3° 23. The velocity, at present, seems to be included between —8.6 and —14.6 km, having an extreme range of only 6 km. The velocity of the binary system seems to be about 12 km.

The determinations of velocity made in 1896 lie entirely outside of the present range of values, and leave no doubt that the

velocity of the binary system is changing under the influence of an additional disturbing force. I think it is certain, therefore, that Polaris is at least a triple system.

The 1896 observations were made at intervals differing but little from multiples of the period of the binary system, and therefore fell near the same point in the velocity curve. Assuming a period of 3d 23, there is no difficulty in selecting the epoch of minimum so that these six observations will fall on the

curve satisfying the 1899 observations. The residuals will be negligible if we assume the observations to fall near the lower part of the curve, as in Fig. 3; and I have no doubt that future determinations of the orbit will definitely place them there. It will be seen, on comparing Figs. 2 and 3, that the velocities of the binary system in 1896 and in 1899 differ about 6 km.

LICK OBSERVATORY,

Sept. 1, 1899.

THE VARIABLE VELOCITY OF POLARIS.

By EDWIN B. FROST.

POLARIS has not been on the regular working list of stars whose velocities are to be determined at the Yerkes Observatory; but in view of the interest in Professor Campbell's important discovery, it seems desirable to give at once the results of the measurements I have just made on three plates recently secured, which confirm the short period variation in the velocity.

The first plate was taken under unfavorable circumstances, with a short camera of 271 mm focus, and would be called a poor plate. The velocity is based upon the displacements of six star lines, titanium being used for comparison. The result, however, may be expected to be reliable within three kilometers.

The second plate is an excellent one, taken with a camera of 456mm focus. The probable error of the above mean of the determinations from 12 lines is 0.59 km. The third plate is equally good. The probable error of the mean of the 16 lines measured is± 0.48 km. The three plates were photographed by

Mr. Ellerman.

On locating the second observation on Professor Campbell's curve by reckoning forward with a period of 3d 23h from his minimum of 15.4km on August 24, it appears that the observation (Sept. 20) falls within a little over an hour of a minimum. It will be seen that the third plate also falls close to its expected position on the curve.

A fourth plate, secured, as was the third, while this was passing through the press, and not yet measured, appears on inspection to give a result accordant with the curve.

The range of variation, 7 km, is within a kilometer of that found by Professor Campbell.

Of course my observations cannot give testimony on the question of a long period motion in the system of Polaris.

YERKES OBSERVATORY,

Sept. 26, 1899.

THE WAVE-LENGTH OF THE GREEN CORONAL LINE, AND OTHER DATA RESULTING FROM AN ATTEMPT TO DETERMINE THE LAW OF ROTATION OF THE SOLAR CORONA.'

By W. W. CAMPBELL.

A DETERMINATION of the law of rotation of the solar corona would no doubt be valuable on many accounts. Interest in this problem was aroused by Deslandres' attempt to solve it at the 1893 eclipse, in Senegal. The reality of his result has been questioned on the ground that the H and K calcium lines, used by him, do not have their origin in the corona, but in the prominences and chromosphere. It seemed proper that another determination should be attempted, basing it upon light radiations which are of unquestioned coronal origin. Accordingly, as one of many problems, it was undertaken by the Crocker expedition sent out from the Lick Observatory to observe the eclipse of January 22, 1898, in India.

It was evident that this investigation, involving the DopplerFizeau principle, and requiring high dispersion, could apply, with any hope of success, only to the bright-line portion of the corona. Existing data seemed to show that the green line and those near XX 423 and 399 were the only reasonably strong and unquestioned coronal lines in the available portion of the spectrum; and further, that the photographic action of the green radiation was vastly stronger than in the case of the other two lines, even though the green line lay in a region of weakness on isochromatic plates. It was therefore decided to base the observations on the green line. The justification of this decision Read at the Third Conference of Astronomers and Astrophysicists, September 7, 1899.

The expenses of the expedition were defrayed by the late Hon. C. F. Crocker, a regent of the University of California. He was also the patron of the expeditions to Cayenne (1889) and Japan (1896).