both extremes in the scale of luminosity. But the orbits assumed for y Leonis and 61 Cygni were probably entirely wide of the mark; and omitting these, they are certainly less luminous than the Sirians; but their position as regards the Capellans is more difficult to define. In fact, there are not enough of them to settle it satisfactorily. There appear to be as many Arcturian as Capellan double-stars, but among known binaries the latter are five or six times as numerous. In like manner, Sirian double-stars are numerous enough, but only a small proportion are known to be binaries. That the vast majority of double-stars are binaries, a careful examination will, I think, lead us to conclude; but good observations on their positions being all of comparatively recent date, we are still without means of proving this fact when the period is very long. Now, ceteris paribus, the greater the mass of the pair, the shorter will be the time of revolution; and the shorter periods of revolution in the case of the Capellan doublestars go far to displace the theory that these stars are unusually small. Indeed, unless they are nearer to us than the others, we should be driven to conclude that their masses were greater than those of either the Sirians or the Arcturians. Capellan stars will, I believe, prove not to be the smallest, but the least luminous class of stars-least luminous, at least, relative to their density; for there are some reasons for thinking that a Sirian star, instead of presenting a much brighter surface than a Capellan of the same mass, presents a much greater extent of surface. The Capellan is rather denser than duller; but, mass for mass, it gives much less light.

These results may not be inconsistent with a theory of stellar development, but if so, it must assume a different form from that which would naturally occur to us. If the Sirian, Capellan, and Arcturian stars represent different stages of stellar development, dependent on cooling and condensation, we must place the Capellan stage last, not second. And if stars pass through what is called the third type of spectrum (designated by M in the DRAPER catalogue), they must do so before reaching the Capellan stage. Very few of the stars with this kind of spectrum possess large proper-motion. The number which I have identified in the PULKOVA catalogue is under fifty, some of which are queried; but only four of these have a proper-motion in Declination of over o'.1, thus giving a smaller percentage than even the Arcturians. Unless they are, on the average, larger stars than the

Capellans, and we would not be likely to find the largest stars in the last stage of cooling-they are evidently more luminous, relative to their density, and must, therefore, represent an earlier stage of evolution.

The stars of the Orion type, designated B in the DRAPER catalogue, have less proper-motion than even the ordinary Sirians, denoted by A. On the development theory, they represent the earliest stage, while the Capellans represent the latest. If we cannot represent the development as taking place in this order, the stars must differ in kind—probably composed of different elements. One circumstance which rather favors the theory of development is, that every star of the Algol type whose spectrum I have succeeded in identifying-including those like Spica Virginis and ẞ Auriga, where no actual eclipse at present takes place are Sirians. On the tidal theory of satellite development, the Algol type of stars belong to the earliest stage, their satellites ultimately developing into the binary systems with moderate periods, which are so frequently found among the the Capellans. The duration of the eclipses in the case of these Algol stars, compared with the intervals between them, indicate. a low density for these stars; and this character may not improbably belong to all the Sirian stars. It is possible that, as they become condensed by the cooling process, and their satellites driven to a greater distance by tidal action, they may become Capellans. But, then, where do the Arcturians come in? I can offer no plausible theory on this subject.

In conclusion, I desire to point out that if Sirian stars are, ceteris paribus, visible at a much greater distance than Arcturians, and the latter visible at a greater distance than Capellans, we must not regard the DRAPER catalogue as indicating the relative frequency with which these classes of stars occur in space. Supposing that the Sirians and Capellans were equally numerous, and that both were uniformly distributed, but that the former were (on the average) visible at double the distance of the latter, we might expect to find eight times as many Sirians as Capellans in the DRAPER catalogue. The actual proportion is about 21⁄2 to 1; and the natural inference appears to be that the Capellan stars are really more numerous than the Sirians. The apparent numerical superiority of the latter is probably a delusive appearance arising from their greater luminosity. And it is evident that in the case of a distant cluster, a number of Sirian stars may

be distinctly visible when the light of their Capellan companions is too faint to enable them to be observed separately. This fact should not be lost sight of in speculations relative to the structure of the Galaxy. Certainly, among the nearer stars which are in the same direction as the Galaxy, many (including the famous a Centauri and 61 Cygni) are of the solar type. Is there any valid reason for regarding the more distant stars in this direction as almost exclusively Sirian? I think not.

THE SPECTRA AND PROPER-MOTION OF STARS. [SUPPLEMENTAL NOTE.]

