for, while the normal pressure at Moncalieri is 29’941 inches, at Genoa on the coast the relatively high normal of 29’992 inches is maintained, the distance of the two places being about 40 miles. To the east pressure rises to 29’970 inches at Venice, and to westward to 30’023 inches at Geneva. Over Scandinavia, along the west coast from the Arctic circle southward, the normal pressure equals or exceeds 29 ’80 inches, the variation being comparatively small; and along the coast from the head of the Gulf of Bothnia to the south-east of Sweden pressure also exceeds 29’80 inches, and the increase from north to south proceeds at a slow rate. In, however, the strictly inland districts to the north-east of Christiania, which lie immediately to the east of the Scandinavian mountains, and sheltered by that lofty range from the winds of the Atlantic, pressure is considerably lower than it is along the east and west coasts of the peninsula. Owing to this peculiar distribution of the pressure, the winds which necessarily result from it give a much finer summer climate to the south-east of Norway and to the strictly inland part of Sweden than would otherwise be the case.

The remarkable curving northward of the isobar of 29’80 inches so as to include Lapland within it points probably to the influence of the White Sea and the wonderful lake system of Lapland in maintaining a higher summer pressure over that country, by which the northerly winds that blow towards the low-pressure region of Central Asia, to the serious deterioration of the summer climate of northern Siberia, do not extend so far to westward as Lapland.

The distribution of the normal pressure over North America is quite analogous to what prevails over Asia, but, the continent being less, the diminution of pressure in the interior ⅛ also correspondingly less. The highest normal pressure, 30’077 inches, is found in the south-east in Florida, and the lowest, 29 ’780 inches, in Utah, the difference being thus 0’297 inch. Another region of relatively high pressure is in the north-western States and British Columbia to the north; the maximum, near the mouth of the Columbia river, reaches 30 Ό62 inches, being thus nearly as high as what occurs in Florida. These two regions are merely extensions of important high-pressure areas which at this season are highly characteristic features of the meteorology of the North Pacific and North Atlantic respectively.

Of these two regions of high pressure the one overspreading the Atlantic between the United States and Africa is the more striking, being not only the region where pressure is highest anywhere on the globe during the months of June, July, and August, but where the normal pressure reaches the highest point attained at any season over the ocean. The highest point reached by the normal pressure over the land at any season occurs, as has been pointed out, near the centre of Asia, or approximately in the middle region of the largest continuous land surface on the globe during the coldest months of the year. On the other hand, the highest pressure over the ocean occurs during the warmest months of the year, and not over the largest water surface, but in the middle regions of the North Atlantic, where the breadth is only about half that of the water surface of the North Pacific.

From the essential differences between these two sets of phenomena it may be inferred that the extraordinarily high pressure which is so marked a feature of the meteorology of Central Asia during the cold months of the year is a direct consequence of the lowering of the temperature of the land of Asia and of the atmosphere resting on it during the time of the year when the effects of solar radiation are at the annual minimum, and of terrestrial radiation at the annual maximum. But the determination of the place and time of highest pressure over the ocean must be regarded as indirectly brought about. The physical conditions under which it occurs are these :—it happens (1) at the time of the year when the earth presents the largest surface of land to the sun, and (2) over that part of the ocean which is most completely surrounded by these highly heated land surfaces. This high summer pressure of the Atlantic has its origin in the upper currents of the atmosphere.

Mean Atmospheric Pressure for the Year.— The distribution of the annual atmospheric pressure may be considered as representing the sums of the influences directly and indirectly at work throughout the year in increasing or diminishing the pressure of the atmosphere. There are two regions of high pressure, the one north and the other south of the equator, which pass completely round the globe as broad belts of high pressure. The belt of high pressure in the southern hemisphere lies nearly parallel to the equator, and is of nearly uniform breadth throughout; but the belt north of the equator has a very irregular outline, and shows great differences in its breadth and its inclination to the equator. These irregularities wholly depend on the peculiar distribution of land and water which obtains in the northern hemisphere.

These two zones of high pressure enclose between them the comparatively low pressure of the tropics, through the centre of which runs a narrower belt of still lower pressure, towards which the tradewinds on either hand blow. Considered in a broad sense, there are only three regions of low pressure, the equatorial one just referred to, and one round each pole bounded by or contained within the zones of high pressure just described. The most remarkable of these, so far as it is known, is the region of loλv pressure about the south pole, which remains low throughout the year, playing the principal rôle in the wind systems of the Antarctic zone, in its heavy snowfall and rainfall, and in the enormous icebergs which form so striking a feature of the water of the Southern Ocean.

The depression around the north pole contains within its area two distinct centres of still lower pressure, the one filling the northern part of the Atlantic and the otner that of the Pacific. Of these two the low-pressure area round Iceland is the deeper, and is probably occasioned by the steeper barometric gradients and stronger winds which prevail over the North Atlantic. The broad equatorial zone of low pressure also contains two distinct regions characterized by still lower pressures. The larger of the two stretches across southern Asia from Assam to the head of the Persian Gulf, and is entirely due to the very low pressures which form so marked a feature in the summer meteorology of that part of Asia. The regions of the middle Indus and upper Ganges occupy the centre of this low-pressure area, where normal pressure falls short of 29'80 inches. The second area of lowest equatorial pressure is in the centre of Africa.

It may be here pointed out that the whole of these areas of low mean annual pressure possess the common characteristic of an excessive amount of moisture in the atmosphere. The Arctic and Antarctic zones of low pressure, and the equatorial low-pressure zone generally, may be regarded as all but wholly occasioned by the comparatively large amount of vapour in their atmosphere. As regards the region of low pressure of southern Asia in summer, it is remarkable that, while the eastern half which overspreads the valley of the Ganges is characterized by a moist atmosphere and large rainfall, the western half of it is singularly dry and practically rainless, and that the central portion of this remarkable depression occupies a region where at the time the climate is one of the driest and hottest anywhere to be found on the globe. Hence, while the vapour is the more important of the disturbing influences at work in the atmosphere, the temperature also plays no inconspicuous part directly in destroying atmospheric equilibrium, from which result winds, storms, and many other atmospheric changes.

The Prevailing Winds of the Globe.— If atmospheric pressure were equal in all parts of the earth we should have the physical conditions of a stagnant atmosphere. Such, however, is not the case. Let there be produced a concentration of aqueous vapour over a particular region, or let one region show a higher temperature than what prevails around it, then from the different densities, and consequently different pressures thereby produced, the equilibrium of the atmosphere is destroyed, and, as might be expected from the laws of aerial fluids, movements of the air, or winds, set in to restore the equilibrium. Now every one of the isobaric maps we have given, as well as every isobaric map which has been made from recorded observations, indicates very considerable disturbance of the equilibrium at the surface of the earth. All observation shows that the prevailing winds of any region at any season of the year are simply the expression of the atmospheric movements which result from the disturbance of the equilibrium of the atmosphere indicated by the isobaric maps for that season and region.

All winds maybe regarded as caused directly by differences of atmospheric pressure, just as the flow of rivers is caused by differences of level, the motion of the air and the motion of the water being both referable to gravitation. The wind blows from a region of higher towards a region of lower pressure,—in other words from where there is a surplus to where there is a deficiency of air; and this takes place whether the differences of pressure be measurable by the barometer, as is generally the case, or not readily measurable, as in the case of sea breezes, squalls, and sudden gusts of wind which are of short duration.

So far as is known, differences of atmospheric pressure,, and consequently all winds, originate in changes occurring either in the temperature or the humidity of the air over restricted regions. Thus, if two regions contiguous to each other come to be of unequal temperature, the air of the warmer region, being specifically lighter, will ascend, and the heavier air of the colder region will flow in below to take its place. Of this class of winds the sea and land breezes are the best examples. Again, if the air of one region comes to be more highly charged with aqueous vapour than the air of surrounding regions, the air of the more humid region being lighter will ascend, while the heavier air of the drier regions will flow in below and take its place. Since part of the vapour will be condensed into cloud or rain as it ascends, heat is thereby disengaged, and the equilibrium still further disturbed. In this way originate gales, storms, tempests, hurricanes, and all the more violent commotions of the atmosphere, except some of the forms of the whirlwind, such as dust storms, in the production of which very great differences of temperature are more immediately and exclusively concerned.

The Trade-Winds. — From fig. 14, giving the isobarics for January, it is seen that atmospheric pressure in the Atlantic is lower near the equator than it is to north and south of it; and the arrows indicate that to the north of the tract of lowest pressure N.E. winds prevail and to the south of it S.E. winds. These are the well-known N.E. and S.E. trade-winds, which thus blow from regions of high pressure towards the tract of lower pressure situated midway between them. The trade-winds do not blow directly to where the lowest pressure is, but in a slanting direction at an angle of about half a degree. The deviation from the direct course is due to the influence of the rotation of the earth on its axis from west to east,—an influence to which all winds and all currents of the ocean are subject.

In virtue of this rotation, objects on the earth’s surface at the equator are carried round towards the east at the rate of about 17 miles a minute. On receding from the equator, however, this rate of velocity is being continually diminished, so that at 60’ N. lat. it is only about 8⅜ miles a minute, and at the poles nothing. From this it follows that a wind blowing along the earth’s surface in the direction of the equator is constantly arriving at places which have a greater eastward velocity than itself. As the wind thus lags behind, these places come up, as it were, against it, the result being an east wind. Since, therefore, the wind north of the equator is under the influence of two forces—one, the low pressure near the equator, drawing it southwards, and the other, the rotation of the earth, deflecting it eastwards—it will, by the law of the composition of forces, take an intermediate direction, and blow from north-east. For the same reason, south of the equator the south is deflected into a south-east wind.

In the Atlantic the north trades prevail between latitudes 7° and 30° N., and the south trades between latitudes 3° N. and 25° S. These limits are not stationary, but follow the sun, being farthest to the south in February and to the north in August. The tract of low pressure between these wind systems is named the region of calms, owing to the calm weather which often prevails there, and it is also characterized by the frequent occurrence of heavy rains. This region of calms varies its position with that of the sun, reaching its most northern limit, lat. 1Γ N., in August, and its most southern, lat. l 0 N., in February. Its breadth varies from 3° to 8°, and it lies generally parallel to the equator. It is to be noted that, in the Atlantic, the region of calms is at all seasons north of the equator.

North and south trades also prevail in the Pacific Ocean, separated by a region of calms, which would appear, however, to be of less breadth and to be less clearly defined than is the region of calms in the Atlantic. In the eastern portion of the Pacific the region of calms lies at all seasons to the north of the equator, but in the western division it is considerably south of the equator during the summer months of the southern hemisphere, this southerly position being in all likelihood occasioned by the extraordinarily high pressure in Asia in its relations to the low pressure in the interior of Australia at this season. During the summer months of the northern hemisphere the region of calms wholly disappears from the Indian Ocean and from the western part of the Pacific Ocean, there being then an unbroken diminishing pressure from the latitude of Mauritius and Central Australia northwards as far as the low pressure of Central Asia.

Regions of light and variable winds and calms occur at the higher limits of the north and south trades. Except in the Pacific, where, owing to the greater breadth of that ocean, they spread over a considerable extent, these regions appear but in circumscribed patches, such as characterize the meteorology of the North and South Atlantic about latitudes 26° to 36°. Of these regions of calms the most important is that marked off by the high pressure in the North Atlantic, between the United States and Africa. This is the region of the Sargasso Sea, where the weather is characterized by calms and variable winds, and the ocean by its comparatively still waters. These are known to seamen as the “horse latitudes,” and are essentially different from the equatorial region of calms. The latter, as has been stated, is the region of low pressure at the meeting of the north and south trades, where the climate is distinguished for its general sunlessness and heavy rainfall. On the other hand, the calm regions in the Atlantic and Pacific Oceans about the tropics have an atmospheric pressure abnormally high, clear skies, and the weather generally sunny and bright, with occasional squalls.

Numerous observations made in all parts of the globe establish the fact that, while the surface winds within the tropics are directed towards the equatorial region of calms in such a manner that the general intertropical movements of the atmosphere or prevailing winds are easterly, the prevailing winds of the north and south temperate zones are westerly. The westing of these great aerial currents is due to the same cause that gives easting to the tradewinds, viz., the rotation of the earth round its axis. For, as an aerial current advances into higher latitudes, it is constantly arriving at regions having a less rotatory velocity than itself; it thus outstrips them and leaves them behind; in other words, it blows over these places as a westerly wind.

While, however, the general prevalence of westerly winds has been established over the extratropical regions of Europe, Asia, Africa, America, and Australia, the directions which in different seasons and at different places are actually found to prevail often differ very widely from west. An examination of the winds at one hundred and fifteen places pretty well distributed over the northern hemisphere reveals the instructive fact that almost every place shows two maximum directions from which winds blow more frequently than from the other directions, and that one of these two directions shows a considerable excess over the other. Thus, for example, the following are, on a twenty years’ average, the number of days at Greenwich each wind prevails during the year:—N., 41; N.E., 49; E., 23; S.E., 21; S., 34; S.W., 103; W., 38; N.W., 24; and calms, 32. Hence S.W. and N.E. winds are there more prevalent than winds from any other direction, and of these two winds the greater maximum direction is S.W. If the two maximum directions be sorted into groups, then the greater maximum direction occurs as follows :— [table]

and the other maximum direction is [table]

This result of observation, so different from what was long accepted as being in accordance with the generally received theory of the movements of the atmosphere, teaches the important lesson that the region towards which the extratropical winds of the northern hemisphere are directed is not the region of the north pole.

Prevailing Winds in January.— On examining fig. 14, which shows the distribution of atmospheric pressure in January, it is seen that pressure is abnormally low over the northern portion of the Atlantic—the lowest occurring between Iceland and South Greenland—from which it rises as we proceed in a S.W. direction towards America, in a S. direction over the Atlantic, and in a S.E. and E. direction over Europe and Asia. Now what influence has this remarkable atmospheric depression on the prevailing winds over this large and important part of the earth’s surface? The arrows in the figure, which indicate the prevailing winds, and which have been laid down from observations, answer this question.

At stations on the east side of North America the arrows show a decided predominance of north-west winds; at the more northern places the general direction is more northerly, whereas farther south it is more westerly. In the Atlantic between America and Great Britain, in the south of England, in France and Belgium, the direction is nearly S.W. In Ireland and Scotland it is W.S.W.; in Denmark and the north-west of Russia S.S.W.; from St Petersburg to Tobolsk S.W.; on the west of Norway generally S.S.E.; and in Greenland, the north of Iceland, and about Spitzbergen N.E. Hence all the prevailing winds in January over this extensive portion of the globe may be regarded as the simple expression of the difference of atmospheric pressure which prevails over the different parts of the region. In truth the whole appears to flow vorticosely, or in an in-moving spiral course, towards the region of low pressure lying to the south-west of Iceland, and extending eastward over the Arctic Sea north of Russia. The only marked changes in these directions of the wind thus broadly sketched out are the deflexions caused by the various mountain systems which lie, so to speak, embedded in these vast aerial currents; of these the winds in the south of Norway afford excellent illustrations.

The influence which this peculiar distribution of the pressure over the north of the Atlantic exercises in absolutely determining the winter climates of the respective countries is most instructive. It is to this low pressure, which draws over the British Islands W.S.W. winds from the warm waters of the Atlantic, that the open, mild, and, it must be added, rainy winters of these islands are due. The same region of low pressure gives Russia and Western Siberia their severe winters; and it is the same consideration that fully explains the enormous deflexion of the isothermal lines from Norway eastwards and south-eastwards over the Old Continent. Finally, the same low pressure draws over British America and the United States, by the N.W. winds which it induces, the intensely dry cold air-current of the Arctic regions. At Portland, Maine, which is swept by these cold north-westerly winds, the normal temperature in January is 23 0, 6, whereas at Corunna, on the coast of Spain, in nearly the same latitude, where south-westerly winds from the Atlantic prevail, the mean temperature of the month is 49° *l, or 25 o, 5 higher.

The region of low atmospheric pressure in the north of the Pacific is accompanied by prevailing winds over the region embraced by it and by climatic effects in all respects similar to the above. In Vancouver Island the prevailing winds in January are S.W., at Sitka E.S.E., on Great Bear Lake E.N.E., in Alaska N.E., in Kamchatka N.N.E., and in Japan N.W. In accordance with these winds the winter climate of Vancouver and adjoining regions is mild and humid, and that of the north-east of Asia dry and intensely cold.

On the other hand, abnormally high pressure rules over the continent of Asia at this season, and as regards this region of high pressure the arrows represent the winds as blowing outwards from it in all directions. Over the interior of Asia, where the highest normal pressures are, observations show a marked prevalence of calms and light winds, but around this central region the prevailing winds in January are—at Calcutta N., at Hong-Kong E.N.E., at Peking N.W., on the Amur W.N.W., S.E. at Nijni-kolynisk and S.S.W. at Ustjansk (in the north of Siberia), and at Bogoslovsk S.W. Hence from this extensive region, where pressure is abnormally high, or where at this season there is a large surplus of air, the prevailing winds flow outwards in all directions towards the lower pressure which surrounds it. Owing to the excessive dryness of the air of Central Asia, terrestrial radiation is less obstructed there than anywhere else on the globe, and consequently the temperature falls very low, the mean of January at Werchojansk being — 55 0- 8, which is the lowest mean monthly temperature known to occur on the earth’s surface. And, since the winds blow outwards from the dry cold climates of the interior, temperatures are low, even on the coasts. Of this China affords good illustrations. Thus the mean January temperature of Peking is 22 o, 7 and of Zi-ka-Wei, near Shanghai, 35 o, 4, whereas at Corfu and Alexandria the normal temperatures for January are respectively 50 o- 9 and 58° O, or 28° ^2 and 22 o, 6 higher than in corresponding latitudes on the coast of China.

The winds of the United States in winter, taken in connexion with the peculiar distribution of pressure already described, are very interesting. There are two regions of high pressure, one in the south-eastern States and the other and larger one in the States around Utah; and between these there is interposed a trough of lower pressure extending from Chicago to the south-west of Texas. On the western side of this depression the winds are northwesterly, but to the east of it they become W., W.S.W., and in some places S.W., and again on nearing the Atlantic seaboard they become north-westerly. In connexion with the region of higher pressure in the west, the prevailing winds are seen to flow outward from it. The normal pressure diminishes everywhere to southward of a line drawn from the Canaries to Bermuda, thence westward in nearly the same latitude to Texas, and then to west-northwest to San Francisco. The tract of lowest pressure stretches from the basin of the Amazon in the direction of the isthmus of Panama in about latitude 8° N., and thence is continued westward for a considerable distance into the Pacific in nearly the same latitude. It follows from this distribution of the pressure that the north trades in a more or less modified form prevail over South America to the north of the Amazon, and in the Pacific to the north of lat. 8° N., probably as far to westward as long. 150° W.

The low-pressure systems which prevail during the summer months in South America and South Africa have each its corresponding system of winds all round. It is, however, in Australia, as being the most compact and isolated continent, that the influence of the summer sun in lowering the pressure is best illustrated. In that continent the lowest pressure occurs in the region situated about midway between the north coast and the tropic of Capricorn, over which the normal pressure does not exceed 29 ’ 80 inches. Further, everywhere in Australia pressure diminishes from the coast on advancing upon the inland districts. It follows from this disposition of the pressure that all round the island the prevailing winds in summer blow from the sea towards the interior; and accordingly it is in these months that the greater part of the rain falls. From the low pressure of the interior southwards to Bass’s Straits pressure rises continuously, the increase in the normal over this space being about 0 ’ 200 inch. To northward it also rises continuously to beyond the north of China, the increase on this side being about f of an inch. In this case the greater part of the increase occurs over the continent, the rate of increase from the north of Australia to the Philippine Islands being only about the rate of increase which obtains southward towards Bass’s Straits. It will be shown when the subject of the rainfall is examined that it is the relative excess of these high pressures, the one in the south of Australia and the other in the south-east of Asia, that determines the position of the area of low pressure in Australia in particular years, and with that position the degree and extent to which the whole of the northern portion of Australia is watered by the rainfall. Thus, when pressure is more than usually high in the south-east of Asia, and either low or not excessive in the south of Australia, then the low-pressure region is pushed farther southward into the interior, and with it the rainfall spreads inland over a wider area and to a greater depth.

Prevailing Winds in July.— In the winter of the southern hemisphere, the geographical distribution of pressure is exactly the reverse in Australia of what obtains during the summer months. Everywhere all round it increases on advancing from the coast into inland districts. The lowest pressure, about 30 Ό0 inches, occurs on the north coast, and the highest over the basin of the Murray river and its affluents, where it rises generally to 30 ’ 18 inches. On the south coast it is generally about 30 - 12 inches, falling, however, at Gabo Island, in the extreme south-east, to 30Ό50 inches, and to 29 ’ 836 in the south of New Zealand. From the Murray river the diminution of pressure is continuous to the north, even to the low pressure of Central Asia. From this arrangement of the pressure, the prevailing winds blow from the interior towards the surrounding ocean all round Australia, with the single exception of the extreme south-west of the continent, where the prevailing winds are south-westerly, being here essentially an outflow of the high pressure which overspreads the Indian Ocean to the westward. As these S.W. winds are from the ocean, the rainfall at Perth in July is fully 6 inches, and it is high over south-western districts of West Australia. The prevailing winds round Australia are S.E. on the north coast, S.W. at Brisbane, W.N.W. at Sydney, N. at Melbourne, N.E. at Adelaide. These all represent an outflow from the high-pressure regions of the interior modified by the influence of the earth’s rotation, and, in correspondence with the reversal of the distribution of the pressure, are directions the reverse of the prevailing winds of January.

In July the central and southern parts of Asia are highly heated by the summer sun, and, besides, the rainfall over southern parts is excessive. Consequently atmospheric pressure is very low, being fully 0'40 inch lower in the Punjab than it is in the south of Ceylon. From the interior pressure rises continuously on advancing to the eastward, southward, westward, and northward, and from all these directions the prevailing winds of summer flow inwards upon the interior, and these bring rain or parching drought according to the vapour they bring from the ocean they have traversed, and according as they advance into warmer or colder regions. The prevailing summer winds of Asia, being an inflow inwards upon the interior, have, generally speaking, exactly the reverse direction of that prevailing in winter.

The winds of Europe are mainly determined by the extraordinarily high pressure of the Atlantic in its relations to the low-pressure systems of Central Asia and Central Africa at this time. The winds in the Spanish Peninsula are north-west; in the north of Africa they are northerly, and again north-westerly in Syria. The winds of the British Islands and western Europe have less southing and more northing than the prevailing winds of winter, and to the east of long. 40° E. they become decidedly north-west. It is to the Atlantic origin of these winds that the summer climates of these large and important regions owe the comparatively large rainfall of this season, it being at this time that the rainfall reaches the annual maximum. The bearing of the low-pressure areas and mountain systems of the north of Italy and Scandinavia on the climates of these countries will be afterwards referred to.

The centre of lowest pressure in North America is over the central States about Utah, from which it rises all round, least to northward and most in south-easterly and north-westerly directions. In California N.W. winds necessarily blow inwards upon this central low-pressure area; and, as these winds pass successively over regions the temperature of which constantly increases, the summer climate is rainless. On the other hand, southerly and south-easterly winds from the Gulf of Mexico blow up the western side of the basin of the Mississippi inwards upon the low-pressure area of the centre, depositing in their course, in a rainfall more or less abundant, the moisture they have brought from the Gulf. To the north of lat 50°, and to westward of Hudson’s Bay, the prevailing winds become easterly and north-easterly, distributing over Manitoba, Saskatchewan, and neighbouring regions, as they continue their westerly course towards the low-pressure area, the rainfall they have transported thither from the wide expanse of Hudson’s Bay. An attentive examination of the arrows of fig. 17 shows that the prevailing winds over all the States to the east of the Mississippi river are rather to be regarded as an outflow from the region of very high pressure over the Atlantic to southeastward. Thus in Florida the winds are S.E., in the southern States S., and in the lake region, in the New England States, and on the Atlantic seaboard S.W. Since the origin of these winds is thus essentially oceanic, and since in their course northwards no mountain range crosses their path, the whole of this extensive region enjoys a large but by no means excessive rainfall, which, taken in connexion with the temperature, renders the summer climate of these States one of the best to be met with anywhere on the globe for the successful prosecution of agricultural industries.

The remarkable protrusion of high pressures from the southern hemisphere, where they are massed at this time of the year, northwards into the Atlantic is, as has already been referred to, one of the outstanding features of the meteorology of the summer months of the northern hemisphere. In the central area of this large region the climate is remarkable for its prevailing calms, light winds, occasional squalls, and clear skies. From this comparatively calm space the wind blows outwards in all directions towards and in upon the surrounding regions of low pressure. These winds, owing to the high temperature, clear skies, and strong sunshine of the region from which they issue, carry with them a great amount of vapour near the surface, by which to a large extent the north of South America, the east of North America, the greater part of Europe, and a large portion of Africa are watered. The prevailing winds over this region are further interesting, not merely from the striking illustration they give of the intimate relation of the winds to the distribution of the pressure, but as being of no small importance in determining the best routes to be taken over this great highway of commerce, and the more so inasmuch as the currents of the ocean are coincident with these prevailing winds.

In the Antarctic regions, or rather to the south of lat. 45° S., the normal atmospheric pressure is low at all seasons, there being a gradual diminution of pressure to 29 - 20 inches about lat. 60° S. Pressure is probably even still lower nearer to the south pole, as seems to be indicated by the observations made by Sir James C. Ross, and in the “Challenger” and other expeditions. Over this zone the prevailing winds are W.N.W. and N.W. This is the region of the “brave west winds,” the “roaring forties” of sailors, which play such an important part in navigation, and which determine that the outward voyage to Australia be round the Cape of Good Hope and thence eastward, and the homeward voyage eastward round Cape Horn, the globe being thus circumnavigated by the double voyage. That the general drift of these winds is inwards upon the south pole is strongly attested by the existence of the enormously thick wall of ice which engirdles these regions, from which are constantly breaking away the innumerable icebergs that cover the Southern Ocean, none of which is. ever seen of a calculated thickness less than 1400 feet. The snow and rainfall which must take place in the south polar regions for the formation of icebergs of such a thickness must be peculiarly heavy, but not heavier than might be expected from the strength and degree of saturation of the “roaring forties” which unceasingly precipitate their moisture over these regions.

To sum up :—so far as the prevailing winds are concerned, it has been shown that where pressure is high, that is to say, where there exists a surplus of air, out of such a region winds blow in all directions; and, on the other hand, where pressure is low, or where there is a deficiency of air, towards such a region winds blow from all directions in an in-moving spiral course. This outflow of air-currents from a region of high pressure upon a region of low pressure is reducible to a single principle, viz., the principle of gravitation. Given as observed facts the differences of pressure, it is easy to state with a close approximation to accuracy what are the prevailing winds, before calculating the averages from the wind observations. Indeed so predominating is the influence of gravitation where differences of pressure, however produced, exist that it may practically be regarded as the sole force immediately concerned in causing the movements of the atmosphere. If there be any other force or forces that set the winds in motion independently of the force called into play by differences of mass or pressure, their influence must be altogether insignificant as compared with gravitation.

It has been abundantly proved that the wind does not blow directly from the region of high towards that of low pressure, but that, in the northern hemisphere, the region of lowest pressure is to the left of the direction towards which the wind blows, and in the southern hemisphere to the right of it. This direction of the prevailing wind with reference to the pressure is in strict accordance with Buys-Ballot’s Law of the Winds, which may be thus expressed :— the wind neither blows round the centre of lowest pressure in circles, or as tangents to the concentric isobaric curves of storms or cyclones, nor does it blow directly towards the centre; but it takes a direction intermediate, approaching, however, more nearly to the direction and course of the circular curves than of the radii to the centre. The angle formed by a line drawn to the centre of lowest pressure from the observer’s position and a line drawn in the direction of the wind is not a right angle, but an angle of from 60° to 80°.

From its importance in practical meteorology Buys Ballot’s law may be stated in these two convenient forms. (1) Stand with your back to the wind, and the centre of the depression or the place where the barometer is lowest will be to your left in the northern hemisphere, and to your right in the southern hemisphere. This is the rule for sailors by which they are guided to steer with reference to storms. (2) Stand with the high barometer to your right and the low barometer to your left, and the wind will blow on your back, these positions in the southern hemisphere being reversed. It is in this form that the prevailing winds of any part of the globe may be worked out from the isobaric charts (figs. 14 and 17).

From the all-important consequences which flow from the geographical distribution of the pressure it is evident that the regions of low and of high normal pressure must be regarded as the true poles of the prevailing winds on the earth’s surface, towards which and from which the great movements of the atmosphere proceed. From the unequal distribution of land and water, and their different relations to solar and terrestrial radiation, it follows that the poles of pressure and of atmospheric movements are, just as happens with respect to the poles of temperature, very far from being coincident with the north pole. Thus during the winter months the regions to which the origin of the great prevailing winds of the northern hemisphere are to be referred are Central Asia, the region of the Rocky Mountains, and the horse latitudes of the Atlantic, and the regions towards and in upon which they flow are the low-pressure systems in the north of the Atlantic and Pacific Oceans, and the tract of low pressure within the tropics towards which the trade-winds blow. In the summer months the reversed conditions of pressure-distribution then observed are attended with corresponding changes in the prevailing winds; and, generally speaking, if the south polar region be excepted, the poles of highest and lowest pressure and atmospheric movements are at no time coincident with the north pole. It is this consideration which affords the true explanation why prevailing winds at so large a proportion of stations in the northern hemisphere do not blow in the directions in which true equatorial and polar winds should blow.

The causes which bring about an unequal distribution of the mass of the earth’s atmosphere are mainly these two— the temperature and the moisture of the atmosphere considered with respect to the geographical distribution of land and water. Owing to the very different relations of land and water to temperature, as already stated, the summer temperature of continents greatly exceeds that of the ocean in the same latitudes. Hence the abnormally high temperature which prevails in the interior of Asia, Africa, America, and Australia during their respective summers, in consequence of which the air, becoming specifically lighter, ascends in enormous columns thousands of miles in diameter. On arriving at the higher regions of the atmosphere it flows over neighbouring regions where the surface temperaturé is lower, and thus the atmospheric pressure of the highly heated regions is diminished.

Surface winds set in all round to take the place of the air removed from the continents by these ascending currents, and since these necessarily are chiefly winds from the ocean they are highly charged with aqueous vapour, by the presence of which, and by the condensation of the vapour into cloud and rain, the pressure over continents at this season is still further and very largely diminished. Air charged with vapour is specifically lighter than when without the vapour; in other words, the more vapour any given quantity of atmospheric air has in it the less is its specific gravity; and, further, the condensation of vapour in ascending air is the chief cause of the cooling effect being so much less than that which would be experienced by dry air. From these two principles, which were established by Dalton, Joule, and Sir William Thomson, it follows that the pressure of vapour in the air, and its condensation, exercise a powerful influence in diminishing the pressure. The great disturbing influences at work in the atmosphere are the forces called into play by its aqueous vapour; and it is to these, co-operating with the forces called into play by the differences of temperature directly, that the low normal pressure of the continents during the summer is to be ascribed. The degree to which the lowering of the pressure takes place is, as was to have been expected, greatest in Asia, the largest continent, and least in Australia, the smallest continent, while in America it is intermediate.

