SOURCES OF LIGHT. 179 



wave-lengths was recorded on the plates, no measurable errors were intro- 

 duced. In fact, in the region considered, the second order ultra-violet of a 

 discontinuous spectrum taken on a film and on a plate could be superposed 

 line for line. 



SOURCES OF LIGHT. 



The sources of light and the selection of photographic material best suited 

 for spectrographic work had previously been investigated by one* of us in 

 an earlier research. 



Nevertheless, a few words relative to these sources may not be superflu- 

 ous in this place. 



For wave-lengths from "above" 0.65,^ to "below" 0. 326,", and for ex- 

 posures of two minutes or less, the Nernst glower was found to be the most 

 satisfactory. Prevailing circumstances made desirable the use of 106 volt 

 glowers, on a circuit carrying about 133 cycles. The emissivity of the 

 Nernst lamp varies so very greatly with the E. M. F. impressed upon its 

 terminals, that it was obligatory to keep in series with the glower a Thom- 

 son A. C. ammeter, having a range from zero to two amperes, and graduated 

 directly to 0.02 ampere. Fluctuations of more than 0.02 ampere invari- 

 ably resulted in a spoiled photograph. Therefore, boxes containing variable 

 metallic resistance were maintained in series with the ammeter, and thus, 

 in spite of large changes of the load on the dynamo, due to other circuits, 

 it was possible to prevent the effective current in the filament from chang- 

 ing by more than 0.01 ampere. The current was always 0.8 of an ampere. 

 The ammeter was appreciably more sensitive to small changes in the termi- 

 nal voltage than a comparably graduated Thomson A. C. voltmeter, because 

 the current shunted through the voltmeter was not negligible in comparison 

 with the current that fed the glower. Among other sources of light the elec- 

 tric arc was given a fair trial and discarded, for two reasons. First, because 

 of the intensity of the carbon and cyanogen bands ; and second, because of 

 the inconveniences resulting from its unsteadiness and great emission of heat. 



For wave-lengths between the strong ultra-violet of the Nernst glower 

 and 0.20/t, a spark discharge in air of about 1 cm. in length was used. One 

 electrode was composed of an alloy of equal parts by weight of cadmium 

 and zinc, and the other was made of sheet brass. The alloy wore away 

 so rapidly that the brass electrode was employed to diminish the mechanical 

 labor attendant upon sparking the terminals. In order to produce a source 

 of ultra-violet light that would have the same intensity from one end of 

 the slit of the spectrograph to the other, for any one wave-length, the 

 electrodes were made in the shape of wedges or chisels, with the sharp 

 edges parallel to the slit. The well-known distribution of a rapidly alter- 



*Uhler and Wood: Atlas of Absorption Spectra. 



