William A. Rogers may be mentioned. But most of it is in the depart ment of astronomical physics, and this it is to which the committee desires to direct attention.

The work in astronomical physics, which has been done in the observatory of Harvard College under Professor Pickering's imme diate supervision, seems readily divisible into three classes:

(1) Stellar Photometry; (2) Stellar Photography; and (3) Stellar Spectrum Photography.

I. The principal subdivisions of the photometric work are as follows:

(1) Observations of Jupiter's satellites at their disappearances in the shadow of the planet, and their reappearances after the eclipse, with the object of determining the times at which these phenomena occur, with greater accuracy than had hitherto been practicable.

(2) The determination of the relative brightness of all stars visible to the naked eye in the latitude of Cambridge.

(3) The determination of the relative brightness of several thousand stars of the first nine magnitudes, in order to fix the value of the arbitrary scales usually employed in estimating stellar magnitudes.

(4) Comparisons of various objects at small apparent distances from each other, such as the components of double stars, different satellites of the same planet, etc.

(5) Numerous special investigations on proposed standards of magnitude, stars which have been used for comparision with variable stars, the light of nebulæ, etc.

All this photometric work has gone uninterruptedly forward with the aid of special photometers ingeniously devised for the purpose by Professor Pickering.

The observation of the eclipse of Jupiter's satellities has long been regarded as important, not only in determining the motions of the satellities themselves, but also in ascertaining the velocity of light. It is obvious that the observation of the disappearance or reappearance of a satellite is necessarily a photometric observation, and, therefore, that the subject is one to which improved means of photometric research are peculiarly applicable. But previous to Professor Pickering's undertaking, no one appears to have recognized the advantages resulting from the use of the photometric apparatus in the observation of these eclipses. Recently, however, these advantages have been recognized in France, and researches similar in character to those of Professor Pickering have been undertaken there.

It is not to be expected that the observations of the eclipses made at the Harvard College Observatory will be fully reduced and published until they have been continued through a complete revolution of Jupiter (about twelve years). The first observation of this sort was made on the 23d of June, 1878. Up to the beginning of the present year, the total number of eclipses observed was three hundred and fifty-eight, thirty

nine of which had occurred since October, 1885. The accuracy with which an eclipse of the first satellite can be observed appears from the preliminary reductions to be very satisfactory, the probable error seeming to be less than a second of time. The eclipses of the three outer satellites will furnish results nearly as useful, since the greater number of comparisons obtained during each eclipse compensates to a considerable extent for the comparative slowness of the changes in the brightness of the object observed.

The meridian photometer of Professor Pickering was first brought into use in the summer of 1879. The work of determining by the aid of this instrument the relative brightness of the stars which are visible to the naked eye was begun on the 25th of October of that year, and was continued through seven hundred series of observations, four thousand two hundred and sixty stars being included in the catalogue. The total number of comparisons made was ninety-four thousand four hundred and seventy-six, four comparisons constituting a complete observation. These results, together with an extended discussion of the estimated magnitudes obtained for the same stars by careful observers, as well as the results of previous photometric determinations of their relative brightness, were published in Volume XIV of the Annals of the Observatory.

The success of this photometer led Professor Pickering to have a larger instrument constructed, of the same kind, but which could be applied to the measurement of telescopic stars. The new photom eter has given perfect satisfaction, and has enabled the brightness of stars to the ninth magnitude to be measured with a deviation from accuracy not exceeding a tenth of a magnitude. Moreover, stars can be measured with this instrument at the rate of forty stars an hour for an entire evening; and, under favorable conditions, even one star a minute may be thus measured. During the year 1886 two hundred and nine. series of measurements were made by Professor Pickering aud his assistant, the total number of separate photometric comparisons being fifty-nine thousand eight hundred. In 1885 fifty thousand such comparisons were made, and in 1884 twenty-seven thousand five hundred. A comparison of the Cambridge results with those obtained at Pulkowa shows that the average deviation of a measurement of the difference in brightness between two stars observed at both places does not exceed one tenth of a magnitude. The principal work of the meridian photometer, the revision of the Durchmusterung magnitudes, is now approaching completion, nine tenths of the observations having already been made.