The influence of the aqueous vapour in diminishing the pressure is well seen in the belt of calms in the tropics between the north and the south trade-winds. Since these winds import into the belt of calms the vapour they have taken up from the sea on their way thither, the climate is characterized by a highly saturated atmosphere and heavy rains. Again the air in regions near the Atlantic contains much more vapour and is of a higher temperature during winter than is observed at places in the interior of continents in the same latitudes. It follows thus that the air over the north of the Atlantic and the regions adjoining is specifically lighter than in the regions which surround them. We have here therefore the physical conditions of an ascending current; and it is plain that the strength of this current will not merely be kept up but increased by the condensations of the vapour into cloud and rain which take place within it, by which a higher temperature and a greater specific lightness are maintained at the surface of the earth and at various heights in the atmosphere than exist over surrounding regions at the same heights. Accordingly it is seen from the winter isobars that an enormous diminution of pressure occurs over these regions, and also over the north of the Pacific and the Antarctic, as compared with the continents.

Since, on the other hand, dry and cold air is specifically heavy, the winter isobars show that where temperature is low and the air very dry pressure is high. Of this Asia and North America are striking examples during December, January, and February, and Australia, South Africa, and South America during June, July, and August.

Since vast volumes of air are thus poured into the region where pressure is low without increasing that pressure, and vast volumes flow out of the region where pressure is high without diminishing that pressure, it necessarily follows that the volumes of air poured into the region of low normal pressure do not accumulate over that region, but must somehow escape away into other regions, and that the volumes of air which flow out from the region of high normal pressure must have their place supplied by fresh accessions of air poured in from above. That the same law of relation observed between sea-level pressures and surface winds obtains between pressures at different heights and winds at the same heights is simply a necessary inference. We are therefore justified in expecting that ascending currents will continue their ascent till a height is attained at which the pressure of the air composing the currents equals or just falls short of the pressure over the surrounding regions at that high level. On reaching this height the air, no longer buoyed up by a greater specific levity than that of the surrounding air, will cease to ascend, and expanding horizontally will thenceforth flow over as an upper current towards those regions which offer the least resistance to its course; that is to say, it will flow over upon those regions where, at that height, pressure happens at the time to be least. Now from the known densities of air of different temperatures and humidities it is evident that the overflow of the upper current will take place towards and over that region or regions the air of which in the lower strata of the atmosphere happens to be colder and drier than that of the other surrounding regions,—because, being denser, a greater mass of air is condensed or gathered together in the lower strata of the atmosphere, thus leaving a less mass of air, or a diminished pressure, in the higher region of the upper current.

If this be so, then the extraordinarily high pressure of Central Asia during winter is to be ascribed to these two causes :—(1) the low temperature and excessive dryness of the air of this extensive region; and (2) its relative proximity to the low pressure of the Atlantic to the northwest, the low pressure of the Pacific to the north-east, and the low pressure of the belt of calms to the south. Similarly, since in summer the temperature of air resting over the Atlantic between Africa and the United States is much lower than that of the land, the ascending currents which arise from the heated lands of Africa, Europe, and North and South America, as well as from the region of calms immediately to the south, all of which are remarkable for a low normal pressure, will on reaching the upper regions of the atmosphere flow towards this part of the Atlantic, because there, the temperature being lower and the density of the air composing the lower strata being greater, pressure in the upper regions is less. And, since the surface winds are constantly flowing outwards from this region of abnormally high pressure, thus draining away the air poured down upon it by the upper currents which converge upon it, extreme saturation does not take place, and the air consequently is relatively dry and cool. That this view generally represents the movements of the upper currents has been strongly confirmed within the last few years by Professor Hildebrandsson and Clement Ley in their researches into the upper currents of the atmosphere based on observations of the cirrus cloud.

From these considerations it may be concluded that the winds which prevail near the earth’s surface are known from the isobaric lines, the direction of the wind being from regions where pressure is high towards regions where it is low, in accordance with Buys Ballot’s law; and that the upper currents may be inferred from the isobaric lines taken reversely, together with the isothermal lines taken directly. In other words, the regions of lowest pressure, with their ascending currents and relatively higher pressure at great heights as compared with surrounding regions, point out the sources or fountains from which the upper currents flow; and the isothermals, by showing where on account of the relatively low temperatures the greater mass of the air is condensed in the lower strata of the atmosphere and sea-level pressure consequently is high, thus diminishing the pressure of the upper regions, point out the regions towards and upon which these upper currents of the atmosphere flow. The facts of the diurnal oscillations of the barometer in the different regions already discussed afford the strongest corroboration of these views.

The term “monsoon” has long been applied to the prevailing winds in southern Asia which blow approximately from S.W. from April to October, and from N.E. from November to April. The term is now, however, generally applied to those winds connected with continents which are of seasonal occurrence, or which occur regularly with the periodical return of the season. Since they are caused immediately by the different temperatures and pressures which form marked features of the climates of continents in winter and summer respectively, they are most fully developed round the coast of Asia, owing to the great extent of that continent. The monsoons of different parts of the coasts of Asia differ widely in direction from each other. Thus in winter and summer respectively they are W.N.W. and E.N.E. at the mouth of the Amur, N. and S.S.E. at Shanghai, N.E. and S.W. at Rangoon, N. and W.S.W. at Bombay, N.W. and S.W. at Jerusalem, and S.S.W. and N.N.E. at Archangel. The Indian winter monsoon generally begins to break up in March, but it is not till about the middle of May, when the normal pressure has been decidedly diminished over the heated interior, that the summer monsoon acquires its full strength and the heavy monsoonal rains fairly set in. In October, when the temperature has fallen considerably and with the falling temperature the pressure of the interior has risen, the summer monsoon begins to break up, and this season is marked by variable winds, calms, and destructive hurricanes. As the temperature continues to fall and pressure to rise, the winter monsoon again resumes its sway. Monsoons, equally with the trade-winds, play a most important part in the economy of the globe. The relatively great force and steadiness in the direction in which they blow, and the periodical change in their direction, give facility of intercourse between different countries; and, besides, by the rainfall they bring they spread fertility over extensive regions which otherwise would be barren wastes.

The winds of Australia are also strictly monsoonal, but owing to the small extent of that continent, and consequently the smaller differences there are between the normal pressure of the interior and that of the surrounding coasts in summer and winter respectively, they are less strongly marked than are the monsoons of southern Asia; and particularly they neither blow with the same force nor so steadily from the same point of the compass. For the same reason the Australian climates are characterized by the occurrence of more frequent droughts than are the climates of southern Asia, and the same remark applies to the climates of southern Africa.

Since the Malay archipelago lies during the summer of the northern hemisphere between the high pressure of central Australia and the low pressure of Asia, and during the winter between the high pressure of Asia and the low pressure of central Australia, it follows that the winds of these islands are eminently monsoonal in their character, being in summer southerly and in winter northerly. The result of this peculiar wind system of the archipelago is to give to these islands a singular diversity of climates, which will be more particularly referred to under rainfall.

At Zanzibar the prevailing wind in July is S.E., but in January, when the low pressure of the interior is situated much farther to southward, it is N.E.: and the same influence is felt, though in a greatly modified degree, as far as Mauritius, where the S.E. trade changes nearly into E. during the summer. On the other side of Africa the S.E. trade of the South Atlantic is changed into a S.W. monsoon on the coast of the Gulf of Guinea.

In the southern, central, western, and northern regions of North America the prevailing winds have a well-marked monsoonal character. The prevailing winds of winter and summer respectively are N.E. and S.S.E. at New Orleans, N.W. and S.W. in Utah, N. and S. at Fort Yuma (California), E.S.E. and N.W. at Portland (Oregon), and S. and E.N.E. at Fort York, Hudson Bay. These winds are readily accounted for by the distribution of pressure over the continent in winter and summer. On the Atlantic seaboard of the United States the prevailing winds of winter vary from N.W. in the New England States to W. in South Carolina; whereas in summer they vary generally from S.S.W. in South Carolina to S.W. in the New England States. Hence over the eastern States the summer winds are not directed towards the low-pressure region of the interior of the continent, but are determined by the relations of their pressure to the high pressure of the Atlantic to the eastward, and to the lower pressure overspreading the Atlantic to the N.E. This influence of the Atlantic may be considered as felt westward through the States as far as the Mississippi.

Though not so decidedly marked, the winds of Europe, except the extreme south, are also monsoonal. In winter they flow from the land towards the region of low pressure in the north of the Atlantic; but in summer the arrows, representing the prevailing winds, show that all but the extreme south of Europe is swept by westerly winds, which flow in a vast continuous stream from the Atlantic towards the central regions of the Old Continent, and which deposit in their course the rains they have brought from the ocean.

Similarly, monsoons prevail on the coasts of Brazil, Peru, North Africa, and many other regions which happen to lie between other regions whose temperatures, and therefore pressures, differ markedly from each other at different times of the year.

These are the chief prevailing winds of the globe when the differences of the normal atmospheric pressure are such as to cause a decided and steady movement of the atmosphere over a large portion of the earth’s surface, resulting in well-marked prevailing winds. But there are other winds which are greatly influenced by local causes, such as the nature of the ground, whether covered with vegetation or bare; the physical configuration of the surface, whether level or mountainous; and the vicinity of extensive sheets of fresh or salt water. An important characteristic of winds in their practical relations to climate is their quality,—they being warm or cold, dry or moist, according to their direction and the nature of the earth’s surface over which they have just passed. Thus in the northern hemisphere southerly winds are warm and moist, while northerly winds are cold and dry. In Europe southwesterly winds are moist and easterly winds dry, while in the New England States and Canada north-easterly winds are cold and raw and north-westerly winds cold and dry.

In particular regions certain meteorological conditions occur at stated seasons intensifying these effects, resulting in excessive drought, heavy rains, intense or great heat, thus giving rise to the following among other well-known winds. The east winds of the British Islands occur chiefly in spring, but also in a less degree in November, being in the latter case often accompanied with fog. The winds here referred to are dry and parching, and their deleterious influence on the health is seen, not merely in the discomfort and uneasiness they impart to the less robust of the population, but also in the largely increased mortality which they cause from consumption and all other diseases more or less connected with the nervous system. In the countries bordering on the north of the Atlantic, atmospheric pressure reaches the annual maximum in May, and it is above the average during the other two spring months. In these months the normal pressure approaches nearer to what obtains farther south, and an examination of daily weather maps shows that this is due to the repeated occurrence in spring of very high pressures in the north of

Europe while pressures much lower prevail to southward. Now these east winds are simply the outflow from these regions of high pressure to northward. Northerly and even westerly winds which are truly outflows from what may be styled Arctic anticyclonic areas bring with them qualities as noxious as those of the east wind itself, and prove as injurious to health and vegetation. The cold dry wind of April 29,1868, which blasted and shrivelled up vegetation in Scotland, particularly in the western counties, as effectually as if a scorching fire had passed across them, was a west wind.

In the south of Europe, during the winter and early spring, peculiarly dry, cold, and violent northerly winds are of occasional occurrence. Of these winds the “mistral” is one of the most notorious, which is a steady, violent, and cold north-west wind blowing from central and eastern France down on the Gulf of Lyons. It is particularly trying while it lasts to invalids who are spending the winter at the various popular sanataria which are scattered along this part of the Mediterranean coast. The great cold that took place in the north of Italy and south of France in the beginning of 1868 was a good example of the mistral. The meteorological conditions under which it occurred were unusually low pressure over the Mediterranean to southward (29 - 450 inches), whilst at the same time pressure rose steadily and rapidly on proceeding northward to 30 ’ 905 inches in the north of Russia. From this geographical distribution of the pressure, northerly winds swept southwards over Europe, carrying with them the low temperatures of the higher latitudes, and became still colder and drier on crossing the Alps before they made the descent on the shores of the Mediterranean. The cold tempestuous winds which descend from the Julian Alps and sweep over the Adriatic, and the dreaded “gregale” of Malta, which is a dry cold north-east wind, are in their character and origin quite analogous to the mistral.

The “northers,” or “nortes,” are peculiarly dry cold strong winds which repeatedly occur from September to March in the States bordering on the Gulf of Mexico, and are perfectly analogous to the mistral. The conditions under which they occur are a pressure lower than usual to the south or south-east over the Gulf of Mexico, together with a pressure even higher than the high normal which is so marked a feature of the meteorology of the Rocky Mountains during the colder months. When, as most frequently happens, they occur in the wake of a storm, their disagreeable qualities of extreme dryness, cold, and violence are all intensified. From a temperature of upwards of 80° experienced as the storm comes up the thermometer rapidly falls to 18° or even lower; and, as the low temperature often occurs with a wind blowing with great violence, the northers prove most deleterious. A violent wind with a temperature of 18° is altogether unknown in the British Islands.

The “pampero” is a strong, dry, cold wind which blows across the pampas of the River Plate of South America, occurring at all seasons, but most frequently during the spring and summer from October to January. They are preceded by easterly winds, a falling pressure, a rising temperature, and increased moisture. A pampero is described by Dr D. Christison, and its appearance figured, in the Journal of the Scottish Meteorological Society, vol. v. p. 342, as seen advancing on the morning of November 28, 1867, in central Uruguay. In the early morning the wind blew rather strongly from north-east, and by and by clouds were seen moving very slowly from the west, throwing out long streamers eastwards. As they advanced, two dense and perfectly regular cloud-masses appeared in front, one behind the other, in close contact yet not intermingling,—the one being of a uniform leaden grey, while the other was as black as the smoke of a steamer. On arriving overhead, the front, though slightly wavy in appearance, was seen to be quite straight in its general direction, and the bands were of uniform breadth. They rushed forward at great speed under the other clouds without uniting with them, preserving their forms unbroken, being borne onward by an apparently irresistible force, as if composed of some solid material rather than vapour. They extended probably 50 miles in length, but as they took only a few minutes to pass their breadth was not great, and they appeared to diminish to mere lines in the distant horizon. At the instant the first cloudband arrived overhead, the wind chopped round from north-east to north and then to south-west; a strong cold blast at the same time seemed to fall from the leaden cloud, and continued to blow till both bands had passed. No rain or thunder occurred at this time, but in the confused rabble of clouds which followed low thunder continued to roll, and in a quarter of an hour rain fell, and for some hours thereafter wind, rain, and thunder continued, but only to a moderate degree. The low temperature and rising barometer and change of wind are the constant and most striking characteristics of the pampero. On one occasion the temperature fell 44° in fourteen hours, and on another occasion the fall was only 4 β . Rain is a usual accompaniment, but on rare occasions the pampero passes off and no rain falls.

Rainfall.— Whatever tends to lower the temperature of the air below the dew-point is a cause of rain. It is therefore to the winds we must chiefly look for an explanation of the rainfall, and the broad principles of the connexion may be stated to be these five :—(1) when the winds have previously traversed a considerable extent of ocean, the rainfall is moderately large; (2) if the winds advance at the same time into colder regions, the rainfall is largely increased, because the temperature is sooner reduced below the point of saturation; (3) if the winds, though arriving from the ocean, have not traversed a considerable extent of it, the rainfall is not large; (4) if the winds, even though having traversed a large extent of ocean, yet on arriving at the land proceed into lower latitudes or regions markedly warmer, the rainfall is small or nil; (5) if a range of mountains lies across the onward path of the winds, the rainfall is largely increased on the side facing the winds, and reduced over the regions on the other side of the range. The reason here is that, the air on the windward side of the ridge being suddenly raised to a greater height in crossing the range, the temperature is further reduced by mere expansion, and a more copious precipitation is the result; whereas on the leeward side as the air descends to lower levels it becomes gradually drier, and accordingly the rainfall rapidly diminishes with the descent.

We have drawn attention to the diminished velocity of the wind over land as compared with the open sea (p. 125). From this it follows that an envelope of stiller air or air of less velocity than that of the prevailing wind broods over the land, and by its presence forces the prevailing wind to a greater height, thus tending to increase the rainfall. If the foreshore rises within a few miles to a height of 200 or 300 feet, the result is very striking when the wind from the sea blows straight upon it. Thus at Spittal, near Berwick, on September 1877, a N.E. wind blew straight ashore at an estimated velocity of 25 miles an hour. To eastward the sky was singularly clear down to the horizon, but to westward all the country beyond a mile from the shore was enveloped in what appeared a dense mist or fog. About 15° to eastward of the zenith of an observer on the shore, the thinnest rack of cloudlets was seen emerging without intermission from the deep stainless blue of the sky, which as they drifted landward increased so rapidly in volume and density that the zenith was three-fourths covered with clouds. A similar phenomenon was seen in September 1879 on board the Orkney steamer at the magnificent cliff of Hoy Island, Orkney. A heavy storm had just cleared away, and a strong W.N.W. wind was blowing right against the cliff. The sky was absolutely cloudless all round, except the upper 300 feet of Hoy Hill, 1570 feet high, which was enveloped in a thick mist that stretched away to windward, some distance to westward of the steamer’s course, which was about 2 miles from land. The western termination of the cloud was the thinnest rack of cloud, which emerged unceasingly from the blue sky at a distance not less than 4 miles to windward of the cliff. The constituent parts of the cloud itself were in rapid motion eastward, but, owing to the fresh accessions it was constantly receiving, the cloud itself appeared stationary. Thus the wind was forced upward into the atmosphere for some considerable distance to windward of the ridge lying across its path.

It is this dragging effect of the land on the wind, and the consequences which result from it, that explain how it is that during storms of wind and rain from the north-east the rainfall over the foreshores of the Firth of Forth, the Moray Firth, and the Pentland Firth looking to the north-east is so much in excess as compared with the rest of Scotland. The same principle explains the heavy rainfall in plains at some distance from the range of hills lying across the wind’s path and on the side of the rain-bringing winds.

For short intervals of time the heaviest rainfalls occur with tornadoes, waterspouts, and some other forms of the whirlwind, the reason being that not only is there rapid expansion due to the rapid ascent of the air, but also great rarefaction is produced by the extreme velocity of the aerial gyrations round the axis of the tornado. On August 1, 1846, 3Ί2 inches of rain fell at Camberwell, London, in two hours and seventeen minutes. Of heavy falls may be mentioned 4'60 inches in London, April 13, 1878; 6Ό0 inches at Tongue, September 7, 1870; 5'36 inches in Monmouthshire, July 14, 1875; 6'62 inches at Seathwaite, Cumberland, November 27, 1848; and 7*12 inches at Drishaig, Argyllshire, December 7 to 8, 1863. But it is in lower latitudes that the heaviest single showers have been recorded. The following are among the most remarkable :—at Joyeuse, France, 31'17 inches in twenty-two hours; at Genoa, 30 Ό0 inches in twenty-four hours; at Gibraltar, 3 3 Ό0 inches in twenty-six hours; on the hills above Bombay, 24Ό0 inches in one night; and on the Khasi Hills, India, 30 Ό0 inches on each of five successive days.

As regards the ocean, there are no available data from which an estimate could be formed as to the amount of the rainfall, since the rainfall statistics of the ocean must be regarded as giving hardly anything more than the comparative frequency of the fall. It is, however, certain that the equatorial belt of calms in the Atlantic and Pacific between the trades is the region where the ocean rainfall reaches the maximum, and the parts of these oceans are the rainiest which are the longest within the belt of calms as it shifts its position northward and southward with season. While the cloud-screen is undoubtedly dense, and the rainfall frequent and heavy, the careful observations of the “Challenger” and “Novara” show that the statements generally made as to these points are greatly exaggerated.

In the regions of the trades the rainfall is everywhere small over the open sea, seeing that the trade-winds are essentially an outflow from anticyclonic regions, and their original dryness is to a large extent maintained because their course is directed into regions which become constantly warmer. Thus at Ascension, lat. 8° 45' S., which is throughout the whole year within the S.E. trades, the mean rainfall for the two years 1854-55 was only 8 , 85 inches. At St Helena, which lies constantly within the same trades, five years give a mean rainfall of 5'36 inches on the coast; but in the same island at a height of 1763 feet the annual amount rises to 23'98 inches. Malden Island and some other islands in the Pacific, about long. 150° W., and for some distance on each side of the equator, have been pointed to by Scott as practically almost rainless, as is shown by their containing extensive guano deposits. These islands are situated somewhat similarly to Ascension with respect to the zone of calms. In Mauritius the annual rainfall on a mean of four years was 30 inches at Gros Cailloux, but at Cluny, only 16 miles distant, for the same four years it was 146 inches; in regard to which Meldrum remarks that at Cluny, which is in the vicinity of mountains and forests, in the southeast of the island, and thus directly exposed to the tradewind as it arrives from the sea, the rainfall in almost any month is from four to six times greater than at Gros Cailloux on the north-west coast, where neither mountain nor forest exists, and where the S.E. trade arrives considerably drained of its moisture.

From what has been said it is evident that the heaviest rains will be brought by the winds which have traversed the greatest extent of ocean within the tropics, and which accordingly of all ocean winds have the highest temperature and humidity. These conditions are most completely fulfilled during the summer months of the northern hemisphere by the winds which, commencing from near lat. 30° S., blow home on southern Asia as the well-known S.W. monsoon of these regions. Accordingly it is by the winds of this monsoon that a larger rainfall is distributed over a larger portion of the earth’s surface than occurs anywhere else in any season; and this large rainfall is in many regions still farther greatly increased by the mountain ranges which lie across the path of the rainbringing winds.

It is on these winds that the rainfall of India chiefly depends. Along the whole of the west coast from the Gulf of Cambay southward, and on the Western Ghats, the rainfall is excessive. The following are some of the more interesting annual means in inches beginning with Bombay and proceeding southwards:—Bombay, 74; Matheram, 247; Mahabaleshwar, 252; Ratnagiri, 104; Baura, 255; Goa, 102; Karwar, 115; Honawar, 139; Mangalore, 134; Cannanore, 132; Calicut, 116; and Cochin, 114. In the west of Ceylon the rainfall is also heavy, being at Colombo 87, at Galle 91, and at Ratnapura, at some distance inland among the hills, 149. Since the S.W. monsoon is drained of much of its moisture in crossing these mountains, a greatly diminished rainfall is distributed over the interior and east side of India, and on the eastern slopes of Ceylon.

If now we cross to the eastern shores of the Bay of Bengal, we again encounter an excessive rainfall along these coasts and up the slopes of the mountains looking down on them. Thus from south northward the following are among the more characteristic rainfalls in inches :— Nancowry, 102; Port Blair, 116; Mergui, 152; Tavoy, 196; Maulmain, 189; Rangoon, 100; Bassein, 98; Sandoway, 212; Akyab, 198; and Chittagong 104. On the other hand, at Thyetmio, inland on the Irawadi, the annual rainfall is only 48 inches.

We have shown how, in accordance with the peculiar distribution of pressure in India in summer, the monsoon is diverted up the valley of the Ganges as an E.S.E. wind, distributing on its way, even to the head of the valley, in a generous rainfall the moisture it has brought from the Indian Ocean and the Bay of Bengal. The rainfall does not extend farther westward than the basin of the Ganges, and the precipitation is most copious along the lower Himalayas, the largest falls being recorded at heights about 4000 feet,—being, as pointed out by Hill, near the level at which the summer monsoon is cooled just below its dew-point. The following are some of the larger rainfalls in inches, beginning with the more western:—Mussooree, 95; Naini Tai, 92; Khatmandu, 57; Darjiling, 121; Kurseong, 154; Buxa, 219; Kuch Behar, 131.

The rainfall is very large in the north-east angle of the Bay of Bengal and thence northwards towards Bhutan, or at the angle where the summer monsoon from the bay curves round to a westerly course on its way up the Ganges. Thus at Noakhally, on the coast, it amounts in inches to 109; at Tura, on the Brahmaputra, immediately to west .of the Garo Hills, 129; at Silchar and Sylhet to eastward, 117 and 155; whilst at Cherrapunji, on the Khasi Hills, it rises to 493 ’ 19 inches on a mean of twenty-four years. This last rainfall is the largest known on the globe, the causes of which are the highly saturated state of the monsoon on its arrival at the lower Ganges, the high mountain range of Burmah to eastward of Bengal, which turns the monsoon to the north, and the protrusion westwards of the Khasi and Garo Hills so as to lie in the line of that branch of the monsoon which passes from the lower Ganges into the basin of the Brahmaputra above Goalpara. The consequence is that the highly saturated air of the monsoon in its passage across the Khasi Hills is suddenly raised to a height of about 6000 feet, and being thereby reduced far below the point of saturation the superabundant moisture is precipitated in unequalled deluges of rain. The amount of the annual rainfall at all these places is determined, essentially if not altogether, by the rains of the summer monsoon, the relative intensity of which over India may be taken to be fairly represented by the rainfall of July.

The rains which accompany the N.E. monsoon of the winter months may be represented by the rainfall for January. These are heaviest in Ceylon, especially on its east slopes, and in southern India, or where the N.E. monsoon arrives after having traversed a large extent of ocean. The fall for the month exceeds 6 inches over a large portion of the east coast, whilst at Colombo in the west the rainfall is only half that amount, and farther north at Pattalum the January rainfall is only 1 ’ 82 inches. In southern India the amount varies from about 1 to 2 inches. Blanford pointed out in 1873 (Phil. Trans., vol. clxiv. p. 618) that, while the surface winds of northern India in winter are northerly, on the Himalayas, especially the northwest portion, southerly winds prevail during the cold months. It is these upper southerly winds which bring the winter rains to the Punjab, Upper India, and the highlands of Assam. It is further to be noted that winter rains also occur in Central India, where the prevailing surface winds are from east and north-east. The mean rainfall of January at Mussooree is 2Ό0 inches and at Naini Tai 2 - 86 inches, and in Assam, at Sibsagar, 1 - 13 inch. Over a large tract of the east side of southern India from Nellore southward, including Ceylon, the maximum rainfall for the year occurs in the months of October and November.

Rainfall of the Malay Archipelago and Australia.— Under the direction of the late Dr Bergsma, systematic observations of the rainfall of the Malay archipelago were begun in 1879, the number of stations being 150. The results of the first three years show that the mean annual rainfall over the archipelago varies from about 60 inches in Timor to upwards of 200 inches at some spots among the western slopes of Sumatra. But the most important feature in the rainfall in its relations to climate is not the absolute amount that falls annually, but rather the manner of its distribution through the months of the year. Over the greater number of the islands rain falls copiously every month; but as regards some of the islands the year is divided into dry and wet seasons as marked as are seen in the climates of India. The key to this essential difference among the climates is the distribution of atmospheric pressure during the months of the year from south-eastern Asia to Australia, with the resulting prevailing winds. During the winter months atmospheric pressure is high in south-eastern Asia and low in the interior of Australia, the difference being about three-quarters of an inch. Since between these two regions the fall in the mean pressure is practically uninterrupted, the Malay archipelago lying between them is swept by northerly winds (fig. 14). As these winds have traversed a great breadth of ocean in their course, they arrive in a highly saturated state, and consequently deposit a copious rainfall, particularly on the northern slopes of the higher islands. Hence in these months the rainfall over the islands without exception is large, the mean monthly amount being in many cases more than 30 inches. These winds continue their course to southward towards the low-pressure region in the interior of Australia, and deposit along the north coasts of that continent a monthly rainfall rising generally to from 14 to 20 inches. On advancing into the interior, the mean amount gradually diminishes at the successive telegraphic stations to 3'50 inches at Alice Springs near the tropic of Capricorn. The amount of the rainfall for any particular year, and the distance from the coast to which the rains penetrate inland, depend essentially on the height of the winter pressure of south-eastern Asia as compared with the low mean pressure of central Australia, by which the strength of the northerly monsoon is regulated.

On the other hand, during the summer of the northern hemisphere pressure is high in the interior of Australia and low in China, the mean difference being about half an inch. Between the two regions the fall in the mean pressure is continuous and uninterrupted, and as a consequence southerly winds prevail over the intervening archipelago. These winds, as they advance from the continent into lower latitudes, are absolutely rainless in the north of Australia, and over Timor and the other Malay islands which are separated from Australia only by a comparatively narrow belt of sea. During the three years no rain whatever fell in Timor in July and A ιgust, and the fall in June, September, and October was small. As, however, the winds pursue their course to northward, they eagerly lick up moisture from the sea, so that by the time they arrive at Amboyna they have become so saturated that the monthly rainfall there rises to nearly 30 inches. Again at some distance to the west of Timor rain falls more or less regularly every month, the amount increasing in proportion to the extent of ocean traversed by the S.E. winds, which advance towards these islands from the direction of Australia. These marked differences among the climates of the Malay archipelago, which, since they really depend on the geographical distribution of land and sea of this part of the globe, must be regarded as permanent differences, have played no inconspicuous part in the singular distribution of animal and vegetable life which characterizes the archipelago.

In July the prevailing wind in West Australia is N.W., and the rainfall reaches the maximum for the year, whereas in January the wind is S.E., and the rainfall is the minimum. Similarly in January since the winds of the southern half of South Australia and Victoria are from the south, and thus blow towards warmer regions, the rainfall is either at the annual minimum, or it is small. But on rounding the coast and proceeding northward, the wind becomes E., then N.E., and ultimately N. in the north of Queensland. With this prevalence of oceanic and equatorial winds, the rainfall at this time of the year rapidly rises over the whole of the eastern slopes, till at Cape York it is about 20 inches. In the basins of the Murray and Darling rivers, which are shut off from the east by the mountain ranges of New South Wales, the rainfall is only about an inch and a half. On the other hand, to south of the latitude of Sydney, including Tasmania, the maximum rainfall occurs in winter over those regions which slope south towards the sea. On crossing the mountain range of Victoria into the basin of the Murray river, the rainfall rapidly diminishes. In the north of New Zealand the winter rainfall is the heaviest; but farther south, where westerly winds prevail with some steadiness through the year, the rainfall is more equally distributed through the months; and, as the prevailing winds are westerly, the heaviest rainfall is in the west of the islands. Thus at Hokitika in the west near sea-level, and not far from a lofty range of mountains to the east, the annual amount reaches 120 inches, and at Bealey inland at a height of 2104 feet it is 106 inches. At Wellington the annual rainfall is 52 inches, at Southland 46, at Dunedin 34, and at Christchurch 25, thus showing, in the rainfall of the two sides of the island, extremes nearly as great as in Scotland.

Rainfall of Europe.— As regards rainfall, Europe may be conveniently divided into two distinct regions,—western and northern Europe, extending in a modified degree through the interior of the continent into Siberia, and the countries bordering on the Mediterranean. A vast ocean on the one hand, a great continent on the other, and a predominance of westerly winds are the determining circumstances in the distribution of the rainfall over western Europe. Hence the rainiest regions are to be found in the west, where mountain ranges stretch north and south. The annual rainfall exceeds 80 inches over a considerable district, including the greater part of Skye and portions of the counties of Inverness and Argyll to the south-east, in the lake district of England, and in the more mountainous parts of North Wales,—these three districts being the wettest in Europe. As Ireland presents no continuous range of mountains opposing the westerly winds of the Atlantic, no Irish rain-gauge shows a mean rainfall of 80 inches. A point of some interest is suggested by the rainfall of the counties of Kirkcudbright and Dumfries in Scotland. These counties offer to the westerly winds a series of valleys sloping south to the Solway Firth, which show successively a diminished rainfall on advancing eastward till at several places in Nithsdale and Annandale it does not exceed 40 inches. But in Eskdale, farther to the f ast, the rainfall instead of falling increases to about 60 inches. The reason is that the westerly winds are obstructed in their onward course by the range of hills by which Eskdale is bounded on the east, in surmounting which the winds are much reduced in temperature, and their superabundant moisture falls in copious rains immediately to westward of the ridge. The cause of the larger rainfall of Eskdale is thus analogous to that of the large rainfall of the coast in the north-east of the Bay of Bengal immediately under the Assam range of mountains. In England the largest annual rainfall is 146 inches at Seathwaite in the Lake district, in Scotland 128 inches at Glencroe in Argyll, whilst in Ireland the largest is only 76 inches. The driest part of the British Islands is an extensive district to south-south-west of the Wash, with a rainfall of about 21 inches. A large extent of England, and all the more important agricultural districts in Scotland, have a rainfall under 30 inches; the greater part of England, and nearly the half of Scotland, have a rainfall not exceeding 40 inches; but in Ireland it is isolated patches only that show a rainfall less than 40 inches.

