REPOKT OF THE SECRETARY. 87 



tend to see if the sun is const^it or variable in its output of radiation. While the 

 results of the past year indicate possibility of considerable ultimate success in this 

 final object, it may probably prove that the situation of this Observatory is vmsuita- 

 ble t(j the most exact studies of this kind, owing to the great and variable absorj^ 

 tion of the air column above it, as well, of course, as to the ground tremors inherent 

 in its present site, surrounded by the traffic of city streets. It seems certain, how- 

 ever, that ^'aluable contributions to the methods of study to be adopted in work at 

 more favorably situated stations, and to the knowledge of terrestrial absorption, can 

 be made here. 



Alterations of the cqyparatus. — For the sake of accuracy in the height of the curves, 

 the speed of taking energy curves has been much increased. Thus we now custom- 

 arily ado])t the speed ratios 2 centimeters of plate and 20 minutes of arc of spectrum 

 in 1 minute of time, while the ratio 1 cm.= V = Im. was the standard one adopted 

 for the detailed study of the infra-red si>ectrum. At the present speed the spectrum 

 between wave lengths 0.45 /< in the blue and 2.5 /< in the infra-red is passed over in 

 fifteen minutes. Such a spectrum energy curve is shown in Plate VIII. At this 

 fast speed much less opportunity offers for "drift," alteration of sensitiveness of the 

 galvanometer, and, what is far more commoix, for clouds or haze to diminish the 

 accuracy of the work. In order to keep the behavior of the apparatus as constant 

 as jjossible and to get the best jiossible detail at tins rapid sj^eed, the great glass prism 

 has been empLjyed instead of rock salt, thus limiting the research to wave lengths 

 less than 3 //. But this includes nearly all the solar radiation which reaches the sea 

 level. To avoid extra reflecting and absorbing surfaces as far as practicable, a single 

 mirror is used for collimation of the spectroscope in place of the two employed for- 

 merly when great detail was sought, and the bolometer case is used oj^en at the end, 

 although a slight amount of accidental disturbance of the record is involved on 

 windy days. In the interest of the greatest detail practicable at the fast speed, it 

 was most desirable to shorten the time of swing of the galvanometer. A new gal- 

 vanometer of most approved design has been constructed and has been in use since 

 Decend^er, at an average time of single swing of 1.5 seconds. This type of galva- 

 nometer is in some respects, I think, new and worth a brief description, which will be 

 found on a later page. To avoid errors from differences of diffraction from the edges 

 of the slit, the slit width has been maintained unaltered at 0.25 mm., subtending 

 10.5 seconds of arc in the spectrum. Its height, 5 centimeters, has been virtually 

 reduced to 2 centimeters in the infra-red portion of the spectrum by the expedient 

 of interposing across it at the proper time a brass grill-shaped diaphragm, which, 

 being fixed upon pivot bearings on the slit mounting, always occupies tlie same 

 l)lace when interposed before the slit. The factor of this diajihragm can be meaa- 

 ureil by the aid of data found on each curve, but the results are constant within the 

 limits of measurement. Bolometer No. 20, of 0.08 millimeter width and snl)tend- 

 ing 10 seconds of arc, has been used for the entire year. This bolometer has been in use 

 nearly all the time since it was made in 1896, and alwa\s witli great freedom from 

 drift and other disturbance. Soon ■aHvy my return from Sumatra I suggested, and 

 you caused to be constructed at the Observatory shop, a more improved type of 

 mechanical rheostat for the bolometer, which lias been used since with great satis- 

 faction. Drift of the galvanometer spot now rarely reaches 4 centimeters in a day, 

 and it is rarely in excess of 2 millimeters upon an energy curve taken in fifteen 

 minutes. So firm is confidence in the behavior of the bolometer that the unshunted 

 galvanometer is left in connection with the circuit weeks at a time. No rebalancing 

 of the circuit is required for a month at a time. One can now prejiare for holo- 

 graphic work one day and the next merely start the siderostat and the driving clock 

 with practical certainty of securing a good holograph. 



Tlie absorption of the ajyparatus. — In order to know the true distribution of the 

 radiations throughout the spectrum, it has been necessary to determine the absorp- 

 tion of the spectroscope. As mirror surfaces deteriorate, this has to be done not 



