72 REPORT OF THE SECRETARY. 



Some decrease in the accidental defections of the "battery record" was effected 

 by the substitution of a "Cupron" battery of 10 cells for tbe great storage battery 

 of 60 cells. 



A new design for a bolometer has been prepared, in which the sensitive threads, 

 balancing coils, and adjusting slide wire are all contained in one compact water- 

 jacketed case with air-tight chamber. This instrument is under construction, and 

 will, it is believed, do away with certain sources of accidental disturbance, and will 

 be far more easy to use satisfactorily than the present form. 



With the improvements above described all or in part installed, there were taken 

 135 holographs between December 1, 1897, and July 1, 1898. Of these 68 were with 

 the great rock-salt prism, 41 with the great glass prism, and 26 with a small prism 

 of fluorite. Fifteen of those taken with tbe rock-salt and 8 taken with tbe glass 

 prism were measured upon the comparator to establish the discovery and positions of 

 the 700 absorption lines in the infra-red already mentioned. Fifteen taken with the 

 fluorite prism were measured upon the comparator to determine the position of about 

 50 absolution lines identifiable on both the rock-salt-prism and the fluorite-prism 

 holographs, with the design of thus deducing the dispersion of rock-salt indirectly 

 from the wave-length determinations of Paschen in the fluorite spectrum. 



Apparatus, including a concave grating, has been arranged for the purpose of 

 directly measuring the dispersion of rock salt, but the actual observations were 

 not begun at the close of the period covered by this report. It is hoped that this 

 research will make it possible to give the wave lengths of the absorption lines dis- 

 covered to the degree of accuracy corresponding with that of the prismatic devia- 

 tions. The Observatory is peculiarly fitted to obtain results of great accuracy, in 

 that, first, it is in possession of such an extraordinary equipment of rock-salt prisms 

 that one great one is provided with a thermometer at its center and used solely to 

 determine the temperature of the optical one; second, a constant temperature may 

 be maintained, and hence the temperature of tbe salt can be certainly known; 

 third, the great sensitiveness of the bolometric apparatus allows of the employment 

 of narrow slit widths; fourth, the holographic method can be employed, which, 

 being independent of circle readings, and involving instead a clockwork of extreme 

 accuracy, gives differences of deviation with extraordinary precision, reaching, as 

 we said in last year's report, to within a second of arc. 



Several energy curves, extending from the violet through the visible and infra-red 

 spectrum as far as 5 /(, were taken with a sheet of bright copper in place of the sil- 

 vered glass mirror at the siderostat. It was found that there was no appreciable 

 difference in quality or amount of reflecting power between the copper and silver 

 surfaces, except in the violet. Here the copper gradually deteriorated as a reflector, 

 which accounts for its red color. 



Observations have been made with the "hot box," a device similar to the garden- 

 er's hot bed intended to obtain a very high temperature from the sun's rays without 

 the use of lenses or mirrors. A temperature of 120° C. was obtained, which is, to 

 be sure, considerably above boiling water, but not in excess of that obtained by 

 Herschel with the same device in South Africa many years ago. The results of the 

 observations are merely tentative. 



A considerable number of observations have been made to determine the accuracy 

 of the bolometer as a heat-measuring device; that is, its capacity for repeating the 

 same measure of radiations under like conditions. For this purpose successive first 

 throws of the galvanometer were observed when the radiations from a student lamp 

 burning good-kerosene oil w r ere alternately allowed to fall on the bolometer and cut 

 off by a water screen at constant temperature. The variations in the deflection 

 were very slight, and indicated rather a variation of the burning of the lamp than 

 any inaccuracy of the bolometer. Thus for ten successive measures the average 

 probable error _of the separate observations was only 0.035 of 1 per cent, or 1 part in 

 3,000; but, as has just been intimated, this is a maximum value, since no absolutely 