In the west of Norway the rainfall in inches is 72 at Bergen, 51 at Aalesund, 46 at the Naze and in the Lofoten Isles, falling to 10 at the North Cape. At Christiania, Upsala, and a large part of the east of Scandinavia the rainfall is about 21 inches, falling to 16 inches on the north coast of the Gulf of Bothnia. In Russia and Siberia it rises only at a few places to 20 inches, several districts of this extensive region having an annual rainfall of 10, 5, 3, or even 2 inches. The rainfall of Spain presents great extremes—from 68 inches at Santiago to 13 inches at Saragossa. In France and the plains of Germany the average varies from 35 to 20 inches, but in mountainous regions these figures are greatly exceeded, rising through all gradations to upwards of 100 inches at some points in the Alps.

An important distinction between the manner of distribution of the rainfall in the west of Europe and at more inland places is that the greater part of the annual quantity of the west falls in winter, whilst in the interior the amount in summer is greater than in winter. The rainfall of January and July shows this in a very forcible manner. The summer climates of the extreme south of Europe and North Africa are rainless, and over extensive regions in the south of Europe adjoining the July rainfall does not amount to an inch. Over these dry regions the prevailing winds of summer are northerly, and hence the drought which characterizes them. On the other hand, the rainfall in the interior of the continent is large. In January the maximum rainfall occurs on the mountains and high grounds overlooking the Atlantic, and the minimum on the plains of Russia.

Owing to the way in which Europe is broken up by the seas which diversify its surface, the time of the year when the rain attains the maximum differs greatly in different regions. This phase of the rainfall occurs, indeed, according to locality, in all months except February, March, and April. The month of occurrence of the annual maximum rainfall over Europe is shown by fig. 18. A similar map representing the month of least rainfall shows still greater uniformity in a regular succession of the months in passing from region to region. Thus the month of least rainfall is January on the lower Volga, February in western Russia and the greater part of central Europe, March in the north of France and south of Great Britain, April farther to the north, May in Scotland north of the Grampians, June in Orkney, Shetland, Iceland, the west of Ireland, and the north-west of France, and July over the whole of the south of Europe. The driest month occurs nowhere in Europe in any of the five months from August to December.

Rainfall of North America.— West of the Rocky Mountains the rainfall is very unequally distributed, the annual amounts varying from 86 inches at Astoria, near the mouth of the Columbia river, to 8 inches at San Diego on the coast, and 3 inches at the head of the Gulf of California. Over the whole of the region between the Cascade and Rocky Mountains the rainfall at all seasons is extremely small, this being indeed that feature in the climate to which the formation of the cahons of that region is chiefly to be referred. On the other hand, in the United States and Canada to east of long. 100° W. the distinguishing feature of the rainfall is the comparative equableness of its distribution, an annual rainfall exceeding 50 inches occurring only over restricted districts, and a rainfall as low as 20 inches being scarcely met with anywhere. The regions where the rainfall exceeds 50 inches are Florida, the lower basin of the Mississippi, and the Atlantic seaboards of Nova Scotia and Newfoundland.

In January the annual maximum rainfall occurs over the whole of the west coast from Sitka to lower California; but in the interior between long. 120° and 95° W. the amount is everywhere small, and over a considerable part in the south-west of this region no rain falls. The region of largest rainfall extends from Louisiana to West Virginia, where the mean varies from 4 to 6 inches. Over nearly the whole of the Dominion of Canada, by much the greater i>art of the winter precipitation is in the form of snow, which has ∣een carefully measured and recorded by the Meteorological Service. The average snowfall for January exceeds 30 inches at St John’s, Newfoundland, in Anticosti, Prince Edward Island, and in many other regions.

In July the rainfall is everywhere small in the west, a large part of this extensive region being absolutely rainless. The remarkable dryness of the climate at this season is due to the N. W. winds that set in towards the low pressure of the interior, which thus blow towards warmer regions. The rainfall to the east of the Rocky Mountains is distributed by the winds which are connected with the low-pressure region of the interior and with the high-pressure region of the Atlantic. The result is two regions of larger rainfall, the one in the south-east of the States and the other to the west of the lakes. The summer winds of the south-eastern coasts are southerly, and as they are anticyclonic in their origin and have in their course traversed some extent of ocean, they arrive well-but not super-saturated, and pour down a rainfall in July of 6 inches and upwards along the coasts and for some distance inland from Louisiana to Chesapeake Bay. Further, since in July these winds attain their maximum force and persistency, the rainfall at the same time reaches the maximum along the whole coast from Boston to some distance west of New Orleans. Since the summer winds blow in the line of the Alleghany mountains and not across them, the rainfall diminishes in ascending their slopes. The comparative equableness of the rainfall over the eastern States is the necessary result of the winds’ passing into higher latitudes, and, therefore, cooler regions. A broad region where the rainfall is less than on each side of it, extends from Michigan to the south-west as far as Canadian River. To the west of the lakes the rainfall rises above 4 inches, and, since over this region the winds become somewhat easterly as they flow towards the low-pressure area, it is probable that the larger rainfall of this prairie region has its origin in no small degree in the evaporation of the lakes. On ascending the higher reaches of the Mississippi, the amount diminishes, but scarcely falls lower than 2 inches, being thus analogous to the summer rains of the Upper Ganges. On crossing the water-parting into the basin which drains into Hudson Bay, we encounter E. and N.E. winds laden with vapour licked up in their passage over Hudson’s Bay, which they distribute in a generous rainfall of probably 3 to 5 inches over the rising colonies of Manitoba ana Saskatchewan. An important point in the climate of the States is that over nearly the whole of the extensive region stretching between Alleghanies and Rocky Mountains, except the south coast already referred to, the annual maximum rainfall does not occur in summer but in spring, the month of largest rainfall in the great majority of cases being May. In the basin of Hudson’s Bay July is the month of largest rainfall.

Rainfall of Central and South America.— The following are, in inches, the larger and more interesting annual rainfalls round the coasts:—Vera Cruz, 182; Belize, 75; Maracaibo, 163; Caracas, 155; Georgetown, 95; Paramaribo, 142; Cayenne, 140; Para, 71; Pernambuco, 109; Buenos Ayres, 34; Bahia Blanca, 19; Puerto Montt, 102; Valdivia, 109; Valparaiso, 100; Serena, 93; Lima, 9; and a large part of Peru, nil. A remarkable feature of the rainfall of South America is the large amounts that fall in the basins of the Orinoco and Amazon; the fall is 91 inches in the upper basin of the Madeira, and 112 inches at Yquitos (lat. 3° 40' S., long. 72° 57' W.). The reason is that this immense region, where pressure appears to be almost constantly low, is open to the highly saturated winds that blow from the equatorial Atlantic. Quite different is the distribution of the rainfall over the La Plata basin. The annual falls, in inches, are 92 at Joinville, 58 at Corrientes, 44 at Monte Video, 36 at Parana, 24 at Santiago, 22 at San Luis, and only 6 at Mendoza. The fall rapidly rises in ascending the eastern slopes of the Brazil mountains facing the South Atlantic; thus, while the amount at Rio Janeiro is 45 inches, on the hills to northward it is 116 inches.

In January northerly winds prevail on the south coasts of the Gulf of Mexico and the Caribbean Sea, and as they have their origin in the high pressure of the American continent, and in crossing the sea pass into lower latitudes, the January rainfall of these coasts is comparatively small. In July, however, the prevailing winds are easterly, and as they have traversed a large extent of the equatorial waters of the Atlantic they are highly saturated, and consequently the July rainfall of these coasts is everywhere very large. The following are, in inches, the January and July rainfalls:—Caracas, 1Ό0 and 14’04; Guatemala, 0’28 and 1079; Vera Cruz, 5’10 and 35’90. The seasonal distribution of the rainfall in the basin of the Amazon is the reverse of this. In January the position of the belt of calms is about lat. 3° N., and as pressure is relatively low over the basin of the Amazon, especially its southern slopes, the trades and the west portion of the region of calms unitedly spread their highly saturated air over the whole region as far as the Andes, resulting in one of the most widespread heavy rainfalls anywhere to be met with. On the other hand, since in July the belt of calms is about lat. 10° N., the saturated atmosphere of the tropical regions no longer flows up the Amazon, but is carried westward into the Caribbean Sea and Gulf of Mexico. Hence at this season the rainfall of the Amazon valley is small. The following are, in inches, the January and July falls:—Para, 6’51 and 3'26; Manaos, 7’33 and 1'82; upper Madeira, 15’90 and 0’30; and Yquitos, 10’24 and 4 ’ 26. On the La Plata in January pressure is low, and as winds consequently blow from the ocean in upon the region of low pressure the rainfall is large; but as pressure is high in the interior in July the rainfall in that month is small. The following are, in inches, the January and July rainfalls:—Buenos Ayres, 2’37 and 170; Parana, 4’63 and 1’32; Corrientes, 5’24 and 2’67; Joinville, 14’26 and 3*55; and San Luis, 2 63 and 0Ό0.

Rainfall of Africa.— As regards the rainfall, Africa presents the greatest diversity in its climates. The following are the annual amounts in inches at various points on or near the coast :—Port Said, 2; Alexandria, 8; Tunis, 12; Algiers 31; Oran, 17; Mogador, 50; mouth of the Senegal, 17; Goree, 21; Sierra Leone, 126; Christiansborg, 23; St Thomas, 40; Gaboon, 106; Loanda, 11; Cape Town, 23; Mossel Bay, 12; Port Elizabeth, 24; Durban, 43; Zanzibar, 58; and mouth of the Zambezi, 61. In the north of the continent, the rainfall rapidly diminishes inland, and over the great desert of Sahara practically none falls. In the interior of Algiers it diminishes, the amount at Laghouat being 17 inches, and at Biskra 9. In Egypt the rainfall is limited to a narrow strip along the coast; at Cairo the annual fall scarcely amounts to an inch. The January and July rainfalls are, in inches, as follows :—Port Said, 0 ’ 46 and 0Ό0; Alexandria, 1'95 and 0’20; Algiers, 4’43 and 0Ό4; Biskra, 0’56 and 0Ό3; St Louis (Senegal), 0’28 and 3'00; Goree, 0'00 and 4'06; Sierra Leone, 0’69 and 24’20; Christiansborg, 0 - 50 and 2’00; Katunga, 0Ί1 and 476; Gaboon, 9’35 and 0’48; Cape Town, 0’28 and 3’83; Durban, 5Ό0 and 170; Pretoria, 6Ό7 and 071; and Zanzibar, 2Ό2 and 2’35. At Zanzibar the heaviest rains occur about the equinoxes, the mean for April being 14’55 inches, and for October 6 ’80 inches.

In the case of this, as the other continents, the explanation of the different amounts is to be had in the seasonal changes of wind. In the north the winter rains are to a very large extent the accompaniment of the Mediterranean storms of that season, but in summer pressure is diminished in the interior and increased in the Atlantic to the north-west, resulting in strong steady northerly winds, which as they advance into hotter regions are unaccompanied with rain. The heavy summer rains from Senegambia to the Gold Coast are due to the strong monsoonal winds which set in towards the interior, thus drawing over these coasts the highly saturated air of the belt of calms and of the trades immediately to the north and south of it. Since in winter the belt of calms is removed 8° of latitude farther to the south, and the temperature of the interior is greatly reduced, it follows that the winds blowing on these coasts from the sea are drier and less strong, and consequently the rainfall is small. At Sierre Leone.the absolutely driest month is February, 0 ’31 inch, and the wettest September, 29’15 inches. On the other hand, at Gaboon (lat. 0* 25'N.) the dry season is from June to August, when the belt of calms is farthest to the north; and the absolutely rainiest about the equinoxes, the mean of March being 1470 inches and October 19’52 inches. At Loanda (lat. 8* 49' S.) the annual amount is only a tenth of what falls at Gaboon, and it falls wholly during the summer months of the southern hemisphere. In South Africa pressure in January is lowest in the interior, towards which prevailing winds from the ocean blow, and as these advance into regions becoming rapidly hotter the rainfall all round the coast and for some distance inland falls to the annual minimum. But in more strictly inland districts which are at a considerable elevation the rainfall reaches the maximum at the same season. Thus the amounts in inches for January and July are—for Pretoria, 6Ό7 and 071; Maritzburg, 4 ’ 23 and 0’21; Graham’s Town, 2 ’ 89 and l ’ 51j Lower Nel’s Poort, 1'33 and 0’49; and Aliwal North, 1 '55 and 0 - 00. In the winter months pressure in the interior is high, and the rainfall consequently small. Though on the coast winds from the arid interior frequently prevail, yet the storms that sweep eastward past South Africa precipitate over large portions of the southern slopes of this part of the globe what must in the main be regarded as a generous rainfall. It follows that the climates of these important colonies range themselves into two perfectly distinct classes,—the climates of the inland regions and the Natal coast, where the rains occur during the hottest months, and the climates of the other regions, where the annual rains occur during the coldest months. Little is accurately known regarding the rainfall of the interior of Africa. It is certain, however, that it is small, or nil, over the extensive region of the Sahara, and that it is large from about 15° N. lat. to some distance south of the equator. Probably the rainiest part of Africa is the region extending from the Victoria Nyanza northwards to and including the gathering grounds of the two great tributaries of the Nile.

Snow.— Snow takes the place of rain when the temperature is sufficiently low to freeze the condensed moisture in the atmosphere. Snow is composed of crystals, either six-pointed stars or hexagonal plates, which exhibit the greatest variety of beautiful forms, one thousand different kinds having been observed. These numerous forms Scoresby reduced to five principal varieties :—(1) thin plates, comprising several hundred forms of the most exquisite beauty; (2) a nucleus or plane figure, studded with needle-shaped crystals; (3) six-sided, more rarely three-sided, crystals; (4) pyramids of six sides; (5) prismatic crystals, having at the ends and middle thin plates perpendicular to their length. In the same snowfall the forms of the crystals are generally similar. The flakes vary from 0Ό7 inch to an inch in diameter, the smallest occurring with low temperatures and the largest when the temperature approaches 32°. If the temperature is a little higher, the snow-flakes are partially thawed in falling through it, and fall as sleet. The white colour of snow is caused by the combination of the different prismatic colours of the minute snow-crystals. The density of snow is far from uniform; it is generally from ten to twelve times lighter than an equal bulk of water, but varies from eight to sixteen times lighter than water.

The limit of the fall of snow near sea-level coincides roughly with the winter isothermal of 52°, since in places where the mean winter temperature is no higher than 52° that of the air falls occasionally to 32° or lower during the winter months. As regards Europe, the southern limit is about Gibraltar; in North America it is Savannah, New Orleans, the mouth of the Rio Grande, the head of the Gulf of California, and San Francisco. In Europe, north of lat. 60°, snow falls generally on an average of from 80 to 110 days in the year. At Upsala the number of days is 61, at Warsaw 45, Aberdeen 42, Oxford 18, Ostend 15, Brussels 27, Tarum (in the south-west of Jutland) 12, Copenhagen 23, Vienna 33, Odessa 19, Sebastopol 12, Milan 11, Trieste 6, Saragossa 5, Madrid 3, and Lisbon 1. In Greenland the number of days exceeds 80, and this figure is nearly reached in Newfoundland and the northeast seaboard of Nova Scotia. At Quebec the mean days of snow are 66, Halifax 64, Winnipeg 54, Detroit 34, Cape Henry 13, St Louis 11, mouth of the Columbia River 7, and Charleston 2. In Russia the time of the year when snow falls most frequently is December and January, except in the south of the empire, where February is the month of the most frequent occurrence of snow. But to the north of a line drawn from the entrance of the

Gulf of Finland through Warsaw, Cracow, Salzburg, and Santiago March is the month of maximum occurrence in the great majority of instances; while to the south of this line it is January and in several cases December.

The largest falls of snow occur in the Antarctic regions, as is well attested by the magnificent icebergs of solidified snow which break off all round from the lofty walls of ice that engirdle the Southern Ocean. Excepting perhaps in the Dominion of Canada, no data have been anywhere collected from which even a rough estimate could be formed as to the mean annual amount of snow that falls in different parts of the globe.

Snow-Line.— The snow-line marks the height below which all the snow that falls annually melts during summer. No general rule can be stated for this height in different climates owing to the many causes determining it. These are the exposure of mountain slope to the sun (and hence, other things being the same, it is higher on the south than on the north sides of mountains), exposure to the rainbringing winds, the steepness of the mountains, and the degree of dryness of the air. Hence the position of the snow-line can be known by observation only. It falls only little on either side of the equator to lat. 20°; from lat. 20° to 70° it falls equably, but from lat. 70° to 78° much more rapidly. To this general rule there are many exceptions. It is 4000 feet higher on the north than the south side of the Himalayas, owing to the larger snowfall on the south, and the greater dryness of the climate of the north side, and therefore the greater evaporation from the snow there. It is higher in the interior of continents than near the coasts, because the precipitation is less and summer heat greater. In the Caucasus it is 11,063 feet high, but only 8950 in the Pyrenees. In South America it rises from the equator to lat. 18°, and more on the west than on the east slopes of the Cordilleras, owing to the large precipitation on the east and small precipitation and arid climate of the west side of that chain of mountains. It is as high in lat. 33° S. as in 19° N., but south of that latitude it rapidly sinks owing to the heavy rains brought by the moist N.W. winds of these regions. In the south of Chili it is 3000 feet lower than in the same latitudes in Europe, and 6000 feet lower than in the extremely arid climates of the Rocky Mountains.

Storms.— If weather charts representing a large part of the northern hemisphere be examined, two distinct systems of pressure are seen which change their forms and positions on the earth’s surface from day to day. The one set are systems of low pressure marked off by concentric isobars enclosing pressures successively lower till the centre is approached; and the other systems of high pressure marked off by concentric isobars enclosing pressures becoming successively higher towards the centre. The former of these are called cyclones, and the latter anticyclones. These areas of low pressure are the distinguishing characteristics of the hurricanes and typhoons of tropical regions, and of the ordinary storms of higher latitudes, and they may all be conveniently grouped under the general name of cyclones. Fig. 19 shows a storm which was passing across northwestern Europe on the morning of November 2, 1863, and it may be taken as fairly representing the general features of cyclones. In the figure the arrows fly with the wind, and the force of the wind is indicated by the number of feathers on the arrows.

It will be seen that the winds indicate, not a circular movement round the centre of lowest pressure, but a vorticose motion inwards upon that centre, the motion being opposite to that of watch-hands. In other words, the wind follows Buys Ballot’s law, already explained. The winds are strongest where the isobars are closest together; or they are generally proportioned to the “barometric gradient,”—a term introduced by Stevenson in 1867. Cyclones have diameters seldom less than 600, and they occasionally exceed 3000 miles; the cyclone of fig. 19 had a diameter of about 1200 miles. The cyclones of the Mediterranean are usually of smaller dimensions than those of northwestern Europe and America. The rates at which cyclones advance over the earth’s surface vary greatly, the average in America being 24 miles an hour, in the Atlantic 20 miles, and in Europe 26 miles. A rate as high as 70 miles an hour has occurred in the British Islands; sometimes they remain stationary, and more rarely their course is for a time retrograde. The temperature and humidity increase at those places towards and over which the front part of the storm is advancing, and fall at those places over which the front part of the storm has already passed. In other words, the temperature and humidity rise as pressure falls and fall as pressure rises. This is the important climatic significance of cyclones. Thus a succession of low pressures passing eastwards in courses lying to northward of the British Islands are the essential conditions of open winters; whereas, if the cyclones follow courses lying to southward, the winters are severe. In a cyclone the broadest feature of weather is an area of rain about or rather somewhat in front of the centre, surrounded by a ring of cloud, outside which the sky is clear. The precise form and position of these areas have been shown by Abercrombie to vary with the type of pressure distribution, with the intensity of the cyclone, and with the rate of its progress, and they are also influenced by local, diurnal, and seasonal variations.

The chief point of difference between American and European storms is essentially the result of the mean winter pressures to the west and north-west of their respective storm-tracks. Owing to the high winter pressure in the interior of America, the barometer rises in the wake of the storms of the United States more rapidly, the wind veers round more quickly and more uniformly to N.W., N.N.W., and N. and keeps longer in these directions, and the temperature and humidity fall to a greater degree, than happens in Europe. In the New England States and Canada the easterly winds of the storms, coming as they do from the Atlantic, are disagreeably cold, damp, and misty in a degree and with a frequency much greater than occurs with the same winds in the British Islands.

The chief points of difference between the hurricanes and typhoons of the tropics and the cyclones of higher latitudes are these :—tropical cyclones are of smaller dimensions, show steeper barometric gradients and therefore stronger winds, and advance at a slower rate over the earth’s surface. Another point of difference is that a large number of the hurricanes of the West Indies and the typhoons of eastern Asia first pursue a westerly course, which gradually becomes north-westerly, and on arriving at about lat. 30° they recurve and thereafter pursue a course to north-eastwards. The tropical cyclones of the Indian Ocean south of the equator also first pursue a westerly course, which gradually changes to south-west, and often on arriving about lat. 30° recurve to the south-east. Many of the cyclones of India have their origin to westwards of the Nicobar Islands, pursue a course to north-westward, and die out in the valley of the Ganges; and, similarly, a considerable number of the cyclones of the West Indies pursue a westerly course through the Gulf of Mexico, and several die out in the States.

The most dreadful attendant on tropical cyclones is the storm-wave, caused by the in-blowing winds and the low pressure of the centre of the storm. When this wave is unusually high and is hurled forward on a low-lying coast at high water it becomes one of the most destructive agents known. The Bakarganj cyclone of October 31, 1876, was accompanied by a wave which flooded the low grounds to the east of the delta of the Ganges to heights varying from 10 to 45 feet, by which more than 100,000 human beings perished.

Tracks of Cyclones of North America, Atlantic, and Europe.— In the Physical Atlas of the Atlantic Ocean, issued under the direction of Dr Neumayer of the Deutsche Seewarte, plate 28 shows by shadings the mean positions of the centres of cyclones and by lines their mean tracks. The following are the regions where the lowest barometer of storms has been most frequently found:—the region to west-south-west of the lakes of the United States; the Gulf of St Lawrence; mid-Atlantic about lat. 35° long. 52°; to the south-west of Greenland; to the south-west of Iceland, which is by far the most important of the whole; to the south-west of the Lofoten Isles; the region embracing Denmark, the south of Scandinavia, and Finland; and, as secondary centres of frequency, the south of the British Islands, Corsica and part of Italy adjoining, and the northeast of the Adriatic. The great importance of these centres, where the lowest barometers are most frequently found, consists in the indication they give of the precise regions either where many storms originate or where they are either retarded or arrested in their course. As regards the origin of storms, the centre west of the Mississippi is the region where most of the United States storms originate, the centre in the Gulf of St Lawrence is where many of the great Atlantic storms have their origin, and the centres in mid-Atlantic and to the south-west of Iceland are the regions where the storms of north-western Europe chiefly originate. The centres on the south-west of Greenland, the Lofoten Isles, Denmark, and the south of the British Islands, all appear to suggest that storms are retarded in their onward courses on coming up against large masses of land,—which may, in part at least, be occasioned by the heavy rainfalls that mark these parts of their courses.

Of all storm tracks the most frequently taken is that by the storms of the United States, which pursue an easterly course through the lakes to the Gulf of St Lawrence. A considerable number of storms follow a course from Nova Scotia to Davis Straits; but the larger number take a north-easterly course through the Atlantic towards Iceland and thence past the north of Norway. Among the less frequent but important tracks are these :—from near New Orleans along the east coast of the States towards Nova Scotia; from mid-Atlantic to south of Ireland and thence through France to the north of the Mediterranean; and from the Atlantic about lat. 42° long. 40° in a northeasterly course quite outside but at no great distance from the British Islands, and thence towards the North Cape. Of the tracks more immediately affecting British weather are one from Iceland in a south-easterly direction through the North Sea and Germany, and four tracks which start from near Scilly:—(1) to the south-east as already described; (2) eastward through the north of Germany; (3) north-east to Christiania; and (4) north through Ireland and the Hebrides. These are the storm tracks which chiefly give the United Kingdom its easterly and northerly winds.

The Inclination of Winds to the Isobars.— The vorticose motion of the winds in a cyclone towards and in upon the centre has been already pointed out. One of the more important practical problems of meteorology is the determination of the angle of inclination of the winds to the isobars in the different segments of the cyclone, not only from the application of the results of the inquiry to the theory of storms but also to practical navigation. The first real contribution to the subject, based on accurate measurements, was made by Clement Ley in 1873. From the observations made at fifteen places in north-west Europe examined by him he showed that the winds incline from districts of higher towards those of lower pressure at a mean angle of 20° 5Γ; that the inclination is much greater at inland than at well-exposed stations on the coast, the respective angles being 28° 53' and 12° 49'; and that the greatest inclinations are with S.E. winds. Then follow S.W., N.E., and N.W. winds, the last showing the least inclination. Whipple has recently compared the winds at Kew with the barometric gradients for the five years ending 1879, with the result that the greatest inclination is 63° with S.E. winds, the least 35° with N.E. winds, and the mean for all winds 52’.

As regards the open sea, Captain Toynbee has shown, from a careful investigation of the great Atlantic storm of August 24, 1873, that the mean angle of inclination calculated from one hundred and eight observations was 29°, the mean at the three selected epochs examined varying from 25° to 31’.

Barometric Gradient and Velocity of the Wind.— In inquiring into the relation of the velocity of the wind to the barometric gradient, it is necessary to have some definite information as to the increase of the velocity with height above the ground. Stevenson recently made observations on this point on winds varying from 2 to 44 miles an hour from the surface up to a height of 50 feet, from which he has drawn the following conclusions :—(1) the spaces passed over in the same time by the wind increase with height above the ground; (2) the curves traced out by these variations of velocity from 15 to 50 feet high coincide most nearly with parabolas (fig. 20)

having their vertices in a horizontal line 72 feet below the surface; (3) between 15 feet and the ground there is great disturbance of the currents, so that the symmetry of the curves is destroyed; (4) the parameters of these parabolas increase directly in the ratio of the squares of the velocities of the different gales. If x be the velocity of the wind at height H above the ground, the parameter of the corresponding parabola is aj 2 ∕(H + 72); and as x varies the parameter will vary as x 2 ∙ or as the square of the velocity of the gale. It follows that, to render wind observations comparable, it is necessary that anemometers be placed at one uniform height above the ground, and that standard height not lower than 15 feet above the. surface. It is very desirable that the inquiry were prosecuted up to a height of 100 feet; and it is of the utmost importance that the variation in the diurnal velocity be at the same time determined at different heights from 15 feet upwards.

Stevenson also made wind observations on the Calton Hill, Arthur’s Seat, and the Pentland Hills, in the vicinity of Edinburgh, up to a height of 1600 feet above sea-level. It is from observations made at stations on knolls and peaks at different heights above the sea, and at different heights above the surfaces of their summits, that the problem of the variation of the wind’s velocity at different heights with the same barometric gradient can be ascertained. In carrying the inquiry to considerable heights, the results cease to be comparable with those obtained at lower levels, unless in those cases where neighbouring heights are available for data from which the barometric gradient at the observed height can be calculated. The results of observations as to the velocity of atmospheric currents at very great elevations in the atmosphere deduced from the apparent movements of the higher clouds arc altogether incomparable with the winds near the surface of the earth, for these among other reasons :—the heights of the clouds can be at best but imperfectly ascertained; the motion of the clouds, particularly the higher clouds, may be only apparent, it being sometimes difficult to distinguish between the formation and dissolution of clouds and their motion; and above all, since the higher clouds are usually the accompaniments of the greater weather changes, their movements are the result of barometric gradients towards a knowledge of which we are absolutely powerless to take a single step.

As regards surface winds, Clement Ley in 1881, and Whipple more recently and with greater fulness, have calculated the mean wind velocities for twelve gradients,— the gradients being derived from the daily weather charts of the Meteorological Office for the five years 1875 to 1879 at 8 A.Μ., and the corresponding wind data being obtained from the hourly readings of the Kew anemograph. The barometric gradient is for 15 nautical miles, and the following are the velocities for the twelve gradients on the mean of the year:— [table]

The influence of season is very strongly marked. The velocities for the same gradients in order are—October to December, 12’5 miles; July to September, 12’6 miles; January to March, 14’8 miles; and April to June, 17’2 miles. From those observations of Whipple it follows that during the six months when the temperature is falling the velocity for the same gradients is least, while the velocity is greatest during the six months when the temperature is rising, and absolutely greatest during the three months ending June, when the greater part of the annual increase of temperature occurs. It is evident that the observed increase in the velocity of the wind for the same gradients is to be referred to the same cause that brings about the diurnal increase in the wind’s velocity, viz., the wind blowing over a warmer surface than itself.

Whipple has also sorted the winds according to the eight points of the compass, with results of the greatest interest. If N.W., N., N.E., and E. winds be grouped together as polar, and S.E., S., S.W., and W. winds as equatorial winds, the mean hourly velocity of the polar winds, for the same gradients, is 1*1 miles in excess of the equatorial winds. Now, since polar winds pass into lower latitudes, the surface of the earth over which they blow is warmer, whereas the surface is colder than the equatorial winds which blow over it. It follows that the increased velocity of polar winds is referable to the same conditions which result in the diurnal increase in the wind’s velocity and the greater velocity for the same gradients of winds when the annual temperature is rising, since in all these cases the winds blow over a surface of a higher temperature than their own.

It is evident from these considerations that for the development of the law of the relation of the wind’s velocity to the barometric gradient with an exactness sufficient to warrant us in expressing that relation in a general mathematical formula much yet remains to be done. In truth, as regards the various formulae submitted by Ferrel, Mohn, Hann, Everett, and others, we have no choice but to allow the justness of Strachan’s criticism (Modern Meteorology, p. 98) that the theoretical values furnished by the formulae do not accord with the actual values, and that therefore a satisfactory formula is yet to be found. Ere such a formula need be looked for, the conditions must be fulfilled for the preliminary work of supplying the observational data required. The “Challenger” observations prove that, with gradients substantially the same, the velocity of the wind is greater on the open sea than near land; and we have seen that the velocity varies with the hour of the day, and generally is increased as the temperature of the surface rises above that of the air blowing over it, and diminished as the temperature of the surface falls below that of the air. It is evident that observations on the open sea will afford data for the simplest solution of the problem; but on land the diurnal, seasonal, and nonperiodic changes of temperature greatly complicate the problem, and render necessary for its solution observations specially designed for the purpose. It is not easy to see how these can be obtained but by carrying out the plan proposed in 1875 by Stevenson of establishing strings of well-equipped meteorological stations planted sufficiently close that the barometric gradients may be determined within the limits of accuracy required. Observations made twelve times daily for a year, at stations so arranged, would supply the observational data for the solution of this fundamental problem in meteorology. Till some such proposal be carried out, the problem remains unsolved, for barometric gradients based on the widely separated existing stations are too uncertain and rough and the wind observations are wanting in that comparability which alone can satisfy the inquiry.

Weather and Weather Maps.— Weather is the state of the air at any time as respects heat, moisture, wind, rain, cloud, and electricity; and a change of weather implies a change in one or more of these conditions. Of these changes the most important as regards human interests are those which refer to temperature, wind, and rain; and, as these are intimately bound up with the distribution of atmospheric pressure, the latter truly furnishes the key to weather changes.

These relations are well shown by the International Monthly Weather Maps issued by the United States Signal Service. Of these that for December 1878 is a striking example. This month was characterized over the globe by unusually abnormal weather. A line drawn from Texas to Newfoundland, across the Atlantic, the north of France, and Germany, thence round to south-east, through the Black Sea, the Caucasus, India, the East India Islands, and Australia to the South Island of New Zealand, passes through a broad and extended region where pressure was throughout considerably below the mean of December, and this low pressure was still further deepened in various regions along the line. Another line passing from Australia, through the Philippine Islands, Japan, Manchuria, Behring’s Strait, and Alaska, also marks out an extensive region where pressure was uninterruptedly below the mean.

