498 Prof. H. Crew on the 



smaller divisions. Such an analysis would, at least, greatly 

 increase the chances of finding new series. 



Such a basis of division has been already hinted at by 

 Kayser*, who found that nearly two-thirds of all the reversed 

 lines of palladium were arranged in triplets having the same 

 frequency-differences. In short, reversibility appears to be 

 a distinguishing feature of a group of palladium lines whose 

 law of distribution is yet unknown. 



It occurred to the writer that possibly another basis for 

 division might be found in the effects produced by surround- 

 ing the arc with various atmospheres of gas, as, for instance, 

 hydrogen, nitrogen, coal-gas, ammonia. 



The experiment t with hydrogen was tried as follows:- — 

 An arc was operated between two metallic electrodes, one of