On the Arc Spectra of some Metals. 497 



probably due to the introduction of the flat rod along the 

 axis of the guard-tube, as the dimensions and distances of 

 the other parts of the apparatus were practically the same 

 as in fig. 1. In these experiments, however, with hot water 

 and with ether, the sensitiveness of the electrometer had been 

 increased to 45 divisions per Clark cell. From these values, 

 taking as before one-fifth of a division per minute as the 

 minimum deflexion which could be relied upon, we get for 

 the minimum loss of charge per second per unit area which 

 could have been detected, 



<7=1'06 X 10 -6 electrostatic units. 



While occupied with the experiments described above my 

 attention has been directed to a second paper by Schwalbe, 

 published this year (Wied. Ann. 1900, Band i. No. 2, p. 294). 

 Besides criticising Pellat's method he gives the results which 

 he has obtained on repeating his own earlier experiments, 

 but using on this occasion potentials similar to those used by 

 Pellat, and giving both positive and negative charges to the 

 liquid. He comes once more to the conclusion that they do 

 not support the theory of a loss of charge by evaporation. 



I desire to take this opportunity of expressing my thanks to 

 Professor J. J. Thomson for the interest he has shown in this 

 work, and for his kindness in giving me every facility for 

 carrying it out. 



Cavendish Laboratory, Cambridge, July 1900. 



XLVIII. On tJie Arc Spectra of some Metals, as influenced by 

 an Atmosphere of Hydrogen. By Henry Crew*. 



THE orderly arrangement of lines in the arc spectra of 

 metals belonging to the first and second groups of 

 MendelejefPs Table, and the apparent disorder among the 

 lines of the remaining groups, constitute one of the most 

 striking features of spectroscopic science in its present state. 

 That this disorder is only apparent is evident not alone on 

 ■a priori grounds but also from the fact that, among these very 

 elements, many pairs and groups of lines with constant 

 frequency-differences have already been discovered. One 

 great barrier to the discovery of more complete order in 

 these spectra is the enormous number of lines with which 

 the metals of these groups are burdened. 



In order to simplify matters, the first step would, therefore, 

 appear to be the discovery of some basis upon which the 

 spectral lines of any one element might be separated into 

 * Communicated by the Author. 



