( 733 ) 



Some time ago one of' iis ') applied lo a spectrograph a peculiar 

 graduation according lo wave-lengths from 5 f//,t to 5 (m, which 

 proved very convenient. For instruments of great dispersive power, 

 however, this principle hardly works well ; more accurate results 

 are obtained with a calibration curve, though this takes more time. 

 For this calibration the lines of the gas spectra of hydrogen, helium, 

 and those of a mercury arc lamp may be used ; also those of the 

 spark spectrum of coj)per and of the tlame spectrum of potassium ; 

 in this way a sufliciently uniform distribution of lines is obtained 

 l)etween 410 and 770 mi. The accuracy of the readings is of the 

 order 0.05 (lu. 



Though from the outset we had been intent upon preventing end 

 play in the mechanism of motion, it pi-oved as yet impossible to 

 avoid this altogether, so that it was necessaiy for the readings to 

 have the motion take place always in the same sense. We hope, 

 however, to remedy this defect by further improvements. 



With the apparatus used as a spectrometer a very satisfactory 

 resolution of neighbouring spectral lines could be brought about, the 

 theoretical dissolving power of the set of pi-isms in the usual sense 

 amuunling to 65000. Thus with the strong eye-piece the yellow 

 helium line is seen resolved into its two components, whose distance 

 apart amounts to about 0,035 /ifx. 



Fabry and Jobin (loc. cit. p. 208) give a comparative table of the 

 breadth occupied in the spectrum by a wave-length interval of 1 mi 

 in the violet at about 434 nn; in the red the dispersion is of course 

 much less : 



APPARATUS 



DISPERSION 



Bruce (Yerkes-Observatory) 



Mills (Lick-Observatory) 



Fabry and Jobin 



DU Bois, Elias and Lowe 



RowLAND-grating of Berlin University, 1st 

 order (5084 lines per cm; radius iWo cm), 



Fabky and Jobin's fourfold focal distance is therefore all but com- 

 pensated by our greater dispersion. 



Though particular care was taken to |)revent reflexes, yet it proved 



1; F. Lowe, Zeitsclir. f. Instr. Kunde 26, p. 832, 1906. 



