and consists of a telescopic camera focussed on Polaris. This star is not at the north pole of the heavens but a little more than a degree distant, and, consequently, it describes a small circle in the heavens during twenty-four hours. When the sky is clear around Polaris its trail upon the photographic plate is continuous, but when the sky is partly or entirely covered with clouds the trail is broken or obscured. Of course the plate is not exposed until after dark, and a shutter is closed by a clock before dawn. The only hourly records of cloudiness at night in the United States are obtained by this instrument on Blue Hill and at Cambridge. It will be objected, perhaps, that the cloudiness derived from observations of the sun or the pole-star is not the amount over the whole sky, but only that in the region of the luminary. This is true, but it is found that the average of the records for a month or a year agrees very closely with the average of estimates of cloudiness over the whole sky during these periods. The use of the pole-star is preferable to that of the sun, because in our latitude it gives values at a point about half-way between the horizon and the zenith; while since the sun travels at a variable height across the sky, when its altitude is low the same mass of cloud may intercept more sunlight than when it shines vertically. From ten years' observations the

D

following deductions have been made concerning the variation in the amount of cloud at Blue Hill. For all the months the diurnal amount of cloud is greatest about one o'clock in the afternoon, on account of the frequency of cumulus clouds near the warmest part of the day, while the next greatest amount, due to the frequency of stratus clouds, occurs near sunrise, or at the coldest time of day. All over the world the least cloudiness is in the evening, when the sum of the combined effects of radiation and isolation is least. The annual period in the cloudiness is complex, because the amount of cloud is connected with changes of humidity at many different levels in the atmosphere, but in the northern hemisphere there is most cloud during the first half of the year and least during the latter half, probably because the increasing warmth at the earth's surface produces increased ascending currents until summer, while the chilling of the earth's surface in the autumn becomes unfavourable for ascending currents. The distribution of cloud over the globe is intimately connected with the general atmospheric circulation, being greater where there are rising currents and less where there are downward currents. The reason, naturally, is that as descending air becomes warmer and therefore relatively drier, the clouds in it evaporate and disappear. A cloudy belt

encircles the earth at the equator, and on either side are two belts of less cloud, but in higher latitudes the cloudiness increases. If we could see our earth from outside its atmosphere, the light reflected from the upper surfaces of the cloud-belts would probably make them appear bright. From the markings on a planet that are known to be caused by condensation, a French meteorologist, M. Teisserenc de Bort, believes that the circulation of its atmosphere can be inferred, for wherever on the surface of the planet bright spots are seen, there the vapour of rising currents should be condensed. If this be true, there is a resemblance between Jupiter, as we see it, and the earth as it would appear from another planet, the bright bands being cloud surfaces, and the dark patches glimpses of the surface of the planet beneath.

Observations of the direction of motion, and apparent velocity of clouds at different heights, have been made at Blue Hill several times a day since 1886. To measure the motion of clouds the nephoscope (Fig. 1) is used. It consists of a horizontal circular mirror with a concentric circle of azimuths and an eye-piece C, movable in a plane BD at right angles to the mirror and also around it, through which the image of the cloud is brought to the centre of the mirror A. It can be proved by geometry that the motion of the cloud-image is

proportional to the movement of the cloud itself, so by noting in what direction and how far the image is displaced in a given time, we have the true direction of motion of the cloud itself and also its relative velocity, comparable with the velocity of all clouds having the same height. If

FIG. 1.-Nephoscope at Blue Hill Observatory.

the height is known, then the relative velocity can be easily converted into absolute velocity, and thus the velocity of currents at different heights in the atmosphere is accurately ascertained.

The height of clouds seems to have been measured trigonometrically from two stations as early as 1644 by Riccioli and Grimaldi, two Jesuits of Bologna, but notwithstanding these measurements and some conclusions derived from observa

tions on mountains, and in balloons, the altitudes of the different clouds were not known with any accuracy until in 1884 Ekholm and Hagström made a series of trigonometrical measurements upon the different kinds of clouds at Upsala, Sweden. About the same time attempts were made at Kew Observatory to measure clouds by photography, and in 1885 probably the first trigonometrical measurements in America were made at Cambridge, Mass., by Professor W. M. Davis and Mr. A. McAdie. In 1890-91 the Swedish methods were employed at Blue Hill by Messrs. Clayton and Fergusson of the Observatory staff, and until recently the measurements there and at Upsala comprised all that was known accurately about the heights and velocities of the various species of clouds.

The trigonometrical measurements at Blue Hill were made as follows: at two stations, one at the Observatory, the other at the base of the hill about a mile distant, two observers determined simultaneously the angular altitude and azimuth of some point on the cloud which was agreed upon by telephonic conversation. If, as is generally the case, the lines of sight did not meet, the trigonometrical formulæ gave the height of a point midway between the crossing of these lines. Such was the accuracy of these measurements that the probable