On the other hand, pressure was above the average, and generally largely so, over the United States to west of longitude 90°, over Greenland, Iceland, the Faroes, Shetland, and a large portion of the Old Continent bounded by a line drawn from Lapland round by Lake Balkhash, Canton, Peking, to the upper reaches of the Lena. Another area of high pressure extended from Syria, through Egypt and East Africa, to the Cape; and part of a third area of high pressure appeared in the North Island of New Zealand. As regards North America, the greatest excess of pressure, 0Ί96 inch above the mean, occurred in the Columbia Valley, from which it gradually fell on proceeding eastward to a defect from the average of 0'146 inch near Lake Champlain and to northward, rising again to near the mean on the north of Nova Scotia. To the north and north-east exceedingly high pressures for these regions and the season prevailed, being 0'635 inch above the mean in Iceland, 0'500 in the south of Greenland, and at the three stations in West Greenland, proceeding northward, 0'445, 0'402, and 0'346 inch.

West Greenland being thus on the west side of the region of high pressure which occupied the northern part of the Atlantic, and on the north-east side of the area of low pressure in the States and Canada, strong south winds set in over that coast, and the temperature at the four Greenland stations, proceeding from south to north, rose to l β ∙l, 8° '8, 12 o, l, and 14° '4 above the means. As the centre of lowest pressure was in the valley of the St Lawrence about Montreal, strong northerly and westerly winds predominated to westward and southward, where consequently temperature was below the average, the deficiency at Chicago and St Louis being 9°'5; and, winds being easterly and northerly in California, the temperature there was also under the mean. On the other hand, in the New England States, the greater part of the Dominion of Canada, and West Greenland temperature was above the average. Pressure was much higher at St Michael’s, Alaska, than to south-westward at St Paul’s, Behring’s Strait, and hence, while temperature at St Paul’s was 2°'9 below the normal, it was 12° '0 above it at St Michael’s, where strongly southerly winds ruled. With these strong contrasts of pressure, America presented contrasts at least as striking in the distribution of the temperature. Along the south of Lake Michigan the November temperature was 13°'7 above the normal, whilst the December temperature was 9°'5 below it, the difference there between the two consecutive months being thus 23° '2∙

As regards Europe, Iceland was on the east side of the patch of high pressure which overspread the north of the Atlantic, and hence northerly winds prevailed there and temperature fell 7° '2 below the mean, presenting thus a marked contrast to the high temperature of West Greenland at the time. In Europe, the area of lowest pressure occupied the southern shores of the North Sea, extending thence, though in a less pronounced form, to south-eastward. Hence over the whole of western Europe winds were N.E., N., and in the south-west of Europe W.; and hence everywhere from the North Cape to the north of Italy temperature was below the normal, in some places greatly so, the deficiency being 10° '4 in the south of Norway and 12°'2 in the south of Scotland. On the other hand, on the east side of this area of low pressure winds were southerly and temperature consequently high. In some localities in Russia the excess above the mean was 15° '0, and over a large proportion of European Russia the excess was not less than 9° '0. This region of high temperature extended eastward into Siberia as far as the Irtish, being coterminous with the western half of the anticyclonic region of high pressure which overspread central Siberia. But over the eastern portion of the anticyclone northerly winds prevailed, with the inevitable accompaniment of low temperatures over the whole of Eastern Asia, the deficiency at Nertchinsk on the upper Amur being 6 o, 8. Here again, just as in America, Greenland, and Iceland, places with atmospheric pressure equally high presented the strongest contrasts of temperature. Thus at Bogoslovsk, on the Ural Mountains, pressure was 0 '211 inch and at Nertchinsk 0'154 inch above the normals, but Bogoslovsk on the west side of the high pressure area had a temperature 15° '0 above, whilst at Nertchinsk it was 6° '8 below the average.

At this time of the year the mean pressure falls to the minimum in Australia, but during December 1878 the usually low pressure was still further diminished. Pressure at this season also falls to the annual minimum in the North Pacific and North Atlantic, and it has been seen that the low pressure of these regions was likewise still further diminished. But in the case of the Atlantic it was attended with a most important difference. The centre of lowest pressure, usually located to the south-west of Iceland, was removed some hundreds of miles to the south-east, and an unwonted development of extraordinarily high pressure appeared to the northward, overspreading the extensive region of Baffin’s Bay, Greenland, Iceland, Faroes, and Shetland. It was to this region of high pressure, particularly in its relations to the low-pressure region to the south-east of it, that the extreme severity of the weather in the British Islands at the time was due. Now this high-pressure region was intimately connected with, and doubtless occasioned directly by, upper atmospheric currents from the widely extended region of low pressure to southward, with its large centres of still lower pressure in the North Sea, mid-Atlantic, and United States, where pressures were respectively 0’307, 0 ’322, and 0’146 inch under the normals. Thus, with the single exception of the high-pressure area about Greenland, the meteorological peculiarities which render December 1878 so memorable over nearly the whole globe arose out of a distribution of the earth’s atmosphere essentially the same that obtains at that time of the year, but the usual irregularities in the distribution of the pressure appeared in more pronounced characters.

Taking the all-important bearings of these areas of high and low pressure on weather and climate into consideration, along with the abnormal concentration of aqueous vapour over extensive regions which they imply, it is evident that, when the meteorologist will be in a position to forecast, on scientific grounds, the weather of the coming season for the British Islands, it is to the Atlantic he will require to look for the data on which the forecast is based.

These questions, which the International Weather Maps of the United States enable us to discuss, are of the first importance in meteorology, whether we consider the amplitude of the atmospheric changes they disclose (these being often so vast as to embrace four continents at one time, besides being profoundly interesting from their direct bearings on the food supplies and commercial intercourse of nations) or regard the larger problems they present, with hints towards their solution, which underlie physical geography, climatology, and other branches of atmospheric physics. The discussion presents the great atmospheric changes as influenced by oceans and continents, including the subordinate but important parts played by mountain ranges, extensive plateaus, and physically well-defined river basins in determining the development, course, and termination of these changes.

Weather Forecasts and Storm Warnings.— It is in tropical and subtropical countries that an isolated observer may, with a close approximation to certainty, predict the approach of gales and hurricanes. In these regions atmospheric pressure and the other meteorological conditions are so constant from day to day that any deviation, even a slight one, from the average of the hour and season in respect of pressure, the direction and strength of the wind, and the direction and amount of cloud, implies the presence of a storm at no great distance. Dr Meldrum has practically worked out this problem at Mauritius with great success. At the Royal Alfred Observatory there the mean pressure at sea-level in January at 9 a.m. is 29 ’ 966 inches, from which it falls to 29’904 inches at 4 p.m., then rises to 29’980 inches at 10 p.m., and again falls to 29’927 at 4 a. Μ. The mean direction of the wind and the diurnal variation, both as regards direction and force, have been stated (p. 125). Suppose then that the barometer is observed to fall after 9 a.μ. more rapidly than is due to the usual daily barometric tide, that in the afternoon it does not indicate the second maximum or that it continues to fall instead of rising,—or suppose, in short, any deviation from the mean daily march,—then it is certain that there is somewhere an atmospherical disturbance near enough to Mauritius to influence the pressure. The direction in which the disturbance is from Mauritius is readily known from the wind, and the distance of the storm closely approximated to by noting the rate and amount of the fall of the barometer, in connexion with the changes of the wind and the clouds,—the rate and progressive motion of the storm being known chiefly from the veerings of the wind. For a good many years past notifications have been sent to the daily newspapers when observations show that a storm is not far from the island, stating its position and probable course from day to day. The scheme of storm warnings at Mauritius has been entirely successful, and the result is of great value, since it shows what may be done at an isolated station in the ocean, or what may be done in ships at sea. In this connexion it is not possible to overestimate the importance to seamen of a knowledge of the hourly variations of the barometer and its mean monthly heights over the ocean tracks of commerce.

In passing from Mauritius to the British Islands we pass from a region where the forecasting of storms and weather is simplest and easiest to the region where it is most complex and difficult, particularly for the western districts of these islands. The great difficulty lies in the fact that the British Islands are immediately bounded by the Atlantic to westwards; and, since practically every storm and nearly all weather changes come from that direction, no telegraphic communication of their approach can be received. The Meteorological Office in London has therefore no choice but to base the forecasts on such of the observations telegraphed to the office as experience has shown to be the precursors of storms and other weather changes. The more important of these observations are the falling and rising of the barometer taken in connexion with changes in the direction and force of the wind. Since on the north side of the track of the centre of the storm winds are northerly and easterly and temperature low, and on the south side winds are southerly and westerly and temperature high, one of the most important points to be ascertained is the probable path the centre of the coming storm will take. Though a good deal remains to be accomplished in the development of this phase of storms, yet much has recently been done in this direction by close examination of the changes of pressure in the region of the anticyclone contiguous to the advancing storm and by the changing positions of the rain area near the centre of the cyclone.

As regards Europe, the facility of forecasting storms increases as distance from the west coasts is increased. Thus to the middle and eastern districts of the British Islands, were a day and night watch established in the west, forecasts of almost every storm could be issued, the exceptions being those small cyclones or satellite cyclones, as they are called, originating within the British Islands themselves, which are frequently characterized at once by their severity and by the rapidity of their onward course. In the United States, the system of weather forecasting is perhaps the best in temperate regions,—a result due to the admirable system organized and developed under the direction of the late General Myer, and adequately subsidized by the Government, but above all to the facilities to detect and track the storms in the region where nearly all of them have their origin, to west of the Mississippi, before they advance upon the more thickly peopled States.

Meteorology sustained a heavy loss by the death in 1877 of Leverrier, who was not only the keenest-sighted of physicists but also the prince of organizers of systems of meteorological observation. His last great service to the science was the establishment of a system of observation, by which the propagation of rain, hail, and other weather phenomena could be followed and recorded from commune to commune over France. This scheme for the investigation of the vitally important bearing on the meteorology of a country of a comprehensive observation of its rainfall, hail, and thunderstorms, through numerous observers possessing sound local information, is not only eminently just in science, but is calculated to be attended with the greatest benefits to agricultural and other public interests. The practical advantages of the scheme, it need scarcely be added, can only be reaped after a very large expenditure of labour and money in organizing a comprehensive parochial scheme of observation, systematically and persistently carried through and discussed.

Further details regarding meteorological phenomena will be found in the articles Atmosphere, Barometer, Climate, Hygrometry Ozone, Raingauge, Sea, and Thermometer. (a. b.)

TERRESTRIAL MAGNETISM.

1. In the preceding portion of this article some account has been given of the influence which the sun and moon exert upon the air, the earth, and the ocean, their strictly tidal effects being left to be separately dealt with. The discussion of the influence of these bodies on what may be termed the movables of the earth will not be complete, however, without embracing an account of the changes which they produce in the earth’s magnetism. An account of the earlier magnetic observations has already been given under the heading Magnetism, and our task will now be to give in the first place a description of the best and most recent instruments by which the magnetic state of the earth is determined, embracing therein observatory instruments, those adapted for travellers whether by land or by sea, and differential magnetometers. We shall next give a short account of the magnetic system of the earth and of its secular variation; and we shall then investigate the changes connected with terrestrial magnetism depending on the sun and moon. In performing this task we shall be led to conclude that the sun’s power is variable, and we shall therefore examine whether this conclusion is likewise borne out by strictly meteorological observations. Finally, we shall venture on remarks embodying a provisional working hypothesis, and our object will be gained if this should be found to suggest certain lines of thought to those interested in the subject which may lead them to examine and discuss the very great mass of observations at present existing.

Instruments for Determining the Magnetic State of the Earth.

(a) Observatory Instruments. 2. Declinometer.— It is that end of the needle which points to the north magnetic pole of the earth of which the position is invariably noted even when the observation is made in the southern hemisphere. The difference of this position from true geographical north denotes what is called the variation or declination (east or west) of the needle. East is often reckoned negative and west positive. The instrument by which this information is obtained is called the declinometer. The unifilar magnetometer, which is the form of declinometer now used, is described and figured in Magnetism, vol. xv. p. 238.

3. Dip Circle.— The instrument by which the magnetic dip or inclination is observed contains a thin needle about 3 inches long, the centre of gravity of which coincides as accurately as possible with the axis of motion of the needle. The needle has two axles consisting of two very fine cylinders of hard steel standing at right angles to the plane of the needle, and great attention must be paid to keep these axles in a state of perfect polish and dryness. By means of these the needle can oscillate freely on two horizontal agate rounded edges, the one axle lying on the one edge and the other axle on the other. If the centre of gravity coincides exactly with the axis of motion, and if there be no adhesion or friction between the axles and the agate edges, the needle must settle into such a position that its magnetic axis lies in the true line of dip.

The position of the ends of the needle is read by means of two microscopes which move round on a cross piece carrying verniers. To view the position of the lower end of the needle we move round the lower microscope until the cross wire in its field of view (extending in the line between the two microscopes) symmetrically cuts the extremity of the needle. The lower vernier is then read. The same process is repeated for the upper vernier, and the mean of the two readings is taken. This mean will accurately denote the position of the needle if the circle is properly set.

The sources of error in a dip observation are—(1) a want of symmetry in mass, the centre of gravity of the needle not being coincident with the axis of motion; (2) the vertical circle being erroneously set; (3) a want of symmetry in magnetism, the magnetic axis not being coincident with the axis of figure; (4) excentricity, the axis of rotation of the needle not passing through the centre of the circle; (5) friction and adhesion of the axles as they rest on their agate supports. This last source of error is guarded against by taking great care of the axles, and by inserting them gently into a piece of cork before each observation; the agate supports ought also to be rubbed with cork. Then, again, when the needle has assumed its position, before reading it is gently raised by means of a lifter, the handle for turning which is shown in the figure towards the right. It is then gently lowered, and this process is repeated until no apparent change of position is produced by the operation.

4. We shall now describe a complete dip observation. The first point is to make the needle to swing in the plane of the magnetic meridian. In order to accomplish this, after levelling the instrument, the verniers are set for 90°, that is, for a vertical position of the needle. The whole instrument is now turned round its horizontal circle until the extremities of the needle are bisected by the wires of the two microscopes, and the position of the vernier of the horizontal circle is then read. The needle is next reversed so that the microscope shall view its other flat side; it is made vertical as before, and the position of the horizontal circle read once more. Next the face of the instrument is turned round 180°, and the same two operations repeated. We have thus four readings of the horizontal circle, and if we take the mean of these we shall have ascertained with sufficient accuracy the position of that plane for which the needle is vertical. Now this plane must be removed 90° from the magnetic meridian, for in such a plane the horizontal magnetic force of the earth would have no resolved portion acting in the plane of the needle’s motion, so that the needle would practically be under the sole influence of the vertical magnetic force, and would therefore point in a vertical direction. By this means therefore we obtain the magnetic meridian, and thus know the plane in which we ought to swing the needle. The needle must now be read in the following positions (a) face of instrument east—face of needle to face of instrument; (j8) face of instrument west—face of needle to face of instrument; (γ) face of instrument west—back of needle to face of instrument; (δ) face of instrument east—back of needle to face of instrument. Finally, the poles of the needle must be reversed, by rubbing them with powerful bar magnets in a direction opposite to that in which they were previously rubbed, and four observations taken corresponding to the above. The mean of the eight observations so obtained will give us the true dip.

The turning round of the face of the instrument from east to west is made to counteract any error due to erroneous setting of the vertical circle. The reversal of the face of the needle is made to counteract any error due to the centre of gravity of the needle not being quite coincident in the direction of the needle’s breadth with its axis of motion, and likewise any error due to want of symmetry of the magnetic axis. The correction for excentricity is made by reading both ends of the needle. Finally, the reversing of the poles of the needle is intended to counteract any error due to the centre of gravity of the needle not being coincident in the direction of the needle’s length with its axis of motion.

Dr Joule has suggested a modification of the dip circle in which the needle is hung on fine threads on which it rolls instead of resting on agate supports.

5. Horizontal Force Magnetometer.— The theory of the instrument for determining the horizontal component of the earth’s magnetic force has already been given in the article Magnetism, vol. xv. pp. 238 sg., and the instrument is shown in two forms, ibid., figs. 28 and 29. The corrections necessary for accurate results are explained in a paper by G. Μ. Whipple (Proc. Hoy. Soc., 1877).

(β) Instruments adapted for Travellers by Land.

6. Declinometer.— For travellers by land the unifilar instrument (§ 2), mounted on a tripod stand and duly levelled, is perhaps the most accurate kind of declinometer.

For this purpose it is furnished with a transit mirror by means of which an image of the sun may be thrown into the field of view of the telescope, and—the geographical position of the station as well as the apparent time of the observation being known—an azimuth thus determined. In order that such an observation may succeed, the following points must receive attention.

In the first place the axis of the mirror must be horizontal; the adjustment for this is made by means of a riding level. Secondly, the normal to the plane of the mirror must be perpendicular to the axis. The adjustment for this is made by a screw attached to the back of the mirror. Take some object sufficiently elevated and reflect it into the telescope, getting the object bisected by the wire of the telescope. Then reverse the mirror in its bearings. If the object remains still bisected by the wire no correction requires tobe made, but if not the screw at the back of the mirror must be moved until the object is in precisely the same position in both observations. Thirdly, the line of collimation of the telescope must be perpendicular to the plane of the mirror. In order to obtain this there is a collimating eye-piece attached to the telescope by which the sun’s light may be made to illuminate the cross wires. Now turn the transit mirror until the reflexion of the illuminated cross wires coincides with the wires themselves, in which case the line of collimation of the telescope must be perpendicular to the plane of the mirror. when this correction has been once made, note the circle reading of a small vernier which moves with the mirror and always set the mirror so as to give this reading.

By these means an accurate reading of the sun’s bearing may be made; and, the position of the place and the time of observation being known, there are tables which enable the azimuth to be at once determined.

7. Lloyd's Method of Determining the Total Force.— While the dip circle and the horizontal force magnetometer may be used by travellers in addition to their use as observatory instruments, the Rev. Dr Lloyd has devised a new method of determining the total force. The ordinary method of obtaining this is first to find the dip and the horizontal force, from which the total force can be at once determined by the equation,—total force=horizontal force x secant dip. This method is, however, open to objection in high magnetic latitudes where the horizontal force is very small and the dip approaches 90°. Now in Lloyd’s method this objection is overcome. Another circumstance which renders his method peculiarly convenient for high magnetic latitudes, where a traveller’s equipment must be kept as light as possible, is the fact that it only requires the addition of two needles to an ordinary dip circle in order to give the required determination. These needles must be carefully kept from contact with other magnets, and their poles never reversed.

Here as before we have two unknown quantities to determine, the one being the magnetic moment of the magnet and the other the total force of the earth. We must, therefore, obtain two results, the one embodying the product of the earth’s total force into the magnetic moment of the needle, while the other gives the ratio between these two quantities.

8. In order to determine the former of these, let the needle have a grooved wheel of radius r attached to its axle as in fig. 21, and over this wheel let an accurately known weight W be suspended by means of a very fine silk thread. The best way of doing this is to have a thread with two hooks of precisely equal weight at each end and then attach the preponderating weight W to one of these hooks. When this is done a new position of equilibrium will be taken by the needle. If we suppose that m denotes the magnetic moment of the needle, that i is the angle of dip at the place, and that ∙η denotes the angle which the needle in its deflected position makes with the horizon, the weighting being so made that η shall be less than i, then it is clear that the needle has been deflected out of its position of equilibrium through an angle i - ∙η. If we call this angle u and designate by R the total force at the place, we obtain the following equation of equilibrium:—

τnRsinu=Wr (1), on the supposition (which is very nearly but not strictly correct) that W denotes a constant force at all latitudes.

9. Next, in order to determine the ratio between this needle’s force and that of the earth, let it be removed and employed to deflect another substituted in its place.

When using it thus as a deflector it should be laid in a frame in an invariable position as in fig. 21. This frame is at right angles to the line between the two microscopes, and as both pieces move together the best plan is to turn the whole round until the deflected needle is visible in the centre of the field of the microscopes, in which position it is of course perpendicular to the deflecting needle. By always keeping to this arrangement we secure an invariable distance between the poles of the two needles. Suppose therefore that we have employed the needle as a deflector in the above manner, and that the deflected needle has thus been made to assume a position denoting an angle η ’ with the horizon. It has therefore been deflected from its position of equilibrium by an angle i - η , (t denoting the dip as before); calling this angle of deflexion u ’ , we obtain the following equation of equilibrium:—

Rsinu' = mU (2),

IT being a function depending upon the distance of the needles and on the distribution of free magnetism in them.

10. If we multiply together equations (1) and (2), we obtain

R 2 sinwsin√=UWr (3), in which u, u ’ are determined by observation, while W and r may be regarded as constants. U is, as we have said, a function depending upon the distance of the two needles and upon the distribution of free magnetism in them.

The magnetic moment of these needles is of course liable to alteration, but if they are carefully guarded from contact with magnets we may imagine that while their intensity alters, becoming weaker for instance, this nevertheless does not sensibly affect the distribution of the free magnetism within them, in which case the function U may be regarded as a constant quantity. The results obtained by this method of Lloyd’s fully confirm this hypothesis regarding IT; but it is essential that the two additional needles, the deflector and the deflected needle, should have their poles at no time either reversed or disturbed.

Assuming therefore the constancy of the quantity U, its value may be easily determined at any baße station where the total force has been determined independently by the ordinary method.

11. Having thus determined the value of IT, or at once of ITWr (which we may call c), let us carry our instrument to a different station and make the requisite observations. We thus obtain

R — J . c .- (4).

V sin u sin u

As this method is specially adapted for high latitudes, the dip circle employed (fig. 21) ought to be one for which the agate supports are horizontal, so as to admit of the needle being visible when the dip is nearly equal to 90°. It will also be noticed that, if the deflecting needle have the same temperature when it is used in equation (1) which it has when used in equation (2), then m in the one case is strictly equal to m in the other, and thus no temperature correction is rendered necessary.

12. A slight modification of the method now described is sometimes adopted. Instead of employing separate weights, which may be easily lost, two small holes are bored in the deflecting needle near each end. The one of these is filled with a suitably heavy brass peg when the observations are to be made in the higher magnetic latitudes of the northern hemisphere, and the other is filled in a similar manner when the observations are to be made near the southern pole. In this case therefore we must readjust the instrument as we pass from the one hemisphere to the other. A slight change must be made in the formula when this method is adopted, for it is clear that the weight will not now act always at the same constant leverage. If the weight be called W and its leverage when the needle is horizontal r, we shall have to modify equation (1) as follows:—

mRsinu≈ Wrcosη (5).

Equation (2) will, however, remain unaltered, and hence equation (3) will become

R 2 sinusinw'=UWrcosη (6).

If the quantity UWr be determined at the base station and called c', we shall have r,√⅛ 5 (7).

V sin u sin u

(γ) Instruments adapted for Travellers by Sea.

13. Azimuth Compass.— At sea the declination is generally observed by means of an azimuth compass invented by Eater. This is exhibited in fig. 22. It consists of a magnet with a graduated compass card attached to it. At the side of the instrument opposite the eye there is a frame which projects upwards from the plane of the instrument in a nearly vertical direction, and this frame contains a wide rectangular slit cut into two parts by a wire extending lengthwise. The eye-piece is opposite this frame, and the observer is supposed to point the instrument in such a manner that the wire above mentioned shall bisect the sun’s visible disk. There is a totally reflecting glass prism which throws into the eye-piece an image of the scale of the graduated card, so that the observer, having first bisected the ’ sun’s disk by the wire, must next read the division of the scale which is in the middle of the field of view. He thus obtains a reading of the sun’s position; let us call this 100°. From this, knowing the geographical position of his station and the time of the observation, he may deduce an azimuth; let us imagine that this is 70° W. Thus a reading of 100° corresponds to a position 70° W. Suppose next that the instrument is so adjusted that when the magnetic axis of the magnet is between the eye-piece and the wire the reading is 0°. It is thus clear that the magnetic meridian is 100° removed from the position

70° W. Let us imagine that the instrument is so graduated that this denotes a position 30° E. We have thus obtained the magnetic declination. If the vessel be at rest the plan generally adopted is to take the reading of the sun when rising and also when setting; a mean between the two will give that which corresponds to a geographical meridian.

14. Fox's Dip Circle.— This instrument, contrived by Robert Were Fox, is more especially useful for observations at sea. In this case it must be placed on a gimball stand and duly levelled before commencing the observation. The following are the peculiarities of this instrument :—(1) the needles have two fine pivots or axles which are inserted into jewelled sockets; (2) in order to avoid parallax there are two graduated circles, the one farther from the eye than the other, and when reading the needle the eye is to bo so placed that precisely the same reading shall be given by both circles,—the true position of the needle being thus obtained; (3) there is a rubber made of bone or ivory and roughened, the object of which is to rub a prolongation of the socket on the back of the instrument,—the friction which this rubbing causes enabling the needle to find its true position;

(4) to avoid as much as possible all effects due to friction ana adhesion, the entire socket arrangement may be turned round. The axles of the needle are thus compelled to be in contact with a different set of particles. Another way of varying the suspension is to use a magnetic deflecting arrangement attached to the back of the apparatus. Suppose that a reading of the position of the needle so deflected is now taken. Next reverse the position of the deflecting arrangement, which is done by turning a movable circle attached to this arrangement 180° round; let the position of the needle be again read. On the hypothesis that the needle is equally deflected on opposite sides of its true position in these two observations, the mean reading will give the true dip. The principle of the method of observing with this circle is precisely the same as that already described for observations on shore with an ordinary inclinometer.

15. Fox's Intensity Arrangement is merely a modification of that introduced by Lloyd, and already described in § 7.

(δ) Differential Magnetometers and Self-Recording Magnetographs.

16. In addition to determinations at fixed intervals of time, it is a point of much interest and importance to keep a continuous record of all the magnetic changes which take place at a few selected stations. This is accomplished by means of differential magnetometers. It is, however, necessary to continue to use absolute instruments side by side with differential magnetometers, because the latter (with the exception of the declination instrument) are badly fitted for recording changes of long period, such as the secular changes of the horizontal and the vertical force. The reason of this will presently be seen.

17. Early in the history of such instruments it was found that hourly observations were exceedingly laborious, and attempts were made to construct a set of self-recording magnetometers. The first set of such instruments which were brought into systematic operation were those devised and constructed by the late Charles Brooke, which have been at continuous work in the Greenwich Observatory since 1848. In 1857 John Welsh devised a fresh set of self-recording instruments, and introduced them into the Kew Observatory. These, with certain slight modifications, have formed the type of instruments supplied to a large number of magnetic observatories all over the globe.

18. As we cannot conveniently record changes of dip by a differential instrument, changes of vertical force are measured instead by a balance or vertical force magnetometer. We have thus in a differential system, whether adapted to eye observation or to continuous photographic registration, three instruments, namely, the declination, the horizontal force, and the vertical force magnetometers or magnetographs as the case may be. The most recently constructed instruments are adapted both for photographic registration and for eye observation through a telescope. The advantage of eye observations is that we see what is taking place at the very moment of its occurrence, whereas we only obtain the photographic record some time after the changes to which it relates have actually happened.

We shall therefore describe—(a) the three instruments of the Kew pattern as adapted to eye observations; (3) these instruments as adapted to continuous registration by photography; (γ) the method of determining their scale coefficients; (δ) the method of determining the temperature coefficients of the force instruments.

19. Kew Instruments—Eye Observations.— Fig. 23 shows us these instruments arranged in the relative positions recommended by Lloyd so as magnetically to interfere with one another as little as possible. We are supposed to be viewing the whole from the south. No. 1 to the right is the declination instrument, No. 2 that for the horizontal force, and No. 3 in the distance behind the central pillar (No. 4) the vertical force magnetometer. Figs. 24, 25, 26 give us the details of these three instruments in the same order as above. Connected with each instrument there is a circular mirror, or rather two semicircular mirrors, made of perfectly plane glass. One semicircular half of each mirror is attached to the magnet and moves with it, while the other half is firmly attached to the marble slab. Each magnet is enclosed in a gun-metal case with windows of perfectly plane glass; each gun-metal case is covered with a glass shade; and the whole is air-tight, and capable of exhaustion. Each magnet too is provided with a copper damper with the view of checking its oscillations. In fig. 23 will be seen two pillars of smaller size. The right-hand pillar carries a telescope, with a scale attached, to record the position of the declination magnet. The scale is reflected from the semicircular mirror moving with the magnet, and the position of this reflected scale as viewed in the telescope indicates the position of the magnet. The optical arrangement for the other instruments is similar, except that the vertical force mirror has a horizontal and not a vertical axis. The telescopes for viewing the force instruments are attached to the left-hand pillar of smaller size.

20. The Declinometer (fig. 24) consists of a magnet about 5 inches long suspended by a silk thread freed from torsion as completely as possible. To keep the state of the thread constant the glass shade should be rendered air-tight, and should contain some substance for absorbing moisture, such as chloride of calcium. It is clear that if the state of the thread remains the same, and if the position of the magnetic axis of the magnet does not change, this instrument should record faithfully the various changes of declination.

The Horizontal Force Magnetometer is exhibited in fig. 25. Here the magnet has been twisted round into a position at right angles to the magnetic meridian. It is suspended by means of two very fine steel wires some little distance apart, and thus the instrument is often called the bifilar magnetometer. These wires have the plane passing through their lower extremities differing very considerably from that of their upper. If the magnet should suddenly lose its magnetism the whole arrangement would be twisted round until the two planes coincided. This difference of plane gives rise to a couple tending to twist the magnet round in one direction while the horizontal magnetic force of the earth constitutes an equal and opposite couple, the two couples keeping the magnet in equilibrium. The couple depending upon the bi-. filar arrangement may for the present be regarded as constant, that depending on the horizontal force of the earth as variable. If the latter increase or diminish, the magnet will be slightly twisted round in one direction or the other.

In the Vertical Force Magnetometer (fig. 26), the magnet is balanced by means of a knife-edge resting on an agate plane. By means of two screws working horizontally and vertically the centre of gravity may be thrown to either side of the point of suspension, or it may be raised or lowered and the sensibility of the magnet when balanced thereby increased or diminished. These screws are so arranged that there is a preponderance of weight towards the south side of the magnet. This is neutralized partly by the magnetic force tending to pull the north end down and partly by a slip of brass standing out horizontally towards the north side. Let us suppose the system to be in equilibrium at a certain temperature; if the temperature rise (since brass expands more than steel), the leverage of the weight at the north side will increase more than that of the weight at the south. There will thus be a slight preponderance towards the north, and this may be arranged so as to neutralize to a great extent the decrease in the magnetic moment which an increase of temperature produces.

21. Magnetographs.

—The arrangement by means of which these instruments are converted into self-recording magnetographs is very simple. In fig. 23 we see a gas flame burning behind a vertical slit and placed endwise in order to render its light more intense. The light from this illuminated slit passes through a lens, and being reflected from the mirror of the declination magnet throws an image of the slit upon some sensitized paper in the central box. To speak more properly, two images are thrown, one reflected from the upper and movable half and the other from the fixed half of the mirror. The sensitive paper is wrapped round a horizontal cylinder (fig. 27), and the two images are therefore thrown upon different parts of this cylinder. But before reaching the cylinder these two images are by means of a hemicylindrical lens (shown in fig. 27) crushed up into two dots of light. The cylinder moves round regularly by clock-work once in twenty-four hours, and hence the course on the moving paper of the dot of light which comes from the fixed halfmirror will be a straight line, while that of the dot from the moving half-mirror will be a curved line depending on the motions of the magnet. When the paper is developed these lines appear black.

