REPORT OF THE SECRETARY. 71 



;m; at a point a little nearer the back of the prism than the center of the faces, and 

 this was done in practice. However, in determining absolute quantities like the 

 index of refraction at A, it was found necessary to be far more particular. 



Measurement of refractive indices at A. — Diaphragms just wide enough for visual 

 resolution of A were placed symmetrically on the faces, and the prism was so placed 

 on the prism table that exactly the same beam of light entered the diaphragm in one 

 j)osition (if the prism as in the other, in measuring l)oth the angle and the minimum 

 deviation. In this way very excellent accord was obtained between several series of 

 measures, and the following constants were fixed for the refractive indices of rock 

 salt and fluorite in air at 20° C. and 7H0 mm. pressure. Average wave length of 

 radiations, 0.7G04/<. 



For rock salt n = 1.536818 ± .000009 



For fluorite n = 1 .431020 ± .000006 



Do all rock-salt prisms have the same dispersion? We were led to believe the affirma- 

 tive upon this very important question by recorded results from many prisms, but 

 we have conclusive evidence in the following comparison of the dispersion of three 

 salt prisms, two from Russian and one from Bavarian salt, between wave lengths 0.4 ja 

 and 4.0 yK. The results indicated the affirmative, for the differences in the refractive 

 indices in all this range never exceeded the probable experimental error of determi- 

 nation. To be more precise, the results at A were as follows: 



Prism R. B. I. n = 1.536818 ± .000009 

 Prism R. B. II. n = 1.536844 ± .000006 

 Prism S. P. L. T. n = 1.536812 ± .000005 



At other points the differences were of the same order of magnitude. 



It follows then, as you have anticipated and elsewhere pointed out, that this most 

 interesting crystal, whose optical application from the time of Melloni to the com- 

 mencement of these observations has been chiefly qualitative as a transmitter of 

 special radiations, can now be used quantitatively with practical convenience in the 

 form of a 60° prism, as a standard of refraction to which all wave lengths may be 

 referred with the same order of precision as to the grating. 



Tlie temprrniure coefficient of refractive indices for rock-salt prisms. — Bolographs were 

 taken at low and high constant temperatures, and from these in connection with former 

 results the temperature coefficient for the whole range of radiations covered by our 

 bolographs was accurately determined. 



Comjmrison of the efficiency of the bolometer and thermopile. — It will be recalled that 

 the thermopile has recently been made far more delicate and efficient In' improve- 

 ments of Rubens, so that with him and with some others it has displaced the bolo- 

 meter for radiation work. A comparison made here between one of these instru- 

 ments and our bolometer, No. 20, shows the latter, though of only one-tifteenth the 

 surface, to give twice the deflection at the galvanometer when substituted for the 

 thermopile. The galvanometer was besides more free from "drift" and "wiggle" 

 with the bolometer, and there was no "creep" to the deflection with it, while such 

 "creep " lasted 5 or 10 seconds with the thermopile. The bolometer has besides the 

 advantage that it can be made more strictly linear and far narrower than the thermo- 

 pile, and is capable of exact setting in the spectrum. To offset these advantages, the 

 thermopile requires no battery or balancing coils, and costsbut about one-thirtieth as 

 much as the bolometer with its necessary accessories. Nevertheless, to make it ecjual 

 to the bolometer as regards " wiggle" and "drift" and capacity for accurate setting 

 it would require a mounting at least a fourth as costly as the bolometer and its acces- 

 sories. On the whole the bolometer has the advantage, except in cost. 