By W. H. S. MONCK.

Having obtained the spectra and proper-motion of a larger number of stars than I had hitherto done by a comparison of the British Association catalogue with the DRAPER catalogue, I thought it desirable to take the different sub-classes into which Professor PICKERING divides the stellar spectra separately. I found a sufficient number of stars with the following spectra to render a comparison feasible; viz. A, B, E, F, G, H, I, K, and M. I compared the proper-motions of the stars of these types (rejecting, in the first instance, all those marked with a note of interrogation in the DRAPER catalogue) in Declination or North. Polar Distance, only ascertaining what proportion of them had a proper-motion of one tenth of a second annually in this direction. All my percentages are somewhat too small, because the divisor included some stars whose proper-motion is not given in the British Association catalogue, but whose spectra I copied into my note-book, with a view of subsequently ascertaining their propermotions from some other source, which I have not yet done. The percentages which I obtained were as follows:

The stars of this last type were not numerous enough to justify the conclusion that the proper-motion is really greater than those of the types E and F.

While the stars classed as A gave a percentage of 8.8, those marked A? (of which there were over 130) gave a percentage of 12.8. This was to be expected, as the intrusion of stars of any other type (except B) would increase the average proper-motion. On the other hand, the stars marked F? gave only 22.6 per cent. and those G? 25.0; but those marked E? gave the high percentage of 36.8. Classing E, F, and G, however, together as Capellan stars, the unqueried Capellans gave a percentage of 31.8, and the queried Capellans a percentage of 28.9, thus confirming the result that the intrusion of stars of any other type among the Capellans will reduce their average proper-motion. H? gave a percentage of 12.6, as compared with 16.0 for H; but I? and K? gave 24.4 and 29.0 as compared with 18.9 and 22.6 for unqueried stars of the same types. On the whole, if we designate the types H, I, and K as Arcturian, the queried Arcturian stars gave a little more proper-motion than the unqueried. The reverse is true of the type M. The extremely low proper-motion of the stars of the Orion type B is remarkable. I had seventyfive of them to compare, the proper-motions of seventy-two being known. Not one of these had a proper-motion of one tenth of a second in N. P. D., while out of twenty-seven stars of the type G, no less than eleven possessed it. It will be seen that the superior proper-motion of the Capellan stars over the Arcturian (with which stars of the type M may be classed) is quite as strongly marked as that of the Arcturian over the Sirian.

This difference of proper-motion will, I think, be found to arise not from the greater actual velocity, but from the greater nearness (on the average) of the Capellan stars. The entire subject, however, calls for further investigation.

THE SAYRE OBSERVATORY, SOUTH BETHLEHEM,

PENNSYLVANIA.

By C. L. Doolittle.

On the first day of September, 1866, occurred the formal opening of the Lehigh University, at South Bethlehem, Pennsylvania. Professor ALFRED M. MAYER, now connected with the STEVENS Institute of Technology, was the first professor of astronomy, which department was then united with that of physics.

Professor MAYER felt that the attempt to teach astronomy with no instrumental means was most unsatisfactory, and mainly through his efforts, Mr. ROBERT H. SAYRE, one of the most prominent men of the place, and a trustee of the University, became interested in the matter, and undertook to provide means or a small observatory. The result was the founding of the SAYRE Observatory, in 1868.

The plan was not an ambitious one, the total cost of building and instruments being only about $5000.00.

The equipment consisted of a six-inch equatorial, by ALVAN CLARK & SONS; a portable transit instrument, by STACKPOLE; a sidereal clock, by BRAND; and, finally, an old zenith telescope, which is said to have been rejected by the U. S. Coast Survey, and which was purchased for a small sum, economy being a matter of necessity. It was with this instrument, repaired by KAHLER in 1875, and afterward by SAGMÜLLER in 1888 and 1892, that a series of latitude observations was carried on by the writer for a number of years, the results of which, it is believed, have been of some service in determining what is now known of the law of latitude variation.

Professor MAYER was more interested in physics than in astronomy, yet he found time for research work at the observatory, mainly in the direction of solar physics. He was succeeded in 1871 by HIERO B. HERR, the chair of astronomy being now connected with that of mathematics. Professor HERR retired in 1874, the vacancy being filled one year later by the election of C. L. DOOLITTLE, who remained in charge for a period of twenty years. He was succeeded, in 1895, by C. L. THORNBURG, the present incumbent.

The original purpose of the observatory was that of instruc