day in its output of radiation within limts of from 5 to 10 per cent in quantity and in irregular periods of from 5 to 10 days. This conclusion I state tentatively. Before it can be accepted without question it must be confirmed by showing that the results obtained day after day at another equally good station, at a great distance, confirm those obtained simultaneously at Mount Wilson. Such a final test, it is now expected, will be made during the coming fiscal year.

1 Other days of observation not yet ready.

General mean, 1.922 calories (15° C.) per square centimeter per minute.
Number of determinations, 405.

Other observations made on Mount Whitney.—Although the main purpose of the Mount Whitney expedition of 1910 was served by proving that the determinations of the solar constant of radiation are independent of the altitude of the observing station, advantage was taken of the unusual opportunity to make several other kinds of observations. Kapteyn's sky photometer was employed there on two successive nights to measure the relative brightness of the different regions of the night sky and to estimate the total quantity of sky illumination per square degree compared with that of a first-magnitude star. Yntema had employed similar apparatus in Holland. He found the average brightness of the Milky Way about two or three times that of nongalactic regions of the sky, such as the north polar region, but that the sky near the horizon was of about the same brightness as the Milky Way. He concluded that the sky at night is illuminated more by some terrestrial sources of light than by the stars.

The results obtained on Mount Whitney at nearly 3 miles elevation agreed in general with those of Yntema. The following is a summary of the principal points. Mean values are given:

The total illumination from 1 square degree of polar sky was found to be 0.0746 that of one first-magnitude star in the zenith. It is possible that the fraction just given may be a little too small, owing to a source of error discovered after the observations were ended.

Computations from the Mount Whitney results confirm Yntema's conclusion that the great increase of brightness toward the horizon can not be due to any arrangement of starlight, but must be caused by some terrestrial source of light, perhaps a continuous faint aurora.

Bolometric measurements were made on Mount Whitney to determine the relative radiation of the sky by day in all directions, as compared with the

sun. These measurements were numerous and seem to have been successful, but are not yet reduced.

The sun's energy spectrum.-A summary has been prepared showing the mean result of determinations of the distribution of the sun's energy in the spectrum, as it would be found outside the atmosphere. The measurements on which it is based include Washington, Mount Wilson, and Mount Whitney work of 1903 38734°-SM 1911-5

to 1910, and have been made with many different optical systems. There is great difficulty in getting an accurate estimate of the relative losses suffered by rays of different wave lengths in traversing the spectroscope. Especially is this the case for the violet and ultra-violet rays, where these losses are greatest. The summary has shown that further determinations are needed to fix the distribution in the extreme ultra violet, and observations for this purpose were made in June, 1911, on Mount Wilson, but are not yet reduced. I give below the summary, excluding the work of 1911.

Intensities in normal solar spectrum, outside the atmosphere.

[Observed at Washington, Mount Wilson, Mount Whitney, 1903-1910.J

The sun's temperature.-If we employ the so-called "Wien displacement formula," which connects the absolute temperature of a perfect radiation with the wave length of its maximum radiation, we may proceed as follows, to estimate the solar temperature, on the assumption that the sun is a perfect radiator:

AmaxT=2930.

If Amax=0.470 μ then T-6230° abs. C.

Another radiation formula is that of Stefan, which connects the total quantity of radiation of a perfect radiator per square centimeter per minute with the absolute temperature. Employing this formula, still assuming the sun to be a perfect radiator, its mean distance 149,560,000 kilometers, its mean diameter 696,000 kilometers, and the mean value of the solar constant of radiation 1.922 calories per square centimeter per minute, we proceed as follows:

2
696,000

76.8X1
8X10-12X149,560,000,

T4=1.922 T=5830° abs. C.

A third means of estimating the sun's probable temperature comes from comparisons of the distribution of the energy in its spectrum with that in the spectrum of the perfect radiator, as computed according to the Wien-Planck formula of spectrum energy distribution. The sun's energy curve and that of the perfect radiator at two temperatures are given in the accompanying illustration (fig. 2). It appears at once from this comparison that the sun's radiation differs greatly from that of the perfect radiator at any temperature. The solar radiation is greater in the infra-red spectrum, and much less in the ultra-violet spectrum, than that of perfect radiators giving approximately the same relative spectral distribution as the sun for visible rays. Taking everything in consideration, the solar energy spectrum seems most comparable with that of a perfect radiator between 6,000° and 7,000° in absolute temperature. The causes of the discrepancies we have noted may be several. First, there is the influence of the selective absorption of rays in the Fraunhofer lines.

effect of this difference may each be twofold: For, firstly, at the center of the sun's visible disk we look probably to deeper-lying and hence hotter layers than at the sun's edge, where the line of sight is oblique; and, secondly, since the transmission of the sun's atmosphere is probably like the earth's,

These lines are much crowded toward and within the ultra-violet spectrum, so that perhaps this indicates a principal reason for the weakness of the sun's spectrum in that region. Second, it seems probable that we are dealing with a mixture of rays from sources at different temperatures. The cause and

1.6

1.8

2.0

2.2

2.4

2.8

FIG. 2.-Energy spectrum of sun compared with black body.

0.0

0.2

0.4

0.6

0.8

1.0

1.2

much less for violet and ultra-violet rays than for red and infra-red ones, we probably get infra-red rays from deeper-lying and hence hotter layers in the sun than we do ultra-violet ones.

We conclude that the solar radiation comes from sources ranging in temperature perhaps between the limits 5,000° and 7,000° absolute centigrade, but mostly from sources between 6,000° and 7,000°.

Washington observations. Further experiments have been made, under Mr. Fowle's direction, on the transmission of radiation of great wave lengths through long columns of air containing known qualities of water vapor. Many of these observations are not yet reduced, so that it is not yet proper to give a numerical summary of results. The length of the column experimented upon has been increased to 800 feet. The measurements cover the infra-red spectrum, from the A line to a wave length of about 17μ. The observations of the water contents of the air column are made by means of pairs of wet and dry thermometers located at a number of points along the path. The air is thoroughly stirred before readings. Check experiments by Mr. Aldrich, in which he drew the air through phosphorus pentoxide tubes and weighed the water absorbed, have confirmed the accuracy of the water-vapor determinations. Mr. Fowle has made a preliminary comparison of the upper infra-red spectrum bands p, σ, 7, 1, ¥, and ?, as observed through the tube with the same bands as observed through the whole atmosphere at Washington, Mount Wilson, and Mount Whitney. The results are most interesting, though not yet ripe for publication, and will probably lead to more exact knowledge of the total quantity of water vapor in the atmosphere, and its variation with the altitude of the observer and the season of the year.

FIG. 3.-Abbot silver disk pyrheliometer.

Reduction of observations.— Upward of 100 days of solarconstant measurements have

been made on Mount Wilson on each of the last several years. Each day requires the equiva lent of three full days of computation. This work is being done at Washington by Messrs. Fowle and Aldrich and Miss Graves and certain graphical parts of it by minor clerk Segal. The solar-constant reductions are computed as far as the middle of the observing season of 1910. Pyrheliometry.—Additional comparisons of the Mount Wilson secondary pyrheliometers have been made with primary standard pyrheliometer No. 3. These are not yet all reduced, but such as have been finished confirm the results of the previous fiscal year, so that we may regard the scale of absolute pyrheli