The arrangement for the horizontal force instrument is precisely similar to that for the declinometer; for the vertical force it is somewhat different, the illuminated slit being horizontal and not vertical, while the mirror oscillates on a horizontal axis and not on a vertical one; the hemicylindrical lens too and the cylinder are vertical and not horizontal. It was found necessary to put the plane of motion of the vertical force magnet 15° out of the magnetic meridian for the following reason. The axes of the telescopes are respectively 30° inclined to the tubes which go from the magneto meters to the central box, and hence had the vertical force magnet swung in the magnetic meridian it would have been necessary to place the mirror inclined at the angle of 15° to the axis of motion of the magnet. This was tried, but it was found that in this position of the mirror the correction for temperature was so excessive that the instrument became a thermometer and not a magnetometer. The mirror was therefore put in a plane passing through the axis of motion of the needle, the needle being made to move in a plane inclined 15° to the magnetic meridian.

22. Scale Coefficients of Differential Instruments.— It is necessary to know the value of one division of the scale in the magnetometer or of one inch difference in the ordinate of the curve impressed on the photographic paper in the magnetograph. In the declination instrument it is only necessary to obtain the angular deviation corresponding to one division, and this may be done at once by a series of measurements. In the horizontal and vertical force instruments we wish to obtain the value of one division in parts of force. There is more than one method by which this can be accomplished, but that of John Allan Broun is probably the simplest, and it is, we believe, the one adopted at most of the various observatories .possessing self-recording instruments. It is given in the British Association Reports, 1859.

23. Temperature Coefficients of Differential Force Instruments.— Broun has devoted a great deal of attention to the subject of these coefficients, and has come to the conclusion that the best and most unobjectionable method of determining them is to compare the instrumental readings on days when the temperature is high with the readings on days when the temperature is low.

24. By differential instruments the components of a force affecting the magnet are determined in three directions at right angles to each other. It does not, however, follow that this force is entirely due to changes in the magnetism of the earth. We know that certain forces connected with the sun affect the earth’s magnetism, and on certain occasions at least these forces manifest themselves as currents in the upper regions of the atmosphere and in the crust of the earth. Now such currents will have a direct effect upon the needle as well as an indirect effect through the changes which they may produce in the magnetism of the earth. The total influence on the needle will therefore be made up of these two elements, the one denoting the direct influence on the needle of the disturbing force, and the other the indirect influence through the change produced in the earth’s magnetism. No attempt has yet been made to separate the action of these two elements.

25. Self-recording instruments after the Kew pattern have been supplied to observatories at the following places :— [table]

There are also self-recording magnetographs of other patterns at Toronto, Montsouris (Paris), Greenwich, Wilhelmshaven (?), Cape Horn, and Havana (?).

We understand that Professor W. G. Adams is at present engaged in making a comparison of simultaneous curves from various stations of these lists?

Magnetic Poles of the Earth—Secular Variation.

26. Magnetic Poles of the Earth.— In the article Magnetism it has been shown that Dr Gilbert of Colchester had at a very early period grasped the important truth that the earth is a magnet, a truth which was afterwards mathematically demonstrated by Gauss. It was reserved for Halley, the contemporary of Newton, to show that the earth must be regarded as having two poles in the northern and two poles also in the southern hemisphere, so that, unlike ordinary magnets, its magnetic system has four poles altogether. Before proceeding further it will be desirable to state what it was that Halley actually did and what are the conclusions to be derived from his investigations. It has been remarked by Professor Stokes that, while in an ordinary bar magnet we may practically regard the pole as having a physical reality and as being the cause of well-known attractions and repulsions, we are not entitled a priori to assume that a point of maximum force in a large spherical magnet like the earth must necessarily be the seat of attractions and repulsions after the same manner as the pole of an ordinary bar magnet. It is to be determined by observation to what extent the earth actually preserves an analogy to an ordinary magnet. Now Halley’s conclusions were derived from the pointing of the declination needle, since in his day there were no observations possible on total magnetic force. He argued that there are two points or poles in the northern hemisphere to which the needle appears to be attracted, one in the upper region of America and one above Siberia. So far this conclusion is hardly anything more than a formal one derived from the grouping together of observations. He asserted that these would be as they are known to be if we imagine two such poles or foci of force each exercising a causal influence on the magnetic needle. And the justification of Halley’s way of regarding the earth is found in the fact that when force observations came to be made two such foci of force were actually found to exist. We do not, however, mean to imply that these foci have causal properties exactly similar to the poles of a bar magnet, for this is not the case.

In order to exhibit the process of reasoning which led Halley to his conclusion, let us first imagine that the earth has only a single pole or force-focus in the northern hemisphere, and that this is coincident with its geographical pole; then, assuming that this pole has a causative influence on the needle’s declination, we should expect all needles to point everywhere due north. If, however, this pole be not coincident with the north pole of the earth, let us draw a meridian circle passing through the magnetic pole and complete it round the earth so as to divide the earth into two halves. At all points in this meridian circle the needle might be expected to point due north, while in the one half of the earth so divided it should point to the east and in the other half to the west of true north. In the next place let us imagine that the earth has two north magnetic poles or foci of equal strength, both being at the same latitude, while their difference in longitude is 180°, and let us draw a complete circle of meridian passing through these poles (fig. 28). Let us start from a point in this circle under one of these poles and pursue our journey eastwards along a circle of latitude. At first the needle will point due north. As we move eastwards the needle will point westwards to the pole we are leaving until we come to a region half-way between the two poles, where it will be equally solicited by each, and will therefore again point due north.

Let us call the space we have travelled over since we set out A. As we proceed the needle will now be under the predominant influence of the second pole to our right, and will therefore point to the east until we arrive at the meridian under the second pole. This second space which we have travelled over let us call B. As we proceed we pass through a space C where the needle again points to the west until being once more equally influenced by the two poles it will point due north. After this we pass through a space D of easterly variation until we arrive once more at the point from which we started.

Thus there are now four spaces instead of two, and these are shown in fig. 28, where the centre of the circle represents the north geographical pole of the earth, and its circumference the equator. If pole 2 be inferior in power to pole 1 the spaces B and C will be smaller in size than A and D.

27. This last is an arrangement of things that agrees very well with the results of observation, if we add that the two poles are not precisely 180° removed from one another in longitude. Fig. 30 represents lines of equal magnetic variation in 1882. There are two lines extending throughout both hemispheres at all points of which there is no variation, and also an oval-shaped district in the northern hemisphere throughout all points in the circumference of which we have no variation. These facts are inconsistent with the hypothesis of a single pole, but they are quite consistent with that of two poles or foci of force, one in northern America and the other in northern Asia, the former being stronger than the latter. In order to see this let us take our stand at the great line of no variation which passes through North America and travel eastwards. We are just south of the American pole or focus, while the Asiatic pole or focus is nearly 180° off, and hence the needle points due north. As we proceed eastwards we leave the American or strongest pole to the westward of us, and hence we have a region of west variation which we have agreed to call A. As we begin to approach the eastern side of Europe we get nearer the Asiatic pole or focus, and at length the line of no variation is reached, the tendency of the American pole to pull the needle to the west being balanced by the tendency of the Asiatic pole to pull it to the east. After this, easterly variation predominates throughout a region B until at length we come to a point in the western boundary of the oval where we may imagine ourselves to be directly south of the Asiatic pole, while the American pole is nearly 180° distant; once more the needle points due north. As we still travel eastwards we leave the Asiatic pole, which is now the predominant one, to our left, and hence we have here a region C of westerly declination. At length we come to the eastern boundary of the oval, where the tendency of the Asiatic pole to pull the ιieedle to the west is balanced by the tendency of the American or stronger pole (acting now towards the right) to pull it to the east, so that we have once more a point of no variation. After this the American pole predominates, and we have a large region D of easterly variation until we travel round once more to the point from which we started.

28. This train of argument receives, as we have already mentioned, corroboration from the fact that in the map of total force we perceive two foci of maximum force, one in northern America and the other in northern Asia, that in America being the strongest. This evidence was not, however, in existence at the time of Halley, and his hypothesis of two poles does the greater credit to his sagacity, inasmuch as he had to deduce it from a comparatively small number of observations of declination and dip, those of force being altogether wanting.

29. We have hitherto spoken of two poles or, more properly, foci of maximum force, the positions of which are of course best pointed out in fig. 29; but we have seen that the existence of such foci was first conjectured from the behaviour of the lines of variation or declination. Now it will be noticed by looking at the variation map (fig. 30) that all the lines of equal magnetic variation appear to converge to a point in the extreme north of the American continent. This point is not, however, coincident with the chief focus of force, which lies decidedly to its south; but it is no doubt coincident with the point denoting a dip of 90°, the locality of which may be inferred from the map of magnetic dip (fig. 31), and it is likewise no doubt coincident with the position of a zero of horizontal force which may be inferred from the map of horizontal force (fig. 32). Thus we have a point to the extreme north of America which has the following properties :—(1) the various lines of declination converge toit; (2) the needle points vertically downwards at it; and (3) the horizontal force vanishes at it. At this point therefore the horizontally balanced needle, having no horizontal force acting upon it, will point in any direction.

This point is, strictly speaking, the pole of verticity, but, inasmuch as there is only one such point in each hemisphere, these may for convenience sake be termed the magnetic poles, so that we speak of two centres or foci of maximum force and one pole in each hemisphere.

In the northern hemisphere Sir Frederick Evans assumes the stronger or American focus to be in 52° N. and 90° W., and the weaker or Siberian focus in 70° N. and 115° E. In the southern hemisphere he assumes the position of the stronger focus to be 65° S. and 140° E., and of the weaker focus probably 50° S. and 130° E., these being thus not far separated from each other or from the magnetic pole. The nearness together of the southern foci is prob ably the reason why the total force is greater at the southern than it is at the northern foci.

The magnetic pole (of verticity) in the northern hemisphere was reached by Sir James Ross in 1831. The position of vertical dip was observed by him to be 70° 5' N. and 96° 43' W. The magnetic pole (of verticity) in the southern hemisphere was nearly attained by the same navigator in a voyage made in 1839-43. Its position is probably 73 f S. and 147⅜ o E.

The line of no dip is called the magnetic or dip equator— its position is given in figs. 29 and 31. The line connecting all the points where the magnetic intensity is least is called the dynamic equator. It coincides very nearly with the dip equator.

30. Secular Variation.— The earth then as a magnet must be supposed to have two sets of centres of force. We shall next attempt to show that these centres cannot be regarded as constant both in position and intensity.

It should be premised that, while there is no well-established evidence to show that either the pole of verticity or the centre of force to the north of America has perceptibly changed its place, there is, on the other hand, very strong evidence to show that we have a change of place on the part of the Siberian focus and also on the part of its analogue in the southern hemisphere.

Table I. (p. 166), given by Gilpin {Phil. Trans., 1806), exhibits the change in the position of the needle in Great Britain from the earliest observations up to the beginning of the present century.

31. Between the dates recorded in this table the needle has been pointing more and more to the west, which implies either a relative increase in the power of the American as compared to the Siberian focus, or a motion of the Siberian focus from west to east. On the first supposition the lines to the eastward of the Siberian focus— for instance, the line of no variation depending on a balance between it and the American focus—should be drawn in towards it, or they should travel westwards; but if the latter supposition is true, or this focus has been moving eastwards while retaining its force, the lines to the east of it should be found moving eastwards also. There is strong evidence that the latter is the case, and that in the northern hemisphere there has been a long-continued progression to the eastwards of the system of magnetic lines on both sides of the Siberian focus. In the southern hemisphere also we have proof that the analogous focus has been travelling, not from west to east, but from east to west.

32. There is some reason to believe that the eastward motion of the Siberian focus has been recently reversed, and that it is now going from east to west. Table II. shows the declination observed at Bushey Heath (Herts) during 1817-20, and at Kew from 1858 to 1882.

It would appear from Table II. that the maximum westerly declination was reached in 1818, and that the needle has since that date been travelling eastwards. A similar change has taken place at other stations; and, although these changes are not strictly simultaneous at the various stations, they have yet been sufficiently general and near together in point of time to indicate that some [table]

change has probably taken place in the movement of one set of the magnetic foci of force.

33. Halley sought to explain the four-pole theory and to account [table] [table] [table]

for the secular change by imagining a solid globe or terella, λ concentric with the earth but rotating independently of the external shell and having a slightly different period of rotation,—the shell having two poles and the terella two others. While continuing to admire Halley’s sagacity, we shall not now be disposed to allow such a constitution of the interior of the earth, but will rather be led to look to some external influence as the cause of the secular variation.

While we have strong evidence that the Siberian focus has changed its position, we cannot assert that the American focus has been absolutely stationary, or that neither focus has experienced any changes of force. On these points we must be content to be guided by observation alone.

34. It has been supposed by some magneticians that it is possible to compute with something like certainty the particulars of the motions of the magnetic foci. Hansteen more especially (1811-19) computed both the geographical positions and probable periods of revolution of this dual system of foci of force round the terrestrial pole. Sir Frederick Evans has discussed in connexion with the subject all the most recent observations, and points out two objections to any such theory as that of Hansteen, viz., (1) that, while a magnetic turning point has been reached in certain regions, there are large portions of the earth in which this change has not yet been accomplished, and (2) that in certain districts of the earth very great changes in force have taken place. “If we turn,” he says, “to the continent of South America and its adjacent seas, we shall find a diminution of the intensity of the earth’s force now going on in a remarkable degree. An examination of the recent observations made by the ‘ Challenger ’ officers at Valparaiso and Monte Video, compared with those made by preceding observers, shows that within half a century the whole force has respectively diminished one-sixth and one-seventh,—at the Falkland Islands one-ninth.” On the whole, while there is strong evidence that the Siberian focus has until recently been travelling eastwards, and its analogue westwards, and evidence less conclusive that recently a turning point in this motion has been reached, we are disposed to think with Sir Frederick Evans that a formal theory like that of Hansteen does not agree with recent observations. We shall revert to this subject.

35. In Tables III. and IV. certain yearly values of declination, dip, and horizontal force are given for various stations.

Inequalities in or connected with Terrestrial Magnetism depending on the Sun.

36. As there is a marked likeness between the ways in which the sun dominates over the two great divisions of terrestrial phenomena, meteorology and magnetism, let us endeavour to describe the sun’s effect upon the latter by referring to its influence on the former, the chief peculiarities of which are well known to all. We find that the temperature of the air at a given station is subject to a diurnal fluctuation having its minimum value shortly before sunrise and its maximum early in the afternoon. We find likewise that the mean temperature for the day, as well as the amplitude of this diurnal oscillation, depends upon the season of the year, both being greatest about midsummer and least about midwinter. Now, if this were the only manifestation of solar influence upon this particular element, it would be possible to predict the temperature tor any hour of any day once the mean temperature, the diurnal variation of temperature, and the modification of these for different seasons of the year had been well ascertained. But this amount of regularity is very far from taking place,—the march of temperature being frequently interrupted, cloaked, perhaps even reversed, by the advent of peculiar weather. Thus we may have very cold weather in midsummer and very warm weather in midwinter, or we may have a very cold afternoon and a very warm early morning, by which means the ordinary conditions of temperature will be completely reversed. In like manner weather interferes even to a greater extent with the diurnal oscillation of the atmospheric pressure, so that, in British latitudes at least, it is only possible to obtain this correctly by means of a long series of observations.

Weather, however, does not consist of a perfectly lawless interference with periodical phenomena, but is subject to laws of its own, some of which we are beginning to discover. Sometimes weather may exalt or depress the diurnal fluctuation of temperature without otherwise affecting its character; but sometimes too the turning-points and the general appearance of this fluctuation are greatly influenced by peculiar weather.

37. Now it is believed that we have something of this· kind in those fluctuations depending on the sun to which the elements of terrestrial magnetism are subject. Let us take the declination as the most easily studied of the three magnetic elements, and suppose that we are engaged in considering the traces denoting the fluctuations of declination as derived from a set of self-recording magnetographs in Great Britain. Here we shall at once be able to recognize in an unmistakable manner the diurnal variation depending upon the position of ‘the sun, in virtue of which a freely-

suspended magnetic needle reaches the easterly extreme of its range about eight in the morning, and the westerly about two in the afternoon. We shall likewise perceive that the range of this diurnal fluctuation is greatest at midsummer and least at midwinter. In fine, the characteristics of this fluctuation, depending as they do upon the hour of the day and the season of the year, are not very different from those exhibited in the diurnal fluctuation of atmospheric temperature. But, however thoroughly we may have ascertained the mean declination and its diurnal oscillation, as well as the modifications of these depending on the season of the year, we shall nevertheless find it impossible to predict the exact position of a freely-suspended magnet at any moment of a particular day. Here then too we have something which may be called magnetic weather, and which interferes with the regular progress of the systematic fluctuations of the magnet. Magnetic weather has, like its meteorological analogue, a set of laws of its own, some of which we are beginning to find out. Sometimes magnetic weather may exalt or depress the diurnal fluctuation of declination without affecting its character, but it is imagined that at other times the turning points and general appearance of this fluctuation may be greatly influenced by peculiar magnetic weather.

38. There is, however, a kind of magnetic change which, so far as we know at present, is not analogous to anything in meteorology, and introduces an additional element of complexity in any attempt to analyse the fluctuations of terrestrial magnetism. We mean the well-known magnetic disturbances or storms which occur simultaneously in places very widely apart. Under these circumstances it becomes a question how we can best deal in a practical manner with this complicated system of things.

We do not think that with our present knowledge any better system can be adopted than that first introduced by Sir Edward Sabine in his discussion of the results of the colonial magnetic observatories. Suppose that we have hourly magnetic observations at a station, then first of all we should arrange these into monthly groups—each hour by itself. We should then reject as disturbed observations all those which differ by more than a certain amount from their respective normals of the same month and hour,—the normals being the hourly means in each month after the exclusion of all the disturbed observations. This method enables us, by its exclusion of disturbances, to ascertain with much accuracy the true form of the solar-diurnal variation of the magnetic elements at a given place corresponding to every month of every year, provided only that the observations are sufficiently numerous. On the other hand it will probably fail in accurately giving us the variations from day to day of the ranges of these diurnal fluctuations caused by the advent of peculiar magnetic weather, —inasmuch as the records of the extreme effects of such weather will probably be cut off from the undisturbed observations and reckoned among the disturbances.

For instance, it is known that the solar influence on terrestrial magnetism varies from year to year, and it is suspected that there are also short-period fluctuations of solar influence. It would not, however, be a safe proceeding to attempt to estimate numerically this last-mentioned element of fluctuation by taking the successive diurnal ranges of those observations at any station, reckoned as undisturbed, by the above process, and plotting them as successive ordinates of a curve, and then supposing that this curve would give us a true graphical representation of solar changes. It would rather probably represent such changes with the tops and bottoms of the larger fluctuations cut off. But if the undisturbed observations fail in this respect we can hardly be wrong in supposing that there has been eliminated from them, as far as possible, all influence due to magnetic storms, and hence that they will afford us a much better means of estimating small fluctuations, such, for instance, as those due to the moon, than we could have had without their aid.

Finally, with regard to that portion of the observations selected as disturbed, we are probably not certain that every such observation represents a true disturbance, or that the absolute times of occurrence of the various observations selected as disturbed at one station will be the same as those at another. Nevertheless Sir Edward Sabine has shown that at the Kew Observatory certain laws of disturbance deduced from the whole body of observations selected as disturbed are closely reproduced when this selection is made on a narrower basis—ninety-five days of prominent disturbance being alone taken. With these prefatory remarks we shall now proceed to discuss the diurnal inequality of terrestrial magnetism.

39. Total Diurnal Inequality Defined.— It will be seen further on that disturbed as well as undisturbed observations are subject to a diurnal variation, but these two variations are different, and the name diurnal inequality is generally given to the compound variation which is the joint resultant of the two. Solar-diurnal variation is that portion of the compound inequality which refers to undisturbed observations, while that which refers to disturbances has received the name of disturbance-diurnal variation. It would appear that in the United Kingdom, and perhaps throughout Europe, the total diurnal inequality is not very greatly different either in character or range from its most important component the solar-diurnal variation, at least so far as the declination is concerned. When the diurnal oscillation of a freely-suspended magnet was first observed, the subject of magnetic disturbances was not understood, and the early individual determinations which have been handed down to us are not such as to justify the expenditure of any very great labour upon them for the purpose of separating the disturbed from the undisturbed observations. Inasmuch, however, as the total diurnal inequality of declination (which is in reality the element given by these early observations) does not greatly differ from the solar-diurnal variation, we may with much justice and little risk of error give the history of these early observations in connexion with that of the solar-diurnal variation of declination, which is by far the best known, and perhaps the most important, of all the various magnetic changes produced by solar influence.

40. Solar-Diurnal Variation of Declination.— Graham, an instrument maker of London, discovered in 1722 that a freely-suspended magnetic needle is subject to a diurnal oscillation of definite character. The next observer was Canton, who in 1756 began a series of nearly four thousand observations, which he communicated to the Royal Society on December 13, 1759, and from which he concludes that the range of the diurnal variation is greater in summer than in winter. Macdonald’s observations at Fort Marlborough in Sumatra in 1795 (Phil. Trans., 1796), and Duperrey’s in the tropics in 1825, were perhaps the first that might lead us to conclude that the amplitude of the diurnal oscillations of the needle is less in the tropics than in middle latitudes, and that the motion of the needle in the southern hemisphere is in the opposite direction to that in which it moves in the northern hemisphere at the same hour.

41. Semiannual Inequality. — The existence of these early observations had led some magneticians prematurely to conjecture that there must be a line somewhere near the equator at which there is no horary variation in declination. In 1847 Sabine communicated to the Royal Society the results of five years’ observations at St Helena, showing that at that station for the half of the year beginning at the vernal and ending at the autumnal equinox the motion of the needle corresponds nearly to that in the northern hemisphere, whilst for the other half it corresponds nearly to that in the southern hemisphere. Sabine afterwards confirmed and extended his conclusions regarding the semiannual inequality by discussing the results obtained at the various colonial magnetic observatories. More recently, as the result of twelve years’ observations at Trevandrum, at an observatory established by the rajah of Travancore, John Allan Broun gave in a very complete form the laws of change of the solar-diurnal variation of magnetic declination near the equator, showing the extinction of the mean movement near the equinox.

42. Perhaps the best way of exhibiting what really takes place is the following, which is that adopted by Sabine.

The mean annual value of the solar-diurnal variation is of what may be called the northerly type in places of middle latitude in the northern hemisphere, and of what may be called the southerly type in places of middle latitude in the southern hemisphere. Now let us take a northern station, and consider the mean form of its solar-diurnal variation for the six months beginning with the vernal equinox. Here we shall have an oscillation of the northerly type with a range greater than the annual range. For these six months, therefore, we may imagine that the annual range has been supplemented by the superposition on it of a variation with a type similar to its own. At the same station, during the other six months, the solar-diurnal variation is less than the mean of the year, as if the anmιal variation had been depressed by the superposition on it of a variation with a type the opposite of its own, that is to say, with a southerly type. At a station in the southern hemisphere, again, the mean annual form of the solar-diurnal oscillation is of the southerly type, reduced during the six months beginning with the vernal equinox by the superposition on it of a variation of northerly type, and increased during the other six months as if by the superposition of a variation of southerly type. Thus when the sun is north of the equator we may superpose a variation of the northerly type upon both hemispheres, with the effect of increasing the range in the northern hemisphere and diminishing it in the southern; and while the sun is south of the equator we may superpose a variation of the southern type upon both hemispheres, with the effect of diminishing the range in the northern and increasing it in the southern, hemisphere.

Near the equator, as at Trevandrum, where Broun made his observations, we find the mean annual value of the solar-diurnal variation to be extremely small, if not altogether evanescent. During the six months beginning with the vernal equinox the type is entirely northerly, while for the remaining six months of the year it is entirely southerly in character. In fine, at this station the solar-diurnal variation changes its character at the equinoxes, at which time we have, as already observed, an extinction of the mean movement,—not indeed an absence of all variation, but rather a variation having an undecided character, which for a few days may be of one type and then of the very opposite. There is movement, but no mean movement.

43. In the following table (V.) the solar-diurnal variation is given for Kew, Trevandrum, and Hobart Town. Of these places the first denotes a station in middle latitude (northern hemisphere), the second an equatorial station, and the third a station in middle latitude (southern hemisphere). [table]

In this table deflexions towards magnetic east are reckoned positive, deflexions towards magnetic west negative. The scale is in minutes of arc.

Also in fig. 33 we have a graphical representation of the solar-diurnal variation for the whole year at these three stations, from which it will be seen that the range at Trevandrum is extremely small, and that the curve for Hobart Town is opposite in appearance to that at Kew.

Finally, in fig. 34 we have a graphical representation of the semiannual inequality or difference from the whole year’s mean of the two half-yearly means of Table V., the one half-year (that with thick lines) commencing at the vernal and the other at the autumnal equinox. It will be seen from this figure that the semiannual inequality is of the same character in both hemispheres, the likeness extending even to its minor peculiarities.

44. Change from Month to Month.— Charles Chambers, director of the Kolaba Observatory, Bombay, remarks (Trans. Roy. Soc., December 10, 1868) that “the regular progression from month to month in the diurnal variation is so distinctly shown in the Bombay observations as to lead, on a first inspection, to the supposition that the law of variation is identical throughout the year, the extent only (including a reversal of direction) varying from month to month. But in this respect a different exposition of the character of the variation in different months shows that the first thought would be inaccurate.” He then proceeds to discuss at length the monthly values of the solar-diurnal variation at Bombay. Broun has likewise (Trevandrum observations) discussed at length the solar-diurnal variation at the Trevandrum Observatory. It would hardly be of service to reproduce here the results of these discussions; but when such analyses become sufficiently extensive they may be expected to throw light upon the cause of the solar-diurnal variation.

In the following table we have mean monthly values of the declination range at the Kew Observatory corresponding to forty-eight points in the year—derived from sixteen years’ observations :—

Table VI.— Containing Monthly Means (ιιnit≈22' - 04) for Fortyeight Points in the Year of the Kew Solar-Diurnal Declination Ranges. Thus January (0) gives the monthly mean of which the middle date is the very commencement of the year, January (1) that for one week after the commencement, and so on. [table]

It will be seen from this table that, while we have a maximum about the summer and .a minimum about the winter solstice, we have unmistakable indications of maxima at or about the equinoxes. This does not take place at a tropical station such as Trevandrum.

45. Behaviour near the Magnetic Pole.— Figs. 33 and 34 exhibit the most prominent features of the solar-diurnal variation of declination in the extra-tropical regions of the northern hemisphere. If an observer stand over the centre of the needle and look towards the marked end, or that which points to the north, he will perceive a deflexion towards his right hand which will reach its extreme about 8 a. μ. and a deflexion towards his left hand which will reach its extreme about 2 p.m. But are these deflexions to the right and left hand of geographical or of magnetical north? This question has been answered by Sabine in his discussion of the results of hourly observations of the magnetic declination at Port Kennedy (Phil. Trans., 1863, p. 660). This station is 72° 0' 49" N. lat. and 94° 19' W. long., and here the marked end of the needle, while it points towards the magnetic pole, points in reality about 35° to the west of south. Now the marked end of the needle when viewed at 8 A. Μ. is seen at Port Kennedy to have moved to the geographical west but to the magnetical east. It would thus seem that throughout the extra-tropical regions of the northern hemisphere the 8 a.m deflexion of the needle is always towards the magnetic east but not always towards the geographical east, while the deflexion at 2 p.m. will always tend towards the magnetical west but not always towards the geographical west. In fine the oscillations have reference to the north magnetic pole of the earth and not to the north geographical pole. No observations of this nature have been made in the southern hemisphere.

46. Long-Period Inequalities of Declination Range.— It was first observed by Lamont that the yearly values of the diurnal range of magnetic declination at Munich presented signs of a long-period variation. In 1852 Sabine (Phil. Trans., 1852, p. 103) showed that this inequality corresponded in its progress with that of the frequency of black spots on the surface of the sun.

The existence of black spots on the disk of the sun was long ago known to the Chinese. In Europe they were first scientifically observed after the invention of the telescope, and it was deduced from their behaviour that the sun revolves about his axis in about twenty-six days. Hof rath Schwabe of Dessau, from a long series of forty years’ observations of the sun, was the first to show that the state of the sun’s surface as regards spots was not uniform, but that their frequency was subject to an inequality the average period of which was about eleven years. Other inequalities both of longer and shorter periods have been supposed to exist, but the eleven-yearly period is the most prominent and is best assured. Although the sun-spot catalogue of Schwabe is the first with pretensions to completeness, yet Professor Rudolf Wolf has endeavoured to render observations of sun-spots made at different times and by different observers comparable with each other, and has formed a list exhibiting approximately the relative number of sun-spots for each year. This list extends back into the 17th century, and is of great value in confirming past all doubt the existence of the eleven-yearly period. It will appear below that the sun is probably to be regarded as giving out most light and heat at those times when sun-spots are most frequent. The most accurate and now universally adopted method of estimating sun-spots is to take the spotted area expressed in millionths of the sun’s visible hemisphere.

To return from this digression,—the correspondence between sunspots and declination ranges detected by Sabine was of such a nature that years of large declination range agreed with those of many sunspots, and vice versa. In the same year with Sabine (1852) Dr Rudolf Wolf and Μ. Gautier independently remarked the same coincidence. Subsequent discussions have entirely confirmed the fact of this connexion, and in May 1879 William Ellis (Phil. Trans., 1880, p. 541) showed that the observations made at the Greenwich Observatory during the years 1841—77 indicated a relation of this nature between the diurnal ranges of horizontal force as well as those of magnetic declination on the one hand and the amount of sun-spot frequency on the other. The general character of this coincidence between sun-spot frequency and declination range is exhibited graphically in fig. 39 below.

47. Ratios of Ranges in Years of Maximum and Years of Minimum Sun-Spot Frequency.— Broun (Trans. Roy. Soc. of Edin., vol. xxvii. ) has shown that the ratios of the diurnal ranges of declination in years of maximum to those in years of minimum sun-spot frequency for places widely apart on the surface of the earth are very nearly alike. This will be seen from the following table :— [table]

48. Closeness of Correspondence—Lagging behind of Ranges.— Stewart has shown from a discussion of the declination ranges at Kew, Trevandrum, and Prague (Proc. Roy. Soc., March 22, 1877, February 8, 1878, May 16, 1878) that this correspondence between the state of the sun’s surface and the diurnal range of declination extends to inequalities of short period as well as to that of which the period is approximately eleven years, but that a particular state of the sun’s surface precedes in point of time that of the declination range to which it corresponds,—in fine, that the solar cause precedes the terrestrial effect, which latter lags behind to an extent that is sometimes considerable. These conclusions have been confirmed by Ellis (ut supra), and have likewise been extended by him to the horizontal force. The close nature of this correspondence, as well as the lagging behind of the terrestrial magnetic effect, will be seen from fig. 35.

There are indications that this lagging behind of the magnetic effect is greater for sun-spot inequalities of long than for those of short period, a method of behaviour quite similar to what we find in meteorological phenomena.

49. Analysis of Lonq-Period Inequalities. — We possess no sun-spot data sufficiently accurate for a discussion, in a complete manner, of questions relating to solar periodicity before the time when Schwabe had finally matured his system of solar observations, which was not until the year 1832. We have, however, a much longer series of the diurnal ranges of magnetic declination, which we have seen to follow very closely all the variations of sun-spot frequency, so it is conceivable that they may give us a better estimate of true solar activity than that which can be derived from the direct measurement of spotted areas.

