scoPy); some are produced by the permanent gases, others by aqueous vapour, which is always present, though in variable amount. The absorption-spectrum of water-vapour has been minutely studied and carefully mapped by Janssen and by Cornu. In Angstrom’s table of normal solar spectra there are numerous groups of lines which appeared most conspicuous when the sun was close to the horizon, and many of these are coincident with the absorption-lines of water-vapour. They are found principally in the red and yellow, and the main group is seen a little to the red side of the D line. In small spectroscopes the water-vapour lines appear fused together into a band on the red side of D, or even as a mere widening of that line. There are several variable bands in the spectrum which come to a maximum of intensity when the sun is on the horizon. These are due to absorption by the permanent gases; the one which in a small instrument appears to separate the yellow from the green is frequently mistaken for the water-vapour band, and this is the cause of many incorrect “rainband predictions.” In 1872 Professor Piazzi Smyth noticed a change in the water-vapour lines before and after a sirocco in Palermo, and the same phenomenon was brought before him very strikingly in France before great rains in 1875 (Edin. Ast. Obs., vol. xiv.), when he named the main group of water-vapour lines the “rainband.”

The rainband may be observed with any spectroscope, but direct-vision instruments of medium size are most convenient. It is important that the spectroscope should have as great dispersion as possible and good definition, especially at the red end. To make an observation the slit, which must be kept perfectly clean, should be narrowed down until the spectral lines are sharp and clear. The instrument should be carefully focused to get the maximum absorption effect; but all observers lay stress on noting whether the intensity of the band decreases rapidly or gradually as the altitude is increased. When a dark band is observed at the horizon and at the zenith, heavy rain is almost certain to follow immediately. It is of little importance whether the sky be clear or covered with high clouds at the point of observation; low clouds or haze make the result untrustworthy by shortening the line of sight, thus reducing the strength of the band and equalizing it in all directions. The utility of the spectroscope in meteorology depends on its power of investigating the hygrometrical conditions of the whole slice of atmosphere looked through, and it affords a means of ascertaining the difference of humidity in different directions. The hygrometer only indicates the state, as regards moisture, of the few yards of air surrounding it. The great difficulty in the way of obtaining accurate results with the rainband spectroscope has been the mental scale of comparison employed by most observers. Although some have found it easy to estimate the intensity of the band from 0 to 20, it is beyond the power of many to describe it in figures even from 0 to 5. Professor Piazzi Smyth noted its strength relatively to that of the dry-air band between the green and yellow. The fixed solar lines E, b, and F have been used as a closer approximation to a scale in instruments not powerful enough to separate the rainband from the D line. This compound line appears to vary from something less intense than E in very dry weather to something considerably darker than F when rain is imminent. Seven degrees can be discerned, and these may be represented by using the sign = to mean “of equal intensity with,” > to mean “darker than,” and < “less dark than,” as, < E,=E, >E<b, = b,>b<F, = F, and > F. When the thin solar lines in the green are seen very distinctly there is less probability of rain falling than when they are indistinct or invisible.

The following table gives an idea of the intensities corresponding to rain-probabilities at Edinburgh. [table]

It appears that the average percentage of fulfilments of predictions of “rain” and “no rain,” made for a period of twelve hours after one observation in the morning, which may be expected in Scotland is about 75. In less variable climates, such as those of the south of Europe and parts of the United States, a much higher degree of accuracy has been attained. The precise strength of rainband which corresponds to the probability of a fall of rain within a definite time depends on the temperature and also on the place of observation; in every case it must be determined by the observer for himself. Very dark rainbands are found to precede rain in more than 95 per cent, of the cases everywhere; entire, or almost entire, absence of the band presages a dry day with equal probability. With a spectroscope powerful enough to split 1) a mental numerical scale must be used in default of a suitable micrometer, and by practice the observer will be able to draw up a table for its conversion into probabilities of rain.

The production of an artificial absorption-line the intensity of which could be varied by known degrees suggested itself to more than one observer as the principle for a rainband-micrometer. A wedge of didymium glass comes very near success in this direction, but the absorption-line is awkwardly situated. Professor Cook of Dartmouth College, the leading meteorological spectroscopist in the United States, has constructed a micrometer for a spectroscope of sufficient power to separate the rainband into lines. A silk fibre is fixed to a frame capable of being moved to and fro in the tube of the spectroscope by a micrometer screw 7 . When brought into focus the fibre appears as two sharp lines, which become fainter and w τ ider as it is withdrawn. When the lines appear of equal intensity with the most prominent line of the rainband (α of the D group in Janssen’s map) the micrometer is read; forty shades of intensity may be indicated by it. The unit proposed for graduating such micrometers on a uniform scale is the intensity of the faintest and least refrangible of a group of three lines in the red ( w. l. 6207) of the spectrum produced by a 1 centimètre column of the gas given off' by heated lead nitrate (mixture of nitrogen peroxide and oxygen) at 25° C. and 760 mm. The spectroscope when reinforced by this micrometer has been found to indicate with unfailing accuracy the existence in certain parts of the sky of banks of invisible cloud, which become visible when a fall of temperature and other necessary conditions allow the vapour to condense. In settled weather such masses of vapour are not to be found, the micrometer readings being the same in all directions at the same altitude. Sometimes the rainband grows gradually darker for several days before a period of steady and long-continued rain, while sudden violent showers may give very short notice.