It would seem to be manifest from the facts above enumerated that practical stellar photometry has now been carried on for the first time by Professor Pickering to that extent and with that completeness which are requisite to give it general interest and importance.

II. The earliest investigations in stellar photography, which were

undertaken by Professor Pickering, were begun in 1882, with the aid of a grant from the Rumford fund of the American Academy. They have been continued subsequently by an appropriation made in June, 1885, from the Bache fund of the National Academy. The work which has been done in this direction has been summarized as follows:

The present research was undertaken in 1882, with a lens having an aperture of only 24 inches. It was shown that photography could be used as a means of forming charts of large portions of the sky, and of determining the light and color of stars in all portions of the heavens. Photographs of the trails of close polar stars no brighter than the eleventh magnitude were obtained without clock-work. In 1885 the investigation was resumed with a telescope having an aperture of 8 inches. With this, one hundred and seventeen stars within one degree of the pole, one of them no brighter than the fourteenth magnitude, left trails. The average deviation of the measures of the brightness of these stars on different photographs was less than a tenth of a magnitude; a greater accordance than is given by any other photographic method. A similar result was obtained from the Pleiades, of which group over fifty left trails. Similar trails are now being obtained of the stars north of -30° in all right ascensions. This work began in the autumn of 1885, at 23h, and has already been completed for more than half of the sky. By photographing on the same plate polar stars near their upper and lower culminations, material has been accumulated for determining the atmospheric absorption each night of observation.

A study has been made of the application of photography to the transit instrument. Measurements of the trails show that the time of a star's transit may be determined from its trail, with an average deviation of 0.03 second, which is about half the corresponding deviation of eye observations.

Charts may be constructed 5° square, having the same dimensions as those of Peters and Chacornac. A single exposure of one hour is required, and it is not necessary that the observer should remain with his eye at the telescope to correct the errors of the clock. Miscellaneous observations have been secured of the Pleiades, of the nebula in Orion, of Jupiter's satellites, and of various other objects.

III. Stellar spectrum photography, in which Professor Pickering has achieved his most marvelous results, was undertaken by him first in 1885, upon a plan entirely novel. Two important features of advantage were possessed by the new arrangement. The first consisted in placing the prism in front of the object-glass; a plan already suggested by Fraunhofer, and tried for eye observations by Secchi. The second consisted in placing the refracting edge of the prism parallel to the motion of the star; i. e., if the star were on the meridian, the refracting edge would be horizontal, and the spectrum north and south. The first device gives not only an enormous increase of light, but renders it possible for the stars over the entire field to impress their spectra upon the plate. Hence, while previous observers have been able to photograph but one star at a time, and have not obtained satisfactory results from stars fainter than the second or third magnitude, Professor Pickering has often obtained more than a hundred spectra on a single plate, many of them of stars not brighter than the seventh or eighth magnitude. By means of the second device, the necessary width of the spectrum is secured. If clock-work were employed, the spectrum of a star would be a line too narrow to show any details. Formerly, a cylindrical lens

was used to widen the spectrum; but by placing the prism as above, and by suitably varying the rate of the clock, any desired width can be given to the spectrum, simply by the motion of the star itself. One millimeter was found to be the best width.