These considerations have induced Messrs Stewart and Dodgson to attempt an analysis of the diurnal ranges of magnetic declination, their method being that which has been pursued by Baxendell and probably other astronomers with observations of variable stars. The observations at their disposal for this research were those which had been used by Professor Elias Loomis in his comparison of the mean daily range of the magnetic declination with the extent of the black spots on the sun (American Journal of Science and Arts, vol. 1. No. cxlix.). These observations are recorded as monthly means of diurnal declination range, and it was found necessary to multiply each by a certain factor, first on account of the change of declination range from one month to another, and secondly to bring them all to the standard of the Prague observations,—Prague being the place where the longest series of such observations has been made. For this latter purpose precisely the same corrections were applied as those made by Professor Loomis.

The result of this analysis has been to indicate the existence of three inequalities,—two dominant ones with periods of about ten and a half and twelve years, and a subsidiary one with a period of about sixteen and a quarter years. By these means the observed annual values of declination range have been reproduced with an average error of 39". The amount of agreement between the observed and calculated values will be seen from the following diagram (fig. 36).

50. Notwithstanding the considerable amount of agreement between the results of observation and calculation which appears in the diagram, it would seem that the series of observed values at present obtainable is too short to render the analysis a very accurate one. It will certainly not bear carrying back forty or fifty years beyond its starting point, which was in 1784, and it would be very hazardous to carry it forward any consider able length into the future. It will be seen that calculation indicates a maximum of declination range about 1884, but not so pronounced a maximum as that of 1871. Here then we have a prevision which observation will either fulfil or contradict, giving us a practical test of the value of this analysis.

51. The remarks now made would seem at first sight to imply that we are not yet furnished with sufficient yearly records either of declination ranges or of accurate sun-spot observations to enable us to analyse the long-period solar inequality with such completeness as to carry our calculations more than a very short distance into the future with any chance of success, and that we may have to wait for another hundred years’ observations before we are able to do so. On reflection, hoλvever, it would seem that long-period inequalities may be caused by the superposition of those of short period, and thus that an analysis of the latter may lead to that of the former. It would relieve us if this were found to be the case; for the observations at our disposal may be sufficient to enable us completely to analyse short-period inequalities, assuming that we have in such the elements of a true periodicity.

A remark made by the authors of the above analysis would seem to indicate that a connexion of this nature between long and short periods does in all probability exist. It is a well-known fact that the so-called eleven-yearly oscillations of declination range are at certain times large and at other times small. Thus, for instance they have been large for the last forty years, but they were small about the earlier part of the present century. Now it is clear from an inspection of the observations (see fig. 36) that a series of large oscillations is accompanied with an exaltation of the base line, or line denoting average efficiency, while a series of small oscillations is accompanied with a depression of the same. The result is a long-period curve of the base line, the beat period, so to speak, of the eleven-yearly inequality.

Now a phenomenon precisely similar occurs in connexion with shorter periods. If we take inequalities having a period of three or four months, we find that such are alternately well-developed or of large range and badly-developed or of small range, and that a large range of such is accompanied with an exaltation of the base line or line of average efficiency, while a small range is accompanied with a depression of the same. The result is a curve of the base line of which the period is roughly speaking eleven years. May we not therefore imagine that the so-called eleven-yearly period, or, to speak more correctly, the ten and a half and twelve-yearly periods into which the eleven-yearly period may perhaps be analysed, may be in reality beat periods for shorter disturbances? Is it not therefore possible that a study of these shorter periods may give us information regarding the nature of the eleven-yearly period, whether for sun-spots or declination ranges, which the small series of actual observations is incompetent to afford?

52. Declination-Range Weather.— Allusion has already been made to magnetical weather as perhaps having laws similar in some respects to those which regulate meteorological weather. Now the diurnal ranges of magnetic declination and those of atmospheric temperature present us with elements of the two weathers that can easily be discussed. Again there is strong evidence for supposing that an element of meteorological weather, such, for instance, as temperature-range, travels as a rule from Λvest to east, so that a peculiar style of temperature-range might be expected to appear first in America and some days afterwards in Great Britain. It becomes therefore a question for inquiry whether this travelling from west to east applies also to magnetical weather as evidenced by the diurnal declination-range. Stewart is of opinion that this law of travelling applies to both, but that magnetical weatheç travels faster than meteorological (see Proc. Roy. Soc., January 10, October 23,1879, and June 9, 1881). From the preliminary discussion made by him it would appear that Kew lags behind Toronto as regards phase of magnetical weather by 1 ‘6 days, that Prague lags behind Kew 0 ’ 7 days, and that Trevandrum lags behind Kew by 9 ’ 7 days. This conclusion cannot, however, be regarded as established until it is confirmed by a more complete discussion of observations.

53. Disturbance-Diurnal Variation of Declination.— Magnetic storms (§ 38) were so named by Baron Humboldt, one of the first observers of such phenomena. From observations at Paris, Berlin, and Freiburg he found that very frequently, though not universally, these three stations were simultaneously affected by such storms. The observation of magnetic disturbances was afterwards pursued in a systematic manner by Gauss and Weber of Göttingen. Term days were instituted for this purpose by these observers,—that is to say, periods each of twenty-four hours length during which observations were simultaneously made at intervals of five minutes at Gottingen and about twenty other stations distributed generally over the continent of Europe. Finally, the establishment by the British Government of the colonial magnetic observatories, and the energy and sagacity of their director, Sir E. Sabine, have very greatly increased our knowledge of these remarkable phenomena..

Sabine has not merely separated the disturbed from the undisturbed observations as explained in § 38, but he has divided the former into two categories—(1) those tending to increase westerly declination and either element of force, and (2) those tending to diminish the same. He finds that these two categories obey different laws, from which he argues that there are at least two sets of disturbing forces. In fact, if we have to give up the idea of a single force of constant type, it is natural to ask if the phenomena of disturbance can be approximately represented as due to the united action of two independent types of force. It was probably some such idea that led Sabine to separate disturbances into these two categories above mentioned. Here there is no attempt to assert that these two types represent an ultimate and complete analysis of the forces concerned. We merely use the separation as the most convenient method at our disposal in the present state of our knowledge for ascertaining whether there be indications of a dual system.

54. Results in the Northern Hemisphere.— Sabine’s method of viewing the phenomena has enabled him to obtain the disturbance-diurnal variation for the following stations [table]

The above stations have been so chosen that Kew may be regarded as on one side and Peking and Nertchinsk as probably on the other side of the Asiatic pole, while Toronto may be regarded as on one side and Port Kennedy and Point Barrow as on the other side of the American pole (§ 29). The question as to what influence, if any, these poles have upon the disturbance-diurnal variation of declination is thus one which may be answered by examining the results obtained at these various stations. For this purpose, instead of recording the aggregate disturbances at the various hours, the result is expressed in ratios, —the mean hourly ratio for the day being taken as unity, or in other words the whole body of disturbances for the twenty-four hours being reckoned as twenty-four. The results of this method are graphically represented in fig. 37, where in the left-hand curves Kew time is used, and in the right-hand curves local time, each starting at 0 h .

55. A t all the various stations one curve exhibits unmistakably a single progression, while the other exhibits more or less distinctly a double progression. At Kew, Toronto, Port Kennedy, and Point Barrow it is the easterly disturbances which exhibit this single progression; while, on the other hand, at Peking and Nertchinsk, stations which are oppositely related to the Asiatic magnetic centre, it is the westerly disturbances which do so. It is imagined by Sabine and others that this peculiar reversal is due to the fact that Kew and its associated stations may be regarded as on one side and Peking and Nertchinsk as on the other side of the movable magnetic system.

Sabine has likewise remarked that the single-progression curves, whether denoting easterly or westerly disturbances, exhibit maxima which take place not far from the same absolute time. We have therefore plotted all the left-hand curves according to Kew time, that the eye may readily see the amount of simultaneity which their corresponding phases exhibit. It will be noticed that there is a very striking.simultaneity between the maxima of Kew, Toronto, Peking, and Nertchinsk, but that the maxima for Port Kennedy and Point Barrow, while both occurring about the same time, fall at a time decidedly if not very greatly different from that of the other maxima. Indeed the time of maximum for Port Kennedy and Point Barrow is not far from the time of minimum for the other stations. Now it has been noticed by Sabine that Port Kennedy and Point Barrow may be regarded as on one side of the American magnetic centre of intensity, while Toronto and the other associated stations are on the other side. It seems therefore possible to connect this last fact with the change in the time of maximum. Sabine has likewise remarked that the aggregate amount of disturbances is much greater at Point Barrow than at any other station. Now Point Barrow is likewise that spot where auroras are most frequent. Thus in the phenomena we are now discussing there is first of all a marked reference to the Asiatic pole; secondly, a reference not so marked, perhaps, to the American pole; and thirdly, a reference to the centre of auroral activity. Sabine, whose experience of such matters is very great, appears to think most of the reference of these phenomena to the Asiatic pole. He thinks that “of the two magnetic systems which are distinctly recognizable in the magnetism of the globe one has a terrestrial and the other a cosmical source,” and that it is “the latter of these two systems which, by its progressive translation, gives rise to the phenomena of secular change and to those magnetic cycles which owe their origin to the operation of the secular change,” concurring with the conclusion of Walker that “the magnetic influence at any point of the globe is the result of two distinct magnetic systems, the principal of which is the magnetism proper of the globe, having its (northern) point of greatest attraction in the north of the American continent, whilst the weaker system is that which results from the magnetism induced in the earth by cosmical action, and of which the northern point of greatest attraction is at present in the north of the Asiatic continent. Thus the direction of the magnet at any point results from the superposition of these two systems, the nearest pole being always predominant over the more remote” (Phil. Trans., 1868). While disposed to think that something of this nature should be accepted as a working hypothesis, we would, however, point out that the Asiatic pole cannot be regarded as accounting for all the phenomena of disturbances, but that the focus of disturbance is probably nearer the focus of auroras than it is to either of the foci of magnetic intensity.

The right-hand curves representing these disturbance-diurnal variations which have two maxima are, except for Port Kennedy and Point Barrow, decidedly irregular. Sabine remarks also that, instead of having a reference to absolute time like those with one progression, their reference is rather to local time. We have therefore plotted all these curves according to local time; nevertheless this reference does not come out with very great distinctness; but it must be remembered that our analysis of disturbances into easterly and westerly, although, in the hands of Sabine, it has given us much new information, has no claim to be regarded as final and complete.

56. Results in the Soτιthern Hemisphere.— Table VIII. shows the disturbance-diurnal variation of declination exhibited for St Helena, 15° 56'7 S. lat., 5° 40'∙5W. long.; Cape of Good Hope, 33° 56' S. lat., 18° 28'75 W. long.; Hobart Town, 42° 52''5 S. lat., 147° 27' 5 E. long.

At St Helena and the Cape the easterly disturbances present the appearance of a single progression, while the same remark slightly modified applies to the easterly disturbances at Hobart Town. Again the times of easterly maxima for St Helena and the Cape are very nearly simultaneous, while Hobart Town, which we may regard as situated on the opposite side of the chief southern magnetic centre from St Helena and the Cape, has its maximum nearly coincident in absolute time with the minimum of the other two stations. It would thus seem that the chief magnetic centre of the south is similar in its action as regards these phenomena to the chief magnetic centre of the north. Again the absolute time of single maximum for the south as determined by St Helena and the Cape is about twelve hours different from the corresponding time for the north as determined by Kew, Toronto, Peking, and Nertchinsk. All this is in favour of the working hypothesis already mentioned.

[table]

Finally, the westerly disturbances at the three southern stations bear greater marks of a double progression and of irregularity just as they aid in the northern hemisphere, and moreover like their northern analogues they are regulated by local rather than by absolute time.

57. Distribution of Declination Disturbance over the Various Months of the Year.— Broun was probably the first to remark in reducing the Makerstoun observations that the disturbances were greatest at the equinoxes and least at the solstices. His method was to find for each month the mean diurnal inequality, and then to consider the difference of each individual observation from the monthly mean for that hour as a disturbance, the summation of all such differences for the month denoting the monthly disturbance value. The following table embodies the results at various stations—those at Toronto, Hobart Town, and the Cape being given by Sabine, and that at Bombay by C. Chambers, who has pursued Sabine’s method of separating disturbances :— [table]

58. A careful inspection of this table, without attempting a more complete analysis, will, it is thought, lead to the following conclusions :—

(1) Although for any station the distribution of the easterly disturbances over the various hours of the day is generally different from that of the westerly, yet the same law of distribution over the various months of the year is followed by the easterly and by the westerly disturbances at any station—the law at one station being, however, different from that at another.

(2) In all stations there is first an annual inequality exhibiting a maximum generally a short time after the summer solstice with a corresponding minimum for the winter solstice, and secondly a semiannual inequality exhibiting a maximum generally a little after each equinox.

(3) The equinox maximum is very conspicuous at Toronto; but the summer maximum is most conspicuous at the other stations.

59. In § 38 it was observed that the observations selected as disturbed at any station may nevertheless be a mixture of what may be termed true disturbances and of the more prominent specimens of magnetic weather. The truth of this statement would appear to be borne out by the laws now given. In one of these we find that disturbances, at all stations, have a maximum about the time of the summer solstice and a corresponding minimum about the time of the winter solstice. But the absolute time of the summer solstice for stations north of the equator corresponds with that of the winter solstice for stations south of the line. It would therefore appear that in so far as this law is concerned such disturbances lack the element of simultaneity. On the other hand, a law of this nature would naturally hold for magnetic weather. For at any station the diurnal range of declination is greatest at the summer solstice, and hence any considerable proportional variation of this would, if represented by a fixed scale, present the appearance of being greatest likewise at this time. The question thus arises whether this law does not rather apply to magnetic weather than to real disturbance.

Again the semiannual inequality of disturbance exhibits throughout the globe a maximum at the equinoxes, and thus presents the element of simultaneity which was wanting in the annual. This law may therefore refer to true disturbance, and this view is supported by the fact that the aurora—which may be regarded as the universal accompaniment of great and simultaneous disturbances— obeys, as we shall afterwards see, in those stations where it has been well observed, this very same law, that is to say, it has likewise maxima at the equinoxes.

60. Distribτιtion of Declination Disturbances over Various Years. —In 1852 Sabine discovered (Phil. Trans., 1852, p. 103) that disturbances have a long-period inequality allied to that of sun-spots in such a way that a maximum and a minimum of disturbance coincide with a maximum and a minimum of sun-spot frequency.

This will be seen from the following table (X.), in which we have the relative values of declination disturbance at Toronto and Hobart Town compared with the number of groups of spots observed on the sun’s disk :— [table]

61. The following table (XI.) exhibits the same thing for Bombay. The first column of this table is derived from the magnetic results of C. Chambers, while the sun-spot areas are those of Messrs De la Rue, Stewart, and Loewy. [table]

We may conclude from these tables that declination disturbances march with sun-spots, but that the alliance between these two phenomena is probably not so intimate as that between declination ranges and sun-spots.

62. Distribution of Declination Disturbances over the Surface of the Globe.— It is well known that disturbances are comparatively small near the equator, while they are great near the magnetic poles, and greatest of all perhaps near the position of maximum auroras. If we adopt Sabine’s system of separating disturbed from undisturbed observations, it is thus clear that the same separating value cannot be adopted at all stations. At first sight this would seem to introduce an element of uncertainty in the estimation of disturbances, but it was soon found by Sabine that no very great nicety is required in this matter. Not only do the laws which regulate disturbances at a given station remain comparatively unaffected by the magnitude of the separating value, but it is likewise easy to tell whether the aggregate disturbance value at one station is decidedly greater or less than at another. Probably at present it would be impossible to obtain more definite information than this.

63. The following table (XII.) exhibits the proportion between the aggregate amount of easterly and that of westerly disturbances of the declination at various stations in both hemispheres :— [table]

64. Annual Variation of Declination.— The declination fluctuations of short period hitherto discussed are not necessarily accompanied by a permanent change of mean position of the needle. We have now to inquire whether there be any fluctuations of long period (besides the secular change discussed in §§ 30-33) tending to alter perceptibly the position of the magnetic needle. This leads us at once to the annual variation, for our knowledge of which we must look to the later-made and more accurate observations, in which all possible sources of error have been carefully eliminated. Broun has made an exhaustive experimental inquiry into the various sources of error which could possibly influence his declination needle at Trevandrum. His conclusion was that the variations of torsion of a well-made thread are not sufficient to produce a sensible effect upon the position of a powerful magnet. In fact Grubb’s magnet, weighing 6000 grains, and Adie’s, weighing 1100 grains, give almost identical results. We may extend these conclusions to other observatories where well-devised instruments have been established, and look with much confidence to such instruments registering correctly the secular as well as the annual change of declination that may be taking place at each locality.

65. The following table (XIII. ), borrowed, with the exception of the Trevandrum and Bombay results, from E. Walker’s Terrestrial Magnetism, shows the annual variation at seven stations :— [table] [table]

Here + indicates that the marked pole of the needle is to the west and - that it is to the east of its mean position for the year.

66. To cancel the irregularities of this table let us take the means from April to September and from October to March, the former embracing the months around the June solstice and the latter those around the December solstice (Table XV.):— [table]

It will be seen from the above that the means for Trevandrum and Bombay present opposite signs to those for the other stations. The whole amount for Trevandrum is no doubt very small, and Chambers does not regard the evidence for Bombay as conclusive; but on the whole it would appear that two observatories near one another present evidence of a similar behaviour in declination, and we are therefore disposed to regard it as a reality.

67. Semiannual Variation of Declination.— If we look at the numbers of Table XIV., we shall see that there are traces of turning points at the equinoxes. Let us, in order to exhibit this, compare together the sums for the six months grouped around the two equinoxes with those for the six months grouped around the two solstices—that is to say, compare the sums for February, March, April, August, September, October, with those for November, December, January, May, June, July—and we thus obtain the following table (XVI.): — [table]

68. Solar-Diurnal Variations of the Horizontal and Vertical Components of Magnetic Force.— Although self-recording magnetographs have been established in many observatories throughout the globe, yet, owing to the peculiar difficulties of the task, and the labour of the process of reduction, very little has been done towards determining the solar-diurnal variation of thehorizontal and vertical components of the earth’s magnetic force. Senhor Capello of the Lisbon Observatory has, however, made progress with nis reductions, and has already published valuable information regarding the solar-diurnal fluctuation of the two force elements at his observatory.

In his attempts to eliminate the disturbances of horizontal and vertical force by the method of Sir E. Sabine, Senhor Capello has experienced considerable difficulty, more particularly with the records of the vertical force magnetograph. This instrument and the bifilar have very often been found by him to change their position of equilibrium after strong perturbations. Again there is generally, for any hour, a variation at the beginning and end of the month from the monthly normal value for that hour owing to change of temperature, and this cannot be completely corrected inasmuch as the coefficient of temperature is not exactly known. These two causes combined tend to falsify the results when the plan adopted is the method of comparison between the individual values of any hour and the normal monthly average of that hour. Senhor Capello has found it necessary to select and extract the disturbances, not directly from the hourly values, but by comparing the variation of an individual day with the average diurnal variation derived from the month.

To illustrate this method by means of an example, let us imagine that the sum of the twenty-four hourly values for a particular day is 24,000, and that the average monthly diurnal variation would indicate that a particular hour of this day should have a value 990, then, if the value for this hour should prove to be greater or less than 990 by more than a certain amount, it would be set aside as a disturbed observation. Senhor Capello rather thinks it will be desirable somewhat to modify this method, and he concludes his remarks by observing that for this and other similar questions it is most necessary that directors of establishments possessing magnetographs should agree together to employ the same method in their reductions in order that their results may be comparable with each other. With the view of adding weight to these remarks, we may quote the observation of Sir William Thomson, that our ability to analyse mathematically that influence which produces the diurnal variation will depend upon our knowing at a certain number of stations the exact nature of this diurnal variation for each of the three magnetic elements. A complete theory of this diurnal influence must therefore wait upon the concerted action of the directors of the various establishments possessing magnetographs.

69. Change in Horizontal Force Range from Month to Month.— Although we do not possess finally accurate determinations of the solar-diurnal variations of either element of the force, yet we are in possession of information regarding the change in the diurnal range of the horizontal force from month to month at the Greenwich Observatory. William Ellis has given us the following table {Phil. Trans., 1880) representing the monthly mean diurnal range of horizontal force at that observatory expressed in ten-thousandths of the whole horizontal force. In the formation of these means, days of great magnetic disturbance were rejected, and also certain other days on which there prevailed a smaller but considerable amount of disturbance estimated according to a general standard formed in the examination of many thousands of photographs. [table]

Thus, like the declination range (§ 43), the horizontal force range has a maximum in summer and a minimum in winter, and exhibits a tendency towards maxima at the equinoxes.

70. Long-Period Inequalities of Horizontal Force Range.— Lagging Behind. — Ellis has compared the diurnal range of the horizontal force as well as that of the declination at Greenwich with the period of sun-spot frequency, his comparisons extending from 1841 to 1877, and he has deduced the following conclusions:—

(1) The diurnal ranges of the magnetic elements of declination and horizontal force are subject to a periodical variation, the duration of which is equal to that of the known eleven-year sunspot period.

(2) The epochs of minimum and maximum of magnetic and sunspot effect are nearly coincident, the magnetic epochs on the whole occurring somewhat later than the corresponding sun-spot epochs. The variations of duration in different periods appear to be similar for both phenomena.

(3) The occasional more sudden outbursts of magnetic and sunspot energy, extending sometimes over periods of several months, appear to occur nearly simultaneously, and progress collaterally.

71. Disturbance-Diurnal Variation of Force Components.— We may derive the following conclusions from the results obtained by Sabine for the observatories of Toronto, Kew, and St Helena. For each element there are two categories, namely, those disturbances which tend to increase and those which tend to diminish the element in question.

(1) At Toronto the disturbances increasing both elements of force well represent single progressions with maxima occurring for both about 4 or 5 hours local time. Again the disturbances decreasing both elements represent fairly well single progressions with maxima occurring for both at about 14 or 15 hours local time.

(2) At Kew the disturbances increasing both elements represent well single progressions with maxima occurring for both about 5 hours local time. On the other hand, the disturbances decreasing the horizontal force represent signs of a double progression and those decreasing the vertical force signs of a single progression, the maximum for the latter falling between the two maxima for the former, and occurring at 14 hours local time.

(3) There is not the same close correspondence between the progress of the disturbances which tend to increase both elements nor between the progress of those which tend to decrease both elements at St Helena as there is for the other stations, nor is there the same likeness between the numbers for St Helena and those of Toronto or Kew as there is between the numbers of Toronto and those of Kew.

72. The fact that the disturbance-diurnal variations of the two force elements at Kew are very like each other while neither of them is very like the corresponding declination variation (§ 54) receives confirmation from a visual inspection of the Kew curves. In the Philosophical Transactions for 1862 Stewart thus describes the result of an inspection of the disturbances of these curves for the years 1858, 1859, 1860 (disturbance years):—

“There are twenty-two cases in which the declination is raised or lowered along with the horizontal force, and only seven cases of an opposite description. Also there are twenty-two cases in which the declination is raised or lowered along with the vertical force, and only eleven cases of an opposite descι iption. Finally, there are thirty-one cases in which both forces are raised or lowered together, and only two cases of an opposite description. There is therefore a decided tendency in the curves of all the elements to be raised or lowered simultaneously, but this tendency is stronger between the horizontal and vertical force curves than between either of these and the declination. It may at the same time be affirmed that with the exception of the disturbance of August to September 1859 there is no very prominent case in which the three element·, do not rise or fall together."

73. Peaks and Hollows. — These are certain small but abrupt magnetic changes which from the fact that they generally fall within the separating value are not usually regarded as disturbances. These changes can only be brought to light where there is a continuous record of magnetic phenomena such as that derived from self-recording magnetographs. They were first studied at the Kew Observatory by Stewart {Phil. Trans., 1862). We have seen that more than one type of force must be concerned in producing magnetic disturbances. This is confirmed by the appearance of the Kew records, from which it may be seen that no disturbance of any magnitude is due to the action of a single force varying merely in amount but not in direction. For if there were only one type of force the distance at any moment of a point in the curve of one of the elements from its normal position should bear throughout a disturbance an invariable proportion to the distance of a corresponding point in the curve of another of the elements from its normal; but this is by no means the case.

But even if several independent forces are at work it may be thought unlikely that at the same moment a sudden change should take place in all; there is thus a probability that sudden changes of force, as exhibited in peaks and hollows, are changes in one of the elementary forces concerned. Even if the change is not a very abrupt one, provided that we confine ourselves to such peaks and hollows as present a similar appearance for all the curves, we may suppose that we are observing changes in one only of the elementary disturbing forces; for it is unlikely that two or more independent forces, changing independently, should produce similar appearances in all of the three curves.

Assuming it as probable that similarity of appearance in the curve variations of the three elements denotes a simplicity in the disturbing force, Stewart has discussed all such peaks and hollows at Kew extending over the first two years of their production, and has obtained a result which is embodied in the following table:—

Table XVIII.— Hourly Ratios and Frequency of the Kew Peaks and Hollows, the Vertical Force Disturbance being taken as Unity . 1 [table]

74. It will be seen from this table that the ratio between simultaneous peaks and hollows of the two components of the force is very nearly constant, the horizontal force disturbance being about double that of the vertical force, so far as size on the curve is concerned. It will also be seen that there is a very marked diurnal range in the ratio which the declination peak or hollow bears to that of the vertical force, this ratio being greatest about 7 a. Μ. About this hour we have also most peaks and hollows, while in the evening and early morning hours there is so great an absence of these phenomena that the ratios are doubtful.

75. A preliminary comparison between the peaks and hollows at Lisbon and at Kew has been made by Capello and Stewart {Proc. Roy. Soc., January 28, 1864) with the following conclusions.

(1) The Kew peaks and hollows are simultaneously produced at Lisbon in all the elements, but to a smaller extent than at Kew.

(2) The direction is the same at both stations for the declination and horizontal force peaks and hollows, but it is reversed in the case of the vertical force, so that a sudden small increase of vertical force at Kew corresponds to a diminution of the same at Lisbon.

It would be manifestly impossible to discuss with any advantage the nature and origin of these peculiar changes until more extensive observations of them have been made. As the peak and hollow force is probably of a simple nature, a further knowledge of its character may be of much importance to the theory of terrestrial magnetism.

It is interesting to remark that we have in peaks and hollows the same close relation between the variations of the two force elements that we find in the larger disturbances.

It is believed too that during violent disturbances a certain change of type is produced in the peak and hollow force, and more especially is this remarkable in the great disturbance in August and September 1859, where the declination would seem to march in the opposite direction from the two components of the force. We have seen that the same peculiarity characterized on this occasion the larger and more apparent magnetic changes. We shall afterwards refer to a circumstance which may perhaps throw light upon this peculiarity (§ 93), meanwhile we conclude by again remarking that during comparative magnetic calms the peak and hollow force shows signs of remaining constant in type, and that it is therefore of great importance that the directors of observatories possessing self-recording magnetographs should take united action to observe this force.

76. Other Inequalities of the Disturbance-Diurnal Variation of the Force Components.— Sabine has shown that disturbances of the force components present a distribution over the various months of the year very similar on the whole to that which is exhibited by disturbances of declination. He has likewise shown that disturbances of the force components present a distribution over various years similar to that exhibited by disturbances of declination. Finally, we may probably conclude that disturbances of the force components are smallest at those portions of the earth’s surface where disturbances of the declination are smallest, and largest at those portions where such disturbances are largest.

77. Annual and Semiannual Variation of Horizontal Force and Dip.— Broun {Trans. Roy. Soc. Edin. for 1861) has discussed the results obtained by Sabine at his magnetical stations, and has shown that differential and absolute observations agree in telling us that the horizontal force is smallest at the equinoxes and greatest at the solstices. Whipple has recently obtained the same result from the Kew observations.

We have deduced the following table (XIX.) from the various absolute determinations that have been made at sundry places. In it the annual and semiannual variations of declination, horizontal force, and dip are exhibited, ‘ ‘ increase” denoting a push to the west, and “decrease” a push to the east. The method of obtaining these has already been indicated in §§ 66, 67.

[table]

78. In discussing the results of this table we shall assume that the sun acts, and in all probability acts indirectly, upon the magnetic system of the earth. This point will afterwards be further examined. Meanwhile, assuming this indirect action of the sun, and assuming, to fix our thoughts, that it is in close alliance with the convection system of the earth's atmosphere, we can readily imagine that such solar action would act most strongly on the earth’s magnetic poles at the solstices, and that in the June solstice the pole or poles in the northern hemisphere and in the December solstice those in the southern hemisphere would be most affected. Now a strong action of this kind upon either magnetic pole may well be presumed to increase the general magnetism of the earth, or at least that portion of it which is most readily affected by external action, that is to say, the induction system. Again, if the solar magnetic influence is connected with the convection currents of the earth, we can readily imagine that the influence in the northern hemisphere where there is much land should exceed that in the southern hemisphere where there is much water.

If these views be reasonable we might expect two things to follow:—(1) the earth’s induction system should be stronger at the solstices than at the equinoxes, and (2) it should be more especially strong at the June solstice, when the sun acts in the northern hemisphere. We must bear in mind, however, that so vast is the earth that a stimulus applied to its particles most susceptible of magnetism may not be instantaneously propagated throughout its mass, but that time may enter as an element of the question, in which case, inasmuch as the action of the sun at the June solstice is in the northern hemisphere, a station near the south pole may not fully partake of the magnetic effects of this action.

79. An hypothesis of this nature would appear to be consistent with the results of Table XIX.

In the first place, if the earth should become stronger as a magnet in one or in both of its magnetic systems this would show itself by an increase of horizontal force at least in all such stations as those at which absolute observations are made. An influence which increases the horizontal force at these various stations is therefore naturally regarded, and was regarded by Broun, as one increasing the strength of one or both of the magnetic systems of the earth— whether of one or of both will presently appear. We may therefore assume from our observations that one or both of the earth’s magnetic systems are strongest at the solstices.

In the next place we may imagine that the changes of declination and dip which the table exhibits as occurring at the solstices are the very changes which would be wrought in these elements by an increase of power in the earth. For we see very well that an increase of horizontal force at the various stations may be regarded as denoting an increase of the earth’s magnetic power. We cannot, however, see with equal facility what changes would be produced in the declination and dip by an increase in power of one or both of the magnetic systems; but we may well imagine that such changes of these elements as are found to accompany an increase of horizontal force are those that denote an increase of the earth’s power.

We have thus ascertained the probable nature of those changes of the three elements which denote an increase of power. Now it will be noticed from the table that the effect at the June as compared with that at the December solstice is of the opposite nature to the effect at the equinoxes as compared with the solstices,—that is to say, the earth is more powerfully affected in June than in December, the only well-established exception to this being Hobart Town in the far south. But, assuming that time is an element in the development of this preponderating influence acting in the north, it is easy to see why Hobart Town should not exhibit its full effect.

It remains to determine from the observations themselves which of the magnetic systems it is that exhibits these oscillations. Analogy would of course point to the induction system, but it is desirable to determine this from the observations themselves.

In § 54, when discussing the disturbance-diurnal variation of declination, it was found that Toronto and Kew may be regarded as on one side of the Siberian pole, while Peking, Nertchinsk, and Bombay are on the other. Now, if it be this pole that is influenced by the oscillations under discussion, we might expect that the influence on declination at Toronto and Kew should be the opposite of that at Trevandrum and Bombay. We find by the table that this is the case, and we are thus inclined to attribute these changes to the Siberian instead of the American pole. It would thus appear that the observations of Table XIX. bear out the provisional working hypothesis which we have ventured to introduce. It is quite possible that these remarks may not stand the test of more complete inquiry, but they are here introduced rather as denoting a method of looking at the subject which ought we think to be pursued than as embodying conclusions of a final nature.