The spectroscopic history of a thunderstorm observed on 9th June 1884 by Professor Cook is extremely interesting. During its approach the water-vapour line observed at 10° north of the zenith towards the storm was darker by ten degrees of the scale than that observed 10° south of the zenith, although to the eye the clouds presented exactly the same appearance at both places. On this occasion the strength of the line varied as follows :— [table]

Some relation has been traced between the variations of the rainband and the appearance of aurorae, but this matter is not yet fully investigated.

The best form of instrument, Glaisher’s rain and snow gauge recommended by the Royal Meteorological Society, is a cylindrical copper vessel 8 inches in diameter and 18 inches high, in which the funnel is placed about halfway down (see fig. 1).

The Scottish Meteorological Society largely employ Howard’s raingauge, a plain glass bottle holding about half a gallon and provided with a long 5-inch copper funnel, which has a collar fitting over the neck of the bottle to prevent rain from being blown in laterally by the wind. In some forms the funnel leads to a long glass tube divided into inches, tenths, and hundredths of rainfall. Mr Symon’s storm raingauge on this principle is intended to be read from a distance, and is only graduated into tenths of an inch. The water collected by a raingauge may be weighed instead of measured, but the latter process being much more simple is always adopted, at a slight expense of accuracy, however, as the variation of volume with temperature is not taken into account. Precipitation and evaporation being complementary phenomena, an atmometer or evaporation-gauge ought, strictly speaking, to accompany each raingauge. But none of the instruments yet devised can be regarded as satisfactory, accordingly a number of devices have been introduced to calculate or to minimize the evaporation from raingauges. Dr Garnett in 1795 proposed to use two gauges of unequal size, and recently Prof. Michie Smith has introduced a simplification by making the area of one gauge exactly double that of the other. If the evaporation is the same from each, the difference between the readings of the two gauges gives the true rainfall in the smaller. If A be the area of the funnel in the smaller, 2 A that of the funnel in the larger, V a certain volume of water placed in each gauge, E the evaporation, and II the inches of rainfall, then V + 2 All - E-{V+ AR-E) — AR, however V and E may vary. The simplest and best method is to use a funnel terminating in a long straight tube, which reaches almost to the bottom of the receiving vessel. Gauges have been constructed for experimental purposes always to face the wind, and with openings capable of being fixed at any angle. For use at sea they may be swung on gimbals; but when so employed the record must be supplemented by the readings of a hydrometer so as to detect and allow for any admixture with sea-spray. Self-registering and self-recording raingauges, as frequently used in meteorological observatories, are constructed on two leading types. In Hermann’s “hyetometrograph,” 1789, a fixed funnel conducts the rain into one of twelve glasses placed on the circumference of a horizontal wheel, which is turned by clockwork so that each glass remains under the funnel for one hour. In Stutter’s more recent instrument the receiving funnel delivers into a smaller funnel, which has a sloping tube and is carried round by clockwork so as to remain for one hour over each of twenty-four fixed glasses arranged in a circle. The second kind of self-registering instrument produces a continuous record of rainfall, indicating the hour of commencement and close of each shower, the amount of rain that has fallen, and the rate at which it fell. In Beckley’s “pluviograph” a pencil, attached to a vessel which sinks as it receives the rain, describes a curve on a sheet of paper fixed round a rotating cylinder; when full the receiver empties itself by means of a siphon and the pencil is carried rapidly upwards, describing a straight vertical line.

The higher a raingauge is placed above the ground, or rather above a broad flat surface, the smaller is the rainfall registered, as the following figures indicate :—

Height of funnel above ground in feet. .0 1 10 20 40 200

Bain registered (average)..............158־0 70־0 90־0 93'0 00־1 07־

When the mouth of the gauge is on or within a few inches of the ground the insplashing of raindrops increases the amount of water. Minute rain-spherules, which usually float in horizontal or oblique planes, are most numerous near the ground, where consequently they coalesce to form regular drops which fall into the funnel. The raindrops also increase slightly in size by condensing moisture as they fall. But the greatest effect is probably produced by wind, which forms eddies round high and isolated objects, thus more or less interfering with the fall of rain into the gauge. It is obvious that all raingauges intended for comparison should be fixed at the same height, and in Great Britain the standard distance of the mouth of the funnel from the ground or from a broad flat surface is one foot. The situation of a raingauge should be perfectly open, especially in the direction of the prevailing rainbringing winds. In measuring rain it is essential to see that the funnel is not indented or deformed in any way, and that the collecting vessel is inaccessible to air or rain except through the funnel. The temperature of rain as it falls should be observed whenever it is possible to do so. The amount of solid matter collected in the raingauge should be ascertained and recorded as bearing on Mr Aitken’s theory of rain (see Evaporation), and it should be examined microscopically for volcanic and cosmic dust.