The first experiments were made in May, 1885, by placing a 300 prism in front of a lens of 24 inches aperture. No clock-work was used, the spectra being formed of the trails of the stars. In the spectrum of the pole star over a dozen lines could be counted, and in that of a Lyræ the characteristic lines were shown very clearly with an exposure of only two or three minutes. In the fall of 1885, two prisms were obtained 20 centimeters in clear aperture, and having angles of 50 and 15° respectively. These could be placed in front of the object-glass of the 8-inch photographic telescope. With this apparatus, systematic operations were begun. The research, however, proved too large for the Bache appropriation. Whereupon, Mrs. Henry Draper came forward early in 1886, and generously offered not only to defray the expenses of the investigation as a memorial to Dr. Draper, but to put at Professor Pickering's disposal the 11-inch contractor, with its photographic corrector, with which Dr. Draper had taken many of his later stellar spectrum photographs. The investigation was thereupon extended by Professor Pickering and divided into three parts. The first included a general survey of stellar spectra, each spectrum being photographed with an exposure of not less than five minutes. These photographs generally exhibited the spectra of all stars brighter than the sixth magnitude with sufficient distinctness for measurement. The second research related to a determination of the spectra of the fainter stars. Each photograph taken received an exposure of one hour, so that the spectra of all the stars not fainter than the eighth or ninth magnitude, and included in a region ten degrees square, were represented upon the plate. For both these investigations, the 8 inch Bache photographic telescope was employed. The third research related to a more detailed study of the spectra of the brightest stars. For this work, the Henry Draper 11-inch corrected refractor was used, Mrs. Draper having provided a small observatory for it in Cambridge, and also the mounting. Four prisms were constructed, each of 159, of which, three had a clear aperture of nearly 11 inches, the fourth being somewhat smaller, the whole weighing more than a hundred pounds, and occupying a cubic foot of space.

The results already attained in these three directions have been satisfactory. In the first research, the first cycle covering the entire sky from zero to twenty-four hours of right ascension has been completed. This cycle contains two hundred and fifty-seven plates, all of which have been measured and a large part of the reduction completed. Eight thousand three hundred and thirteen spectra have been measured on them, nearly all of which have been identified, and the places of a greater portion of the stars brought forward to the year 1900 and en

tered in catalogue form. In the second cycle, which will complete the work, sixty-four plates have been taken, of which fifty-one, including two thousand nine hundred and seventy-four spectra have been meas. ured and identified. In the second research, which has been carried on in the intervals when the telescope was not needed for other purposes, ninety-nine plates, with the exposure of an hour, have been obtained, on which four thousand four hundred and forty-two spectra have been measured; in one case, over three hundred spectra being shown on a single plate. In both these researches, taken together, fif teen thousand seven hundred and twenty-nine spectra of bright and faint stars have been measured. In the third research, the first results of which were obtained in October last, a considerable number of excellent photographs have been obtained, the results of which can be stated only after a long series of measurements and a careful reduction and discussion of them. Some points of interest, however, are shown on inspection. A photograph of a Cygni, taken November 26, 1886, shows that the H line is double, its two components having a difference in wave-length of about one ten-millionth of a millimeter.* A photograph of Omicron Ceti shows that the lines G and H are bright, as are also four of the ultra-violet lines, characteristic of spectra of the first type. The H and K lines in this spectrum are dark, showing that they probably do not belong to that series of lines. The star near X' Orionis, discovered by Gore in December, 1885, gives a similar spectrum, which affords additional evidence that it is a variable of the same class as O. Ceti. Spectra of Sirius show a large number of faint lines beside the well-known broad lines.

Professor Pickering has also devised a new method of enlarging these original photographs, which also removes the inequalities in density in the negatives, appearing as bands parallel to the length of the spectrum. A cylindrical lens is placed close to the enlarging lens, with its axis perpendicular to the spectrum lines. In this way the length of the spectrum is increased only five times, while its width is increased nearly one hundred, i. e., from one millimeter to about four inches. This arrangement greatly reduces the difficulty arising from the feeble light of the star.

Until very lately, the spectra in the original negatives were made very narrow, since otherwise, the intensity of the starlight would have been jusufficient to produce the proper decomposition of the silver particles. The enlargement being made by daylight, the vast amount of energy then available is controlled by the original negative, the action of which may be compared to that of a telegraphic relay. The copies, therefore, represent many hundred times the original energy received from the stars.

The deed of trust under which this committee is acting provides "that

*It has long been known that the H line is also double (bright) in the spectrum of the solar chromosphere.