80. Effect of the State of the Sim's Surface upon the Absolute Magnetism of the Earth. — We have now to consider whether the state of the sun’s surface permanently influences the magnetism of the earth. It will at once be seen that any such action will apparently manifest itself as an oscillation in the secular change. We must, however, carefully guard ourselves against prematurely concluding that it implies a variation in the amount of true secular change. There may be two distinct things—true secular change due to one cause, and action depending on sun-spots due to another. These, from the nature of the case, are necessarily mixed up together in the yearly changes which we examine; it does not, however, follow that there is any real identity between them. We shall now give one example of the method to be pursued in the attempt to detect a solar influence of this nature. Let us turn to Table 111., and take the declination yearly values at Toronto from 1856 to 1871. Subtracting the value for 1856 from that for 1871, we find that the westerly declination had increased in fifteen years 5Γ - 6, that is to say, at the mean rate of 3' - 44 per annum. Again, the average declination for the sixteen years 1856-71 is 2° 20' ' 8 corresponding to the epoch at the commencement of the year 1864. Taking the average value and epoch, and also the average yearly increase above given, we are able to construct the following table (XX. ), in which calculated and observed values at Toronto are compared together :— [table]

It may be gathered from this table that the years which correspond to minimum sun-spots have in the last column a greater negative or lower positive sign than those which correspond to maximum sun-spots, and hence we may conclude that at Toronto the tendency of many sun-spots is to increase the westerly declination.

81. Performing a similar operation for all those cases in which we have a sufficiently extensive series of observations to work upon, we obtain the following table :—

Table XXL— Effect of Numerous Sun-Spots σ ∣ ι the Values of Magnetic Elements. [table]

82. We have good grounds for supposing that the sun is most powerful when there are numerous spots on his surface, and therefore the above table represents a state of things which we may imagine to be caused in one way or another by increased solar power. Now the most natural hypothesis is to imagine that an increase of spots acts in producing an increase of disturbances, and that for those stations at which the disturbances tend on the whole to affect the elements in a definite direction there will be left behind a permanent effect in this direction. A comparison of Table XXL with Table XII. will, however, show that this explanation is not valid. For instance, at Toronto and Kew disturbances tend rather to diminish than to increase the westerly declination, while the effect of numerous sun-spots is to increase it. Again, at the Cape the tendency of disturbances is to increase the westerly declination, while that of numerous sun-spots is to decrease it. At Trevandrum again (if we judge of it by Bombay) the effect of disturbances will be to increase the easterly declination, while that of sun-spots is to decrease it. Again, it is believed that at Kew and Toronto the supposed disturbance effect on the dip agrees in character with the sun-spot effect. On the whole, therefore, there is no definite relation between the two effects.

Now if we take Hobart Town, the Cape, and Trevandrum in the above table, we find from Table XIX. that these stations seem to indicate that the magnetic state of the earth is most powerful at times of maximum sun-spots. Kew and Toronto, however, so far as declination and dip are concerned, appear to go the other way. If, however, we suppose that during the several years of maximum sun-spots the American pole as well as the Siberian is affected, and that on such occasions of long continuance the former has more influence than the latter, we shall be able to reconcile our results with the hypothesis of increased solar action. We can understand too that time must be an important element in any influence communicated to the American pole, and that, although such influence might be apparent at Toronto and Kew, which are comparatively near the pole, it would not be apparent at the other stations of Table XXL We shall recur to this subject λvhen discussing secular change.

Various Phenomena connected with the Sun and with Terrestrial Magnetism.

83. Closeness in Time between Solar Changes and Magnetic Disturbances.— Loomis (American Journal of Science, vol. 1.) has registered the extent of sun-spots for the six days preceding and following each of the great magnetic disturbances at Greenwich, and has compared these values with that for the very day of the disturbance. In this manner he has treated all the days of great magnetic disturbance at Greenwich for a period of twenty-three years, with the exception of those cases in which very few observations of sun-spots were made. The cases of disturbance thus treated amount to one hundred and thirty-five, and the following result has been obtained:— [table]

From this result Loomis draws the following conclusions:— (1) great disturbances of the earth’s magnetism are accompanied by unusual disturbances of the sun’s surface on the very day of the magnetic storm; (2) the great disturbance of the sun’s surface which accompanies a terrestrial magnetic storm is generally heralded by a smaller disturbance three or four days previous, succeeded by a comparative calm which immediately precedes the magnetic storm.

84. There is one instance on record of a sudden solar change which was practically simultaneous with a magnetic disturbance. On September 1, 1859, a little before noon, R. C. Carrington was observing by means of a telescope a large sun-spot, when, to quote his own words—

“Within the area of the great north group (the size of which had previously excited general remark) two patches of intensely bright and white light broke out. . . . I noted down the time by the chronometer, and, seeing the outburst to be very rapidly on the increase, and being somewhat flurried by the surprise, I hastily ran to call some one to witness the exhibition with me, and on returning within sixty seconds was mortified to find that it was already much changed and enfeebled. Very shortly afterwards the last trace was gone; and, although I maintained a strict watch for nearly an hour, no recurrence took place. . . . The instant of the first outburst was not fifteen seconds different from ll h 18 m Greenwich mean time, and ll h 23 m was taken for the time of disappearance. In this lapse of five minutes, the two patches of light traversed a space of about 35,000 miles. . . . It was impossible, on first witnessing an appearance so similar to a sudden conflagration, not to expect a considerable result in the way of alteration of the details of the group in which it occurred; and I was certainly surprised, on referring to the sketch which I had carefully and satisfactorily finished before the occurrence, at finding myself unable to recognize any change whatever as having taken place. The impression left upon me is that the phenomenon took place at an elevation considerably above the general surface of the sun, and accordingly altogether above and over the great group in which it was seen projected.

“It has been very gratifying to me to learn that Mr Hodgson chanced to be observing the sun at his house at Holloway on the same day, and to hear that he was a witness of what he also considered a very remarkable phenomenon.”

At the very moment when Carrington observed this phenomenon the three magnetic elements at Kew were simultaneously disturbed. This disturbance occurred as nearly as possible at Uh 15m a.μ., affecting all the elements simultaneously, and commencing quite abruptly. The first or most abrupt portion of the disturbance lasted only about three minutes for all the elements; but after that there was a more gradual change in the same direction before the curve turned. This more gradual continuation of the first sudden movement lasted about seven minutes for all the elements. This magnetic disturbance was, however, in reality a small one, and was followed by a very great disturbance which took place not many hours afterwards.

85. Simultaneity of Changes of Horizontal Force at Various Parts of the Earth.— We have already (§ 79) alluded to the superiority of the horizontal force in indicating by its changes what is taking place in the magnetic system of the earth. If this system be strengthened as a whole we shall no doubt find the horizontal force increased in value at our various stations, while if the earth’s power be weakened as a whole we shall find the horizontal force diminished.

Broun has discussed this subject at’ great length in a memoir already mentioned, and has embodied his observations in numerical results from which the following table has been extracted :—

Table XXIII.— Daily Means of Horizontal Force at Makerstoun, (Μ.), Trevandrum (T.), Singapore (S.), and Hobart Toum (H.). [table]

This table shows a considerable likeness between the daily changes of the horizontal force at the four stations. For instance, we have a minimum which occurs on March 5 at Makerstoun and March 6 at the other stations; we have likewise a well-defined maximum occurring at all stations on March 14, and another occurring at Trevandrum on March 25, and at the other stations on March 26. Finally we have a well-defined minimum occurring at Trevandrum on March 31, and at the other stations a day earlier.

Broun has extended a similar treatment to daily means for every hour, and fig. 38 conveys a good idea of the amount of simultaneity which obtains in the changes of such values of horizontal force at stations far apart.

86. Recurrence of Disturbances at Intervals of about Twenty-six Days.— Broun and likewise Hornstein have observed that there is a tendency in large magnetic changes to recur at intervals of about twenty-six days. At first it was natural to suppose that we have here a magnetical indication of the true time of the sun’s synodical rotation, the interval between two disturbances denoting that which elapses between two presentations to the earth of a peculiarly powerful solar meridian. It seems unlikely, however, that there is a really permanent one-sidedness of this kind in our luminary; but the result of observation seems to show that for a limited period, say two or three years, certain meridians of the sun appear to be peculiarly powerful. The cause of this we shall not here discuss, but simply treat the phenomenon as a fact derived from observation. Broun in his paper above quoted (Phil. Trans., 1876) makes the following remarks :—

“We have seen that when one side of the sun is presented to the earth the magnetic force of the latter is greater titan when the other side is turned towards us; we may even say that the intensity is greatest for a given solar meridian; this, however, may be simply an integral effect resulting from the actions due to all the meridians. But can we suppose when a great and sudden increase or diminution of the earth's magnetic force occurs that this is produced by some change occurring on a particular solar meridian? This does not seem at all improbable.

“In order to examine the facts, all the cases were noted during the years 1844 and 1845 in which the daily mean horizontal force diminished one-thousandth of its whole value within an interval of three days; they were found to be twentyeight in number. If wc call the solar meridian presented to us on the 1st January 1844 tlιe zero meridian (0), and consider the time of rotation to be twenty-six days, and that there are twenty-six meridians, we find that the solar meridians presented to us when these great movements occurred may be arranged in a ftw groups, as in the following table” [Table XXIV., p. 177].

“An examination," continues Broun, “of this table will show that nearly half of the great changes began when the eighth meridian after the zero had passed, while five began near the twelfth after, and five near the zero itself. . . . If any doubt existed as to the possibility of these being mere accidental coincidences, it would be removed. I think, by a consideration of the marked succession occurring between July 31 (No. 18) and December 11, 1845 (No. 28). . . . If we neglect the two cases of July 31 and August 26, which commence at ⅛6 and 4-5 respectively, we have five cases of successive solar rotations in which the diminutions of intensity began on the +8 day. This exact recurrence at the end of twenty-six days of a marked diminution of force proves, it seems to me, that the actions are all due to the sun, whose time of rotation must be nearly twenty-six days.

“An examinalion will show that the sudden diminutions of terrestrial magnetic force are in nearly every case preceded by a sudden increase.”

In the above extract we have given the author’s exact words, but, while thinking with him that these actions are due to the sun, it does not appear to us to follow that the time of the sun s rotation must be nearly twenty-six days. This assumes that the meridian of peculiar power is fixed on the solar surface. It does not, however, seem impossible to imagine that such a meridian may have a proper motion of its own, and indeed the planetary hypothesis of the origin of sun-spots would rather lead to this conclusion. But if this be the case we shall be unable to deduce from recurrent magnetic disturbances the true value of the period of solar rotation.

87. Repetitions of Magnetic Changes.— J. B. Capello, director of the Lisbon observatory (Proc. Roy. Soc., October 1868), has remarked that at periods of disturbance there are nearly synchronous movements of the declination needle during corresponding hours for two, three, or more days. He thus describes these phenomena:—

“In some cases the repetition is only in two or three parallel movements; in others there are true periods of repetition of some hours in duration. The repeated periods are not entirely similar, their phases being so modified that in some cases their identity can only be recognized by a very minute investigation. The same periods, when repeated, have not always the same total duration; nor do they recommence at the same precise hour, but sometimes earlier and sometimes later, the differences varying from a few minutes to two or three hours. We also see that the greatest number of repetitions belong to the night hours, that is to say, those hours when the movements of the needle are easterly. In the morning hours there do not appear to be any well-marked repetitions. There are twenty-four examples now given, fifteen of which show repetition on two days, eight on three days, and only one where the curve appears repeated fur four days. It appears that all the facts exhibited in these examples agree with the cosmical theory; the cause (existing in the sun or in space) appears to continue sometimesdιιringtwo, three, or several days without undergoing remarkable transformations. The repetition, being sometimes earlier sometimes later, seems also to indicate that the cause possesses a proper movement; the cause persists, but only comes again into operation when the earth by its diurnal rotation is placed in a similar position or conjunction to that of the preceding days.”

Stewart, having compared Capello’s curves with the corresponding traces of the declination at Kew, found that the Lisbon disturbances are almost invariably reproduced at Kew at the same time, only to a greater extent, and also that the same amount of similarity which the various Lisbon curves exhibit is exhibited in the corresponding Kew curves. The strongest point in favour of the hypothesis is, he thinks, “not so much the repetition of a single disturbance as the repetition of a complicated disturbance in most if not all of its sinuosities.” Several examples of this occur in the diagrams. It would seem that something of the above nature was suspected by Humboldt, the earliest investigator of disturbances. Humboldt was astonished to discover the frequency with which nocturnal perturbations occurred, sometimes recurring at the same hour on several successive Table XXIV.— Cases in which the Earth's Magnetic Force diminished One-thousandth of its Value or more in 1844-45 (§ 86). [table]

nights (Walker’s Magnetism, p. 80). We would make two suggestions before dismissing this subject.

(1) If we imagine that these changes are caused by the solar influence acting vertically on some susceptible region of the earth, then, inasmuch as they occur at the evening or early night hours, this region must lie considerably to the west.

(2) The region must also have a proper motion of its own (see Capello’s remark). Is it possible that this proper motion is on the whole from west to east,—a motion which we know is pursued by meteorological weather, and in which it is imagined (§ 52) that magnetical weather as defined by us likewise participates?

88. Comparison of Declination Changes at Stations near each other.— Messrs Sidgreaves and Stewart (Pro. Roy. Soc., October 1868) have compared together certain curves of the Kew and Stonyhurst declination magnetographs. These magnetographs are of the same pattern, and it was found that on ordinary occasions the declination traces at both stations were precisely alike. This was confirmed by placing the curves the one over the other, when they were found to coincide even in their most minute features. In times of disturbance, however, it was found that the motions exhibited by the Stonyhurst curves were greater than those at Kew, and this excess of Stonyhurst over Kew depended not so much on the absolute size of the disturbance as on its abruptness.

This feature of the comparison is exhibited in the following table (XXV. ), in which the excess of Stonyhurst over Kew in scale divisions is compared with the abruptness of the disturbance, this element being measured by the changes occurring in unit of time:— [table]

It is very desirable that further comparisons of this nature should be made.

89. Auroral Displays.— These are very frequent if not continuous near the magnetic poles, while in middle latitudes they are the invariable accompaniments of all considerable magnetic storms. N ear the equator they hardly ever occur.

There is a considerable variety in the forms assumed by these displays, and it is possible that this may denote a corresponding variety in the cause or causes which give rise to this phenomenon.

Loomis (Smithsonian Report for 1865) specifies five such varieties: (1) a horizontal light like the morning aurora or break of day; (2) an arch of light which frequently extends entirely across the heavens from east to west and cuts the magnetic meridian nearly at right angles,—in the polar regions five such arches have been seen at once; (3) slender luminous beams or columns well-defined and often of a bright light; (4) the corona, the centre of which is invariably near the magnetic zenith, but not always exactly coincident with it; and (5) waves or flashes of light.

90. Auroras exhibit the same annual variation as magnetic disturbances, and are most frequent about the equinoxes—a fact first observed by Maizan. Kaemtz in his Meteorology gives the following table, which is applicable to European auroras. [table]

Loomis again in the memoir already quoted gives the distribution of American auroras over the various months derived from one hundred and thirteen years’ observations at New Haven and Boston, twenty-five years’ observations at New York, and two years’ observations in Canada. His results are represented in the following table :—

Table XXVII.—Monthly Frequency of American Auroras. [table]

It appears from this table that American like European auroras exhibit a maximum of frequency about the equinoxes.

91. Since auroras and magnetic disturbances go together, it is natural to imagine that we should have great auroral displays in years of maximum sun-spots. This is found to be the case, and in the following table (XXVIII.) Wolfs proportional numbers denoting sun-spot frequency are compared with the number of auroras witnessed in Europe and America as compiled by Loomis. It will be seen from this table that years of maximum auroras coincide very well with years of maximum sun-spots. [table]

In fig. 39 a graphical representation is given of the likeness which subsists between the progress of auroral frequency, spot frequency, and declination ranges.

92. While the results now given leave little doubt as to the fact of a connexion of some sort subsisting between sun-spots on the one hand and magnetic disturbances and auroras on the other, yet it is desirable to obtain evidence as to the closeness of the connexion between auroras and sunspots similar to that which was exhibited in § 82, and which showed the close connexion in point of time between sun-spots and disturbances. Loomis has with this view treated auroras in precisely the same way in which he treated disturbances, and has obtained the following table':— [table]

From which he concludes that “auroral observations in the middle latitudes of America are generally accompanied by a maximum disturbance of the sun’s surface on the very day of the aurora.”

93. Earth Currents.— These are electrical currents which take· place in the moist crust of the earth, and were first detected by W. H. Barlow {Phil. Trans., 1849). At a later period they were systematically observed and studied by C. V. Walker {Phil. Trans., 1862). They are now continuously recorded by photography at the Royal Observatory, Greenwich. Earth currents are particularly strong during magnetic disturbances. Sir George Airy has· graphically compared together certain magnetic disturbances as recorded by the Greenwich self-recording magnetographs and the simultaneous earth currents recorded by appropriate galvanometers {Phil. Trans., 1868), and finds it almost impossible to avoid the conclusion that the magnetic disturbances are produced by terrestrial galvanic currents below the magnets. The likeness between the two-systems of graphical representations is unquestionably very striking. But, while there is no doubt an intimate connexion between earth currents and magnetic disturbances, there is one circumstance which should make us pause before assigning the former as the complete and efficient cause of the latter. It is thus indicated by Lloyd: —

“When we examine the curves in which Mr Barlow has represented the course of the galvanonιetric deflexions caused by the earth currents, we observe that the regularity of that course is continually interrupted by rapid reciprocating movements in which the needle oscillates from one side to the other of the zero alternately. These movements are similar to those of the magnetometers with which we are familiar; but they are much more rapid, and bear a larger proportion to the regular changes. . . . I have selected for calculation the observations made during the six hours commencing at 3 a.m. on May 29, 1848, that being a period of comparative disturbance. The sum of the changes of the galvanometer needle during that period, on the Derby and Rugby line, was> equivalent to 571 divisions of the instrument—the mean daily range for the entire week being 11'4 divisions and the ratio=50. . . . The sum of the changes of the Greenwich declinometer during the same period was only 57 minutes, the mean daily range being 12’4 minutes. In like manner the sum of the changes of the horizontal force was - 0158 and the mean daily range '0034. The ratio is accordingly the same for the two magnetic elements, and its amount is 4’6, or less than one-tenth of the corresponding ratio in the case of the galvano-metrical changes. We learn therefore that the rapid changes of the earth currents are much greater in proportion to the regular daily changes than the-corresponding movements of the magnetometers."

We shall return to this subject in a subsequent part of this article.

94. Inequalities in Terrestrial Magnetism caused by the Moon.— Kreil in 1841 was the first to point out that the moon has a small influence on the position of the declination needle, and shortly afterwards the same fact was independently discovered by John Allan Broun. The more recent observations of Sabine and of Broun, but especially those of the latter, have thrown much light upon the nature of this action. As the lunar influence is not generally large, it is necessary to free the observations from the results of other inequalities, and this has beendone by the two observera above mentioned. The results given in Table XXX. have been obtained by Sabine (see Walker’s Magnetism}.

95. Thus (1) the mean effect of the moon upon the declination needle is to cause in each lunar day a double oscillation, and Sabine has shown that the lunar influence upon the other magnetic elements is of a similar type. (2) The turning points for both hemispheres are in all cases not far removed from the lunar hours

[table]

In this table 4- indicates that the north end of the magnet is to the east and — that it is to the west of its mean position.

0, 6, 12, 18. (3) In the northern stations we have a maximum westerly deflexion about the hours 0 and 12 and a maximum easterly deflexion about the hours 6 and 18, while in the southern stations the action is the reverse of this. (4) The oscillations would appear to be most decided at those stations, such as Toronto and Hobart Town, that are far removed from the equator.

96. Annual Variation of Lunar Effect.— Broun has recently studied with much success the peculiarities of the lunar influence at Trevandrum, and has obtained some very unexpected results. His first result was that the nature of the lunar influence upon the declination needle at Trevandrum depends upon the time of the year, and that the southern type of lunar action predominates at Trevandrum during the winter and the northern type during the summer months. If we take the mean of the whole year, then probably the southern type will be found to predominate.

97. Mean Lunar-Diurnal Variation during the Day and during the Night.— Broun has shown that the action of the moon on the declination needle at Trevandrum is greater in every month of the year during the day than during the night. The following table (XXXI.) gives the day and night ranges for the various months and their ratios :— [table]

It would appear from Broun’s observations that there is a difference of type as well as a difference of range.

98. Lunar-Diurnal Variation with Reference to the Moon’s Distance.— Both Sabine and Broun have shown that this variation is greater for perigee than for apogee. Broun has found that the mean ratio of the apogee effect to the perigee effect is as 1 to 1 '24 nearly. He remarks tlιat “the ratio of the moon’s mean distance from the earth in the half orbit about apogee is to that in the half orbit about perigee nearly as 1 '07 is to 1; as the cube of 1 ‘07 is 1’23 nearly, we see that the mean ranges of the curves for the two distances are in the approximate ratios of the inverse cubes of the moon’s distance from the earth, as in the theory of the tides.”

99. Lunar-Diurnal Variation with Reference to the State of the Sun’s Surface.— Kreil in a memoir presented to the Imperial Academy of Science in 1852 considers that the observations at Prague and Milan tend to show the existence of a solar period in the lunar variations. Sabine {Phil. Trans., 1856) has discussed the Toronto observations and also {Phil. Trans., 1857) the Hobart Town observations with the view of deciding this important point, and has come to the conclusion that there is no systematic difference in the lunar-variation corresponding to the decennial period of the solar-diurnal variation, but merely such casual fluctuations as might be reasonably expected, considering the shortness of the periods which they represent.

We have made a preliminary discussion of Broun’s Trevandrum observations with the view of throwing a little more light on this subject. For this purpose we have taken the ranges of the lunar-dnιrnal variations recorded by him for each month of each year. We have in the first place grouped these ranges together into threes, representing quarterly results, and have then compared together these quarterly results for years of small and for years of great sun-spot frequency,—assuming the years 1854-56 and 1863-64 to represent the former, and the years 1857-62 to represent the latter. We have thus obtained the following result:—

Table XXXII.—Relation between Lunar-Diurnal-Ranges of Declination at Trevandrum and Sun-Spot Fregtιency. [table]

It would appear from this table that such ranges are greater at years of maximum than at years of minimum sun-spot frequency. Nevertheless the proof is not conclusive, inasmuch as associated with such lunar ranges we may have remnants of solar disturbance, the tendency of which might possibly be to increase the apparent range. Now such disturbances are more frequent at times of maximum sun-spots, and it might therefore be conjectured that this tendency would be to increase the apparent lunar range at such times above the range corresponding to years of minimum sun-spots. On the whole we are not disposed to think that the evidence already adduced is sufficient to decide this question as a matter of fact either in the one direction or in the other.

100. Variation in the Diurnal Range of Declination depending upon the Age of the Moon.— Capello {Annals of Lisbon Observatory, 1876) and Stewart {Proc. Roy. Soc., 1877) have separately found that the range of declination is greatest about the times of new and full moon, a result recently confirmed by C. Chambers of Bombay. The following are the results recorded by Stewart derived from 197 lunations at the Kew Observatory :— [table]

101. Earth Currents caused by Lunar Influence.— Mr Alexander J. S. Adams {Nature, March 3, 1881) has made use of a telegraph wire the ends of which were connected with the earth, the one at Cardiff and the other at London. He observed the currents on this wire every quarter of an hour from March 28 to April 26, 1879. He believes that such are earth currents, and that they have a lunar-diurnal variation. There are during the lunar day, according to this observer, four electrical maxima, two positive and two negative, each maximum being divided from the next by a zero or point of no current. His results likewise exhibit a regular retardation or lagging of the earth-current variations behind the corresponding phase of the moon to the extent of nearly three hours, but these require confirmation from further observations.

The Effect of Solar Variability on the Meteorology of the Earth.

102. We may learn from the preceding paragraphs that the sun exercises a more powerful influence upon the magnetism of the earth at times of maximum than at times of minimum sunspots. It remains now to find whether a similar variability obtains in the phenomena of terrestrial meteorology. For this purpose we may divide the meteorological elements into the four divisions of pressure, rainfall, wind, and temperature, inquiring in what manner these are affected by solar inequalities.

103. Pressure.— In 1871 Baxendell {Memoirs of the Lit. and Phil. Society of Manchester, 1871-72), from an analysis of eleven years of the Radcliffe observations, Oxford, came to the conclusion that in years of maximum sun-spots the maximum barometric pressure occurred under north-east winds and the minimum under southwest, while in years of minimum sun-spots the maximum and minimum pressures took place respectively under north and southeast winds. He came likewise to the conclusion that, besides this difference in distribution of the convection currents of the earth, the forces which give rise to the movements of the atmosphere appear to be more energetic in years of maximum than in years of minimum sun-spots.

104. A study of the isobaric lines would appear to be the best method of obtaining information upon this important point. It is well known that during summer the interiors of large continents have a peculiarly low and the surrounding oceans a peculiarly high atmospheric pressure; while a disposition exactly the reverse obtains during the winter months. All this is no doubt due to solar action, and we might therefore imagine that when the sun is peculiarly powerful these peculiarities of distribution will be found to be increased in prominence, while they will be diminished at times of comparatively weak solar influence.

The distribution of pressure has been studied with much success by the Indian meteorologists, including Messrs Archibald, Blanford, Broun, Charles and Frederick Chambers, Eliot, and Hill, and the following conclusion is the result of their labours. We may assume that the Indo-Malayan region has for the mean of the year a barometric pressure probably below the gefteral average of the earth. λVe might therefore imagine that during years of powerful solar influence this peculiarity would be increased. Now these observers have found that in this Indo-Malayan region the barometer is abnormally low during times of maximum sunspots. Again, western Siberia is a district which in the winter season has a pressure decidedly above the average, and we should therefore imagine that during years of powerful solar influence this winter pressure should be peculiarly high. But this is what Blanford has found in his discussion of the Russian stations to correspond with years of maximum sun-spots.

Again, Frederick Chambers has enunciated the following laws as resulting from his discussion of various meteorological records:—

(1) Variations of the sun-spot area are succeeded some months afterwards in the Indo-Malayan region by corresponding abnormal barometric variations, a high barometer corresponding to a minimum of sun-spots.

(2) This lagging behind is greater for easterly than for westerly stations. In other words, this, like other meteorological phenomena, appears to travel from west to east.

We may therefore conclude that the barometric evidence as far as it goes is in favour of the hypothesis that the sun is most powerful at times of maximum sun-spots.

105. Rainfall— Heights of Rivers and Lakes. — In 1872 Meldrum of the Mauritius Observatory brought forward evidence showing that the rainfalls at Mauritius, Adelaide, and Brisbane were on the whole greater in years of maximum than in years of minimum sun-spots. Shortly afterwards it was shown by Lockyer {Nature, December 12, 1872) that the same law was observable in the rainfalls at the Cape of Good Hope and Madras.

Meldrum has since found that the law holds for a great number of stations, including eighteen out of twenty-two European observatories, with an average of thirty years’ observations for each. The results are exhibited in the following table (XXXIV.) :— [table]

It would, however, appear from the observations of Governor Rawson that the rainfall in Barbados forms an exception to this rule, being greatest about the times of minimum sun-spots.

106. Gustav Wex in 1873 s showed that the recorded depth of water in the rivers Elbe, Rhine, Oder, Danube, and Vistula for the six sun-spot periods from 1800 to 1867 was greater at times of maximum than at times of minimum sun-spot frequency. These conclusions have since been confirmed by Professor Fritz.

Quite recently Stewart {Proc. Lit. and Phil. Soc. of Manchester, 1882) has treated the evidence given by Fritz as regards the Elbe and Seine in the following manner. He divides each sun period, without regard to its exact length, into twelve portions, and puts together the recorded river heights corresponding in time to similar portions of consecutive sun periods. He finds by this means residual differences from the average representing the same law whether we take the whole or either half of all the recorded observations, and whether we take the Elbe or the Seine. The law is that there is a maximum of river height about the time of maximum sun-spots and another subsidiary maximum about the time of minimum sun-spots. There is some reason too to think that the Nile and Thames agree with those rivers in exhibiting a maximum about the time of maximum sun-spots and a subsidiary maximum about the time of minimum sun-spots, only their subsidiary maximum is greater than it is for the Elbe and Seine.

107. In 1874 G. Μ. Dawson came to the conclusion that the levels of the great American lakes were highest about times of maximum sun-spots. In this investigation the value of the evidence derived from rivers and lakes is no doubt greater than that derived from any single rainfall station, inasmuch as in the former case the rainfall of a large district is integrated and irregularities due to local influence thus greatly avoided.

108. Dr Hunter, director-general of statistics in India, has recently shown {Nineteenth Century, N ovember 1877) that the recorded famines have been most frequent at Madras about the years of minimum sun-spots—years likewise associated with a diminished rainfall.

109. Winds and Storms.— Meldrum of the Mauritius Observatory found in 1872, as the result of about thirty years’ observations, that there are more cyclones in the Indian Ocean during years of maximum than during years of minimum sun-spots. The connexion between the two is exhibited in the following table :—

Table XXXV.— Comparison of the Yearly Number of Cyclones occurring in the Indian Ocean with the Yearly Number of Spots on the Sun. [table]

In 1873 Μ. Poëy found a similar connexion between the hurricanes of the West Indies and the years of maximum sun-spots. He enumerated three hundred and fifty-seven hurricanes between 1750 and 1873, and stated that out of twelve maxima ten agreed.

110. In 1877 Mr Henry Jeula, of Lloyd’s, and Dr Hunter found that the casualties on the registered vessels of the United Kingdom were 17⅛ per cent, greater during the two years about maximum than during the two years about minimum in the solar cycle.

111. Temperature.— Baxendell, in a memoir already quoted, was the first to conclude that the distribution of temperature under different winds, like that of barometric pressure, is sensibly influenced by the changes which take place in solar activity. In 1870 Piazzi Smyth published the results of an important series of observations made from 1837 to 1869 with thermometers sunk in the rock at the Royal Observatory, Edinburgh. He concluded from these that a heat wave occurs about every eleven years, its maximum being not far from the minimum of the sun-spot cycle. Sir G. B. Airy has obtained similar results from the Greenwich observations. In 1871 E. J. Stone examined the temperature observations recorded during thirty years at the Cape of Good Hope, and came to the conclusion that the same cause which leads to an excess of mean annual temperature at the Cape leads equally to a dissipation of sun-spots. Dr W. Koppen in 1873 discussed at great length the connexion between sun-spots and terrestrial temperature, and found that in the tropics the maximum temperature occurs fully a year before the minimum of sun-spots, while in the zones beyond the tropics it occurs two years after the minimum. The regularity and magnitude of the temperature wave are most strongly marked in the tropics.

112. The evidence now given appears at first sight to be antagonistic to that derived from the other elements both of magnetism and meteorology, and to lead us to conclude that the sun heats us most when there are fewest spots on its surface. This conclusion will not, however, be strengthened if λve examine the subject with greater minuteness.

Scientifically we may regard the earth as an engine, of which the sun is the furnace, the equatorial regions the boiler, and the polar regions the condenser. Now this engine may be supposed to work in the following manner. Hot air and vapour are carried along the upper regions of the atmosphere from the equator to the poles by means of the anti-trade winds, while in return the cold polar air is carried along the surface of the earth from the poles to the equator, forming what is known as the trade-winds. Whenever the sun’s heat is most powerful both trades and anti-trades should be most powerful likewise. But we live in the trades rather than in the anti-trades—in the surface currents, and not in the upper currents of the earth’s atmosphere. λVhen, therefore, the sun is most powerful is it not possible that we might have a particularly strong and cold polar current blowing about us? The same thing would happen in the case of a furnace-fire; the stronger the tire the more powerful the hot draught up the chimney, the more powerful also the cold draught from without along the floor of the room. It might thus follow that a man standing in the furnaceroom near the door might be chilled rather than heated when the furnace itself was roaring loudest. In fact temperature is a phenomenon due to many causes. Thus a low temperature may be due (1) to a deficiency in solar power, (2) to a clouded sky, (3) to cold rain, (4) to cold winds, (5) to cold water and ice, (6) to cold produced by evaporation, (7) to cold produced by radiation into space.

Blanford has recently shbwn that at certain Indian stations a low mean temperature occurs when there is an unusually large rainfall and a great amount of clouds, a result in accordance with the conclusions previously enunciated by Professor Piazzi Smyth. Records of maximum and minimum temperature must not therefore be too closely associated with a maximum and minimum of solar power.

113. Considerations of this nature have induced Stewart to imagine {Nature, June 16, 1881) that the true connexion between sun-spots and terrestrial temperature is more likely to be discovered by a study of short-period inequalities of sun-spots than by that of the elevcn-year period in which there is time enough to change the hygrometric state of ths atmosphere and the whole convection system of the earth. He has accordingly discussed at some length two prominent sun-spot inequalities of short period (about twenty-four days), and endeavoured to see in what way they affect terrestrial temperature. From this it appears that a rapid increase of sun-spots is followed in a day or two by an increase of the diurnal temperature range at Toronto. Now an increase of diurnal temperature range most probably denotes an increase of solar energy, and we are thus led to associate an increase of solar heat with a large development of spots. This, however, is a point which requires further investigation.

114. Gênerai Conclusion.— On the whole we may conclude that the meteorological motions and processes of the earth are probably most active at times of maximum sun-spots, and that they agree with magnetical phenomena in representing the sun as most powerful on such occasions, although the evidence derived from meteorology is not so conclusive as that derived from magnetism.

Hypothetical Views regarding the Connexion between the State of the Sun and Terrestrial Magnetism.

115. Principles of Discussion.— In the following discussion we claim only to advance a working hypothesis, with the view of suggesting further inquiries into the subject of terrestrial magnetism. It seems therefore desirable that we should limit ourselves to such probable or possible causes as are known to exist and to operate on the earth. These various agents or causes will be described, and we shall endeavour to show that converging lines of evidence point in several cases to certain of these as being most likely to produce that particular type of effect which is exhibited in terrestrial magnetism. This course will in our view most readily suggest further inquiries with the view of confirming or disproving the various points of this working hypothesis. Believing that the introduction of any unknown cause can only be justified when known causes have been found insufficient to account for the phenomena in question, we have not advocated any direct magnetic action of the sun upon the earth. We have refrained from this for two reasons,—first, because from what we now know of the sun it appears to us unlikely that it should exercise an influence of this nature upon the earth, since a body at a high temperature possessing very strong magnetic properties is unknown to us; and, secondly, we shall see further on that such an influence will not explain the best-understood magnetic changes, nor is there in our opinion any magnetic phenomenon for the explanation of which it appears absolutely necessary to resort to this hypothesis. In fine, without' presuming to deny the possibility of unknown influences of this nature, ft does not appear to us that the time has yet arrived when we are called upon to resort to such as necessary aids to the discovery of further truth.

. 116. Nature of Solar Variations.— It is quite certain that there is a variability in the visible appearance of the sun’s disk, which exhibits sometimes a comparatively large amount of spotted area while on other occasions it is entirely free from spots. Now it has been remarked by Thomson that were the sun an incandescent solid its surface would become cool in a few minutes. The astonishing property which our luminary possesses of pouring out continuously a vast amount of radiant energy must unquestionably depend upon machinery of great power by means of which fresh hot particles are rapidly brought from the interior to the surface, while those particles which have given out their light and heat are rapidly hurried downwards to be recruited from the great storehouse of heat in the sun’s centre. In fine, a gigantic system of convection currents of this nature forms the essential condition without which the suιι would not be able to continue shining as it does.

The mottled appearance of the sun’s disk as seen through a telescope denotes no doubt the existence of a vast system of ascending and descending currents, the hot matter rising from beneath being denoted by the brighter portions and the cold matter descending from above by the darker portions of the structure. On certain occasions and in certain regions of the sun the scale of these phenomena is greatly increased, and we have a huge up-rush of bright and a corresponding down-rush of black matter—in fine, the well-known sun-spot with its bright faculous appendages. Whenever sun-spots are very frequent we should therefore expect the convection system of the sun to be particularly powerful, and the great velocity and size of the red flames or the higher portions of the convection system observed around the sun’s limb on such occasions confirm us in this supposition. And if the convection system of the sun be particularly powerful when there are most spots on its surface, it would seem to follow that the radiation from our luminary should on such occasions be particularly powerful also. The spectroscope leads us to the same conclusion. It would appear from the observations of Lockyer and others that at times of maximum sun-spots certain definite regions of the sun when examined spectroscopically present all the appearances of a very high temperature.

We are unable to confirm these conclusions by direct observations of the sun’s heating power. Actinometrical determinations have not yet been made with sufficient accuracy and persistence to decide this point experimentally. We have, however, evidence of an indirect nature derived from terrestrial magnetism and meteorology all tending to make us think that the sun is most powerful during times of maximum sun-spots. We have seen that on such occasions the solar influence upon the magnetism of the earth is peculiarly powerful in more than one way, and that its influence on meteorology is then peculiarly powerful also, although we are not so certain of this latter fact as of the former.

We may therefore take it to be most probable that the sun is most powerful at times of maximum sun-spots, and proceed from this basis to propound the two following questions :—in the first place, what is the nature of the solar influence upon terrestrial magnetism? and, secondly, why is this influence so much more easily perceived than certain forms of solar influence upon meteorology?

117. Diurnal Magnetic Variations—Hypotheses regarding them.— The various speculators on the cause of these phenomena have ranged over the whole field of likely explanations. (1) It has been supposed that the sun acts directly as a magnet upon the magnetism of the earth. (2) It has been imagined that convection currents established by the sun’s heating influence in the upper regions of the atmosphere are to be regarded as conductors moving across lines of magnetic force, and are thus the vehicle of electric currents which act upon the magnet. (3) Faraday, reasoning from his discovery that oxygen is paramagnetic, and becomes weaker in its power when heated, and stronger when cooled, supposed that the sun by heating certain portions of the atmosphere renders them less magnetic while others, not subjected to any heating influence are rendered more magnetic. The action is equivalent to a displacement by means of the sun of the magnetic matter of the earth, and involves a displacement of the lines of force. Here too the solar action is associated with the great mass of the atmosphere. (4) It has been supposed by Christie and by De la Rive that the heat of the sun produces in the atmosphere and in the earth thermoelectric currents which produce the daily magnetic variations. It is not easy to perceive how we could have thermo-electric currents in the upper regions of the atmosphere, but there is no obvious objection to the generation of such currents in the crust of the earth. Thus the first hypothesis has no reference to the atmosphere whatever; the second deals with the upper atmospheric regions, the third with the great body of the atmosphere, while the fourth, as we have ventured to modify it, has reference to the crust of the earth.

118. Discussion of these Hypotheses.— Dr Lloyd and Mr C. Chambers have both shown that direct solar magnetic action will not account for the peculiarities of the diurnal magnetic variation. Again (§ 48) we have strong evidence that changes in the range of the daily magnetic variation lag behind corresponding solar changes in point of time. Now this kind of behaviour is inconsistent with direct magnetic action, and points rather to an indirect magnetic effect caused by the radiant energy of the sun.

119. Let us therefore dismiss the hypothesis of direct action and consider that of Faraday. We know both from observations of the declination and horizontal force (Proc. Roy. Soc., March 22, 1877, and Phil. Trans., 1880, p. 541) that the action of the sun in producing diurnal variations of these elements is one and a half times as powerful at epochs of maximum as it is at epochs of minimum sun-spot frequency. It is hardly credible that there should be such a great difference on these occasions in the sun’s heating effect upon the great bulk of the atmosphere. Meteorologists have never observed such a difference, nor is there any marked corresponding inequality of diurnal temperature range. Meteorological evidence is thus against the diurnal magnetic changes being due to the heating up by the sun of the great mass of oxygen which constitutes the magnetic portion of the earth’s atmosphere. Again, as there is a preponderance of hot oxygen in the northern hemisphere during the June and in the southern hemisphere during the December solstice, there ought according to this theory to be a well-marked annual variation of the magnetism of the earth, the northern hemisphere being at the same time differently affected from the southern. But there are no traces of such a phenomenon, the annual and semiannual variations which we have already described (§§ 64-67) being of quite a different nature, and none of them very large.

120. Precisely the same objections apply with even greater force to the fourth hypothesis. It seems impossible to allow that any heating effect of the crust of the earth caused by the sun can be one and a half times as great at epochs of maximum as it is at epochs of minimum sun-spot frequency.

121. We are thus driven by the method of exhaustions to look to the upper regions of the earth’s atmosphere as the most probable seat of the solar influence in producing diurnal magnetic changes, and it need hardly be said that the only conceivable magnetic cause capable of operating in such regions must be an electric current. Now we know from our study of the aurora that there are such currents in these regions—continuous near the pole and occasional in lower latitudes. A good deal has been said about the difficulty of imagining a daily set of currents to be generated in regions of such imperfect conductivity, but we shall see by and by (§ 134) that there seems ground for imagining that their conductivity may be much greater than has hitherto been supposed.

122. Analogies between the Meteorological and Magnctical Systems of the Earth.— We have in the first place a zone of maximum terrestrial temperature, the middle line of which is nearly coincident not only with the geographical but likewise λvith the magnetical equator. Again, there are possibly in the northern hemisphere two poles of greatest cold, which possibly do not greatly differ in position from those spots which we have called magnetic poles or foci. About the southern hemisphere we have no information.

Furthermore we believe that the hot air is carried from the zone of greatest heat to the place or places of greatest cold by means, no doubt, of the return trades which blow in the upper atmospheric regions. The hot air divides at this zone, one part blowing northwards in the northern and another southwards in the southern hemisphere. Now this zone, from which the anti-trades divide, has an annual motion of its own, being found farthest north at the June solstice and farthest south at that of December. Probably too the northern system is strongest in June and the southern system in December. If we now turn to the solar-diurnal variation of magnetic declination, we find here also a northern and a southern system (§ 41), the type of the one being antagonistic to that of the other. We find also that the northern system is strongest in June and the southern system in December.

Again, it seems probable from what we have now said that the anti-trades, strictly speaking, have reference not to the geographical equator and poles but to the zone of maximum and the poles of minimum temperature. Now, turning once more to the diurnal oscillations of the declination needle, it seems probable that the directions east and west must be interpreted as having a reference not to the geographical but to the magnetical pole (§ 45).

These analogies must be taken for what they are λvorth. Our object in introducing them has reference to the previous discussion, from which we concluded that the magnetic influence of the sun is probably due to currents in the upper region of the atmosphere— the cause of which we were content to leave in abeyance. Now these analogies would lead us to suggest that this cause, whatever it is, may perhaps be found to be related to the convection system of the earth on the one hand and to the magnetic system on the other.

123. Analogies between Meteorological and Magnetical Weather.— These remarks are borne out by the further analogy which appears to subsist betwixt what we have termed meteorological and magnetical weather. Let us take the solar-diurnal variation of declination. Not only is this variation similar in form to the diurnal variation of atmospheric temperature (§ 37), but the ranges of the two have a similar annual variation. And, as the element of meteorological weather affects the orderly march of the temperature range, just so the element of magnetical weather affects the orderly march of the declination range.

Furthermore, just as temperature-range weather progresses from west to east (§ 52), so declination-range weather would seem to progress in the same direction as the other (§ 52) although at a greater rate. It will doubtless require a more extended investigation to make us quite sure of this latter point; nevertheless we do not perceive the validity of the objection that is sometimes made to the hypothesis of progress in magnetic weather on the ground that magnetic influences are known to affect all portions of the globe simultaneously. It will, we think, be perceived that in the above statement no supposition whatever is made with respect to the rate of propagation of a magnetic influence through the earth; this may be instantaneous or it may not. It is supposed that we have here a travelling cause of excitement, say a travelling cause of currents in the upper regions of the atmosphere which progresses from west to east and always produces its most marked effect above those regions where it passes—just as the sun itself in passing from east to west produces a magnetic effect the various phases of which travel from east to west with the sun which causes them. We think too that this hypothesis of travelling causes of magnetic change is strengthened by the facts observed by Capello and described in § 97.

124. If, however, the objection made to this hypothesis refers to the fact disclosed by Broun (§ 85) that changes of horizontal force appear to take place simultaneously at distant parts of the earth’s surface, then we think that analogy should lead us not to deny the possibility of a travelling magnetic excitement, but rather to suggest the possibility of there being some meteorological influence which, like the magnetical one above mentioned, may be found to take place simultaneously at different parts of the earth’s surface. Now Broun (Proc. Roy. Soc., May 11, 1876) has given us preliminary evidence for supposing that there are simultaneous barometric variations. For instance, there was a barometric maximum at Hobart Town, Peking, the Cape, St Helena, Makerstoun, Singapore, Madras, Simla, Ekaterinburg, and Bogoslovsk about the end of March or first day of April 1845. There appears to have been a simultaneous increase of the horizontal force of the earth at various stations much about the same time, and there also appears to have been a short-period maximum of spots on the solar surface. Broun has likewise registered simultaneous barometric variations at Singapore, Madras, and Simla, for the first three months of 1845. From these it would seem that simultaneous barometric maxima are possibly coincident with rapidly increasing sun-spot areas.

Again is it not absolutely certain that if there is a sudden increase of solar power this must mean an increase of heat communicated to the earth, although it may be difficult or even impossible to obtain experimental evidence of such a fact? All these are subjects which require further investigation.

125. Further Remarks on the Solar-Diurnal Variation of Declination.— In § 24 we have asked how far the action of the solar-diurnal force upon a freely-suspended magnet is due to currents acting directly upon the magnet and how far to a change produced in the magnetism of the earth. Some light appears to be thrown on this point by the behaviour of the needle at places near the magnetic pole where the dipping needle is nearly vertical. On opposite sides of this locality the declination needle points in opposite directions. Now suppose that we have a set of such needles placed all round this region. It seems a legitimate generalization from the observations described by Sabine (§ 45) to conclude that if we place ourselves above the centre of any of these needles at 8 A. μ., and look towards its marked pole, we shall find it in every case deflected towards the right, while if we look towards the same pole at 2 p. μ. we shall find it deflected to the left. Now if we imagine that at 8 A. μ. there are above these magnets (in the upper atmospheric regions) electrical currents of which the horizontal components form a set of positive currents flowing from the pole on all sides, then by the known laws of such currents the marked pole of all these needles will be deflected towards the right. And if at 2 p.M. the resolved portions of such currents should be flowing towards the pole, then the marked poles of all these needles will be deflected towards the left. It thus appears that this peculiar magnetic behaviour might easily be explained by a hypothetical distribution of currents. And in fact in such regions we have indubitable evidence of the existence of currents in the upper regions of the atmosphere. On the other hand this behaviour could not easily be explained by the hypothesis of some definite temporary magnetic system set up by the solar influence in the earth, for in such a case we should imagine that similar poles of all the needles ought to be deflected towards the pole of this temporary system, which is not the case.

126. Another point for consideration is the possible complexity of the solar-diurnal variation. For we may imagine (1) that the sun acts in such a manner as to produce a diurnal variation; (2) it may also act like the moon (§ 94) and produce a semidiurnal variation; (3) these possible actions may be accompanied by induced currents in the upper regions of the atmosphere and in the crust of the earth; (4) it is possible that the sun’s rays may affect these variations or some of them in the way in which Broun found that the lunar variation at Trevandrum was affected by the sun. It was found by him that the lunar action was considerably increased when the sun was above the horizon of the place.

127. We have pointed out (§ 119) that, while there is a marked likeness in many respects between the diurnal variation of declination and that of atmospheric temperature, we have yet no long-period fluctuation of the diurnal range of temperature at all comparable in magnitude to the magnetic fluctuations. It does not, however, seem difficult to account for this difference if we imagine that the magnetic fluctuations take their origin in the upper atmospheric regions, while the temperature fluctuations are due to the lower regions of the earth’s atmosphere. For, as the sun increases in power from times of minimum to times of maximum sun-spot frequency, we may imagine that a continuously increasing amount of aqueous vapour will be taken into the earth’s atmosphere.

Now the experiments of Tyndall and others induce us to think that the air would under such circumstances become more and more opaque for certain rays of the sun, and thus a continuously decreasing proportion of the sun’s heat would be able to penetrate into the lower atmospheric regions. This latter influence would therefore operate to cloak, perhaps to a considerable extent, the effect of the sun’s increasing power; and this may very well be the reason why the temperature range at the earth’s surface does not exhibit the same eleven-yearly inequality as the declination range.

128. There seems, however, reason to believe that if we go from long to short period inequalities there is a much greater similarity in the range of the magnetical and the meteorological changes (§ 113). The explanation seems tobe that in the short-period changes the sun has not time to alter sensibly the constitution of the atmosphere, and hence the proportional increase of effect experienced in the upper atmospheric regions is more nearly the same as that experienced near the surface of the earth.

129. Magnetic Disturbances.— There is strong evidence that the most important disturbances break out very nearly simultaneously at widely different parts of the earth, and that they even affect both hemispheres at the same time. Very little, however, is known about the modus operandi of the forces concerned in producing such disturbances. For instance, it is not known whether a disturbance permanently affects the magnetic state of the earth, e.g., whether one of the magnetic elements before a disturbance begins is sensibly different in value from what it is after the disturbance has ceased to exist. On the other hand we know (1) that disturbances break out on the very day when there are rapid changes taking place on the sun’s surface (§ 83); (2) that they generally begin by momentarily increasing the horizontal force, but that the type quickly changes, so that during most disturbances the horizontal force is diminished (§ 86); (3) that large disturbances take place more particularly about the equinoxes, when, we have reason to believe, the horizontal force of the earth is at a minimum (§ 77). May we not possibly conclude from these habits of action that at times of disturbance the earth is magnetically in a delicate state of equilibrium, perhaps having more magnetism than its surroundings would strictly λvarrant, and being therefore inclined to part with some, and that a sudden increase of solar activity, tending, as such changes probably do, at first to exalt the magnetism of the earth, nevertheless destroys its magnetic balance and gives it ultimately the opportunity of parting with some of its magnetism? This can only be regarded as a speculation, inasmuch as we do not know whether or not a disturbance produces any permanent influence upon the magnetism of the earth.

130. Auroras and Earth Currents.— There is no doubt that these phenomena denote electric currents in the upper regions of the atmosphere and in the moist conducting crust of the earth. The point in dispute is with respect to the origin of such currents. Some are inclined to regard auroras as peculiar manifestations of atmospherical electricity in high latitudes, while others imagine that such displays are rather of the nature of induced currents generated by small but abrupt changes taking place in the magnetism of the earth. The advocates of the first view do not deny that currents taking place somehow in the upper atmospheric regionswill have their conditions modified, to some extent at least, by the inducing influence of magnetic changes. Nor will the advocates of the induction hypothesis be disposed to deny the possibility or even the certainty that displays due to atmospherical electricity and not dissimilar to some kind of aurora take place in some region of the atmosphere. But the first party regard auroras rather as the cause than as the effect of magnetic changes, whereas the advocates of induction regard such displays rather as the effect than as the cause of changes somehow produced in the magnetism of the earth. And here it is desirable to remark that the advocates of the induction hypothesis take for granted the magnetism of the earth and the changes thereof as phenomena for which they do not profess to account, whereas unless we go to some absolutely unknown cause (and this is against our present programme) we must look to atmospherical electricity as likely to throw light upon the origin of terrestrial magnetism. We cannot therefore dispense with regarding atmospherical electricity as an agent which may have played an important part in the development of the present magnetical condition of the earth, but we are yet of opinion that, under the present state of things, the theory which holds by atmospheric electricity must largely be supplemented by the induction hypothesis if it is to explain the peculiarities in type or form of the phenomena which observation brings before us.

131. Professor Tait in his essay on thunderstorms attributes one kind of aurora to atmospherical electricity. Such an aurora is, he believes, the manifestation of almost continuous discharges, like those given by a Holtz machine in a vacuum tube. The cause is condensation of vapour going on very slowly in very large spaces of air. The electricity is due to previous contact of particles of air and vapour. The result is that the air-particles in the mixture in time acquire a definite difference of potential from those of vapour,— so that, when the latter aggregate, a misty region well charged is the result, and this discharges to the oppositely electrified air all round.

132. Again, Professor Stokes, without attempting to account for the origin of atmospherical electricity, has produced an hypothesis with the view of explaining the intimate connexion subsisting between auroral displays, earth currents, and magnetic changes on the one hand and outbursts of sun-spot activity on the other. His idea is that two somewhat distant atmospheric regions A and B are charged, let us say, with positive and negative electricity respectively; A induces in the ground below it a charge of negative, B a charge of positive electricity. At first things are held in this state: A cannot discharge either through the upper atmospheric regions to B or through the lower regions to the ground beneath it, while B is in a position precisely similar. Presently, however, an increase of the radiative power of the sun is produced. Such an increase would probably imply not merely an increase in general radiation but a particular increase in such actinic rays as are absorbed in the upper regions of the earth’s atmosphere. The layer of atmosphere between A and B will therefore greedily absorb such rays, its temperature will rise, and, as is known to be the case for gases, the electrical conductivity of the stratum will be increased. A discharge will therefore ultimately take place in the upper regions between A and B; this will relieve the charges of negative and positive in the ground immediately beneath A and B, and these charges will therefore rush together through the ground, producing an earth current. This earth current will be in the opposite direction from the atmospheric current, and the two will combine to represent, virtually at least, if not absolutely, a closed circuit. This will of course affect the earth’s magnetism and. produce a disturbance.

133. This hypothesis certainly affords a good explanation of the promptness with which disturbances follow increased solar activity (§ 83). Unless we are to resort to some unknown cause it is difficult to think of any other possible explanation of this fact. Such an explanation appears too to receive corroboration from the fact (§ 97) that the lunar influence on the earth’s magnetism as observed at Trevandrum is greater during the day than during the night,— greater possibly too at times of maximum than at times of minimum sun-spots. We are therefore disposed to accept this explanation of the way in which increased solar activity produces magnetic disturbance as the best that has been brought forward.

134. This does not, however, decide the disputed point how far these elevated currents are due to atmospherical electricity and how far to induction. The argument against the possibility of induced currents in these regions is derived from experiments with vacuum tubes, such as those recorded by Messrs De la Rue and Miiller, which would seem to indicate that enormous differences of potential would be required to produce electrical currents in elevated regions, where the atmosphere is very rare.

Indeed, on account of these experiments, the measurements of the old observers, who sometimes assigned a height of more than 100 miles to the aurora, have been called in question, and it has been supposed against direct observation that these phenomena must always occur in regions much less elevated. It would appear too that such reasons were influential in determining Professor Stokes to regard the aurora as produced by atmospherical electricity which, as we know from ordinary lightning, presents us with enormous differences of potential; but it is to be remarked that he has carefully guarded himself against the possibility of laboratory experiments with vacuum tubes not being strictly analogous to that which takes place in the upper atmospheric regions. Now it would appear that recent experiments by Hittorf throw some doubt upon the strictness of this analogy. The high difference of potential required to force the current through vacuum tubes is, according to this observer, due in great part if not entirely to the passage of the fluid from the terminal to the residual air of the tube, so that the potential requisite to pass a current through a tube of double length is not sensibly greater than that required for a tube of single length. The whole subject is one which demands further investigation; meanwhile we are not disposed to assert the impossibility of induction currents taking place in the upper atmospheric regions.

135. Let us now consider whether the form or type of the earth currents observed during disturbances favours the presence of induction to any sensible extent. The remarks of Dr Lloyd already quoted (§ 93), which are confirmed by the Greenwich observations, seem to be decisive in this respect. These may be interpreted in the following manner. In a magnetic disturbance we have frequently a general displacement of the various elements—the horizontal force, for instance; now on the curve which represents this slow but considerable displacement a large number of comparatively small but very abrupt changes are superimposed. These latter appearances are invariably accompanied by quick and strong alternations from positive to negative of the earth currents, while the former slow motion, although it may be of large range, hardly appears to have any galvanic equivalent at all. This would appear to favour the induction hypothesis, according to which small but abrupt magnetic changes should give rise to strong earth currents alternately positive and negative without reference to the position of the magnet above or below its normal at the time.

136. Another fact bearing upon this hypothesis is that mentioned in § 88. From this it would appear that on ordinary occasions the curves recording the progress of the declination needle at Kew and Stonyhurst are as nearly as possible identical, but on occasions of disturbance the range at Stonyhurst is greater than that at Kew by an amount not apparently depending so much on the magnitude of the disturbance as on its abruptness. The introduction of the element of abruptness would appear to be in favour of the mixing up to some extent of induced currents with the phenomena in question.

137. Sir George Airy has not been able to detect any resemblance in form between the regular diurnal progress of the magnet and that of the earth currents. It seems, however, possible that the peaks and hollows alluded to in § 73 may form an important and integral part of the daily magnetic movement, and there even appears to be some evidence that the diurnal progress of the earth currents bears a nearer resemblance to that of the peaks and hollows than it does to the progress of the smoother curve which is usually held to represent the diurnal variation. But this is a question which can only be decided by more prolonged investigations.

138. To conclude, there can be no doubt that at times of great magnetic disturbance we have currents in the upper atmospheric regions and in the crust of the earth which, so far as we can see, must either be due to atmospherical electricity or to induction, or to a mixture of both. The proportions of this mixture can only be decided by further inquiry and by the multiplication of stations where atmospherical electricity and earth currents may be observed. It ought to be mentioned that the experience of the Kew observers, as far as this extends, seems unfavourable to the hypothesis of a connexion between auroras and atmospheric electricity.

139. Lxmar-Semidiurnal Variation.— From the fact observed by Broun (§ 98) that the moon’s magnetic influence is as nearly as possible inversely proportional to the cube of the moon’s distance from the earth, it is impossible to refrain from associating it either directly or indirectly with something having the type of tidal action, but in what way this influence operates we cannot tell. Is it possible that the earth currents observed by A. Adams (§ 101) are induction currents generated in the conducting crust of the earth by the magnetic change caused by the moon,—inasmuch as these currents were found by him to be strongest in one direction about the lunar hours 3 and 15, when the lunar-diurnal magnetic effect is changing most rapidly in one direction (§ 95), while they were found to be strongest in an opposite direction about the lunar hours 9 and 21, when the lunar-diurnal magnetic effect is changing most rapidly in an opposite direction?

140. We might perhaps expect from the analogy of the tides that the sun should possessa semidiurnal magnetic effect similar in type to that of the moon. Now Sir George Airy in his analysis of the earth currents observed at Greenwich {Phil. Trans., 1870) during days of tranquil magnetism has detected in such currents a semidiurnal inequality having maxima in one direction at solar hours 3 and 15, while it has maxima in the opposite direction at solar hours 9 and 21. The reference to solar hours in this inequality is thus precisely similar to that which the inequality observed by Adams bears to lunar hours.

141. If there are induced currents of this nature in the crust of the earth, we might naturally suppose that there will be corresponding currents in the upper regions of the earth’s atmosphere, and in accordance with the suggestion made by Professor Stokes (§ 132) we might perhaps suppose that these currents will be strongest when the upper atmospheric regions are heated by the sun and thereby rendered better conductore. Is it not possible to suppose that the influence of daylight upon the lunar magnetic effect discovered by Broun (§ 97) may be due to this cause, and may it not also induce us to recognize the possibility of a maximum lunar influence (§ 99) at times of maximum sun-spots, when there is reason to believe that solar radiation is most powerful?

142. Secular Variation.— Sabine and Walker are agreed in regarding this variation as cosmical in its origin, and they are apparently of opinion that it is caused by some change in the condition of the sun. It seems difficult if not impossible to attribute it to anything else, since the terella of Halley cannot be now regarded as having a physical existence. Again it is more than possible—it is probable—that there are solar variations of much longer period than eleven years. On the other hand the evidence given in § 81, tending to show that an access of sun-spots produces a change in the magnetic state of the earth consistent with the hypothesis that the magnetizing power of the sun has then been augmented, requires to be confirmed by more observations; and even then it is certain that this magnetic change produced by a considerable change in spotted area is extremely small. We cannot therefore regard the very large secular magnetic change as due to a non-cumulative magnetic influence of some long-continued solar variation; nor does it seem possible to attribute the change to solar influence at all unless we regard this influence as producing results of a cumulative nature.

It is possible, however, to regard solar influence as producing a cumulative effect in one of two ways, or by a combination of both. For (1) time is necessarily an element in any influence acting upon the hard-iron system of the earth—presuming the earth to possess such a system. There are in fact indications in the results of § 82 that a system of this kind is perhaps connected with the American pole; yet, even allowing the influence of time, it seems difficrdt to account for the peculiarities of the secular variation by an hypothesis of this nature. But (2) any long-continued variation of solar power would no doubt act cumulatively in producing an increase or diminution of the large ice-fields round the poles of the earth. In the course of time this cumulative change in the extent and disposition of these might perceptibly alter the distribution of the convection currents of the earth—and these, according to the views herein indicated, might in their turn perceptibly alter the earth’s magnetic system.

143. Concluding Remarks.— If we agree to look for an explanation of terrestrial magnetism and its changes to strictly terrestrial processes, we may derive some assistance in our search from such considerations connected with symmetry as enable us, for example, at once to perceive that when two perfectly similar things are rubbed together we cannot have electrical separation, because there is no reason why the one should be positively and the other negatively electrified. Suppose then that an observer stands at the equator and looks towards the north, and then turns his back upon the north and looks towards the south. In the first position let him regard the northern system of meteorological processes and motions, and in the second the southern. Now if symmetry obtained absolutely in these systems—that is to say, if the observer, whether he regarded the northern or the southern system of things, had in either case precisely similar phenomena at his right hand and at his left—then we should see no reason why the earth should be a magnet, or why one hemisphere should be the seat of magnetism of the one kind rather than of the other. If then we regard meteorological processes and motions as being in some way the cause of terrestrial magnetism, we must direct our attention to that peculiar element which causes a want of perfect symmetry such as we have described in meteorological phenomena. This element can hardly be anything else than the rotation of the earth, which is from left to right to an observer facing the north, but from right to left to an observer facing the south.

144. Now if we look upon the terrestrial meteorological system modified by the earth’s rotation as having produced somehow in the past the magnetic state of the earth, it seems most natural to regard the system which formerly produced this magnetic state as being likewise that which at present maintains it in its efficiency, and which also accounts for the various magnetic changes which take place. It would seem therefore that terrestrial meteorology and terrestrial magnetism are probably cognate subjects, and that they ought to be studied together in the well-founded hope that the phenomena of the one will help us to explain those of the other.

Furthermore, if these meteorological processes—deriving their one-sided character from the earth’s rotation—are to be regarded as accounting not only for the origin but for the maintenance of the earth’s magnetic system, we can hardly fail to imagine that these processes must derive part of the energy which they exhibit from that of the earth’s rotation. Tidal energy we know is derived from this source; but we must likewise regard part of the energy displayed in convection currents whether in the air or in the ocean as derived no doubt from the same source. And we may perhaps allow that in the phenomena of tidal action, as well as in those of convection currents of the air and ocean, there may be, not merely a transmutation of actual energy directly through friction into heat, but likewise a transmutation of it, ultimately perhaps into heat, but first through the intermediate agency of electrical currents which serve to maintain the magnetic state of the earth and to produce magnetic changes.

Now if this be the case, if there be a large and complicated system of tidal and convection currents all tending to change the rotative energy of the earth ultimately into heat, whether directly through friction or indirectly through the medium of electricity, it is surely impossible with the present state of our knowledge to calculate with the smallest pretensions to accuracy at what rate this transmutation is taking place, and hence at what rate the velocity of the earth’s rotation is being slowly diminished. (B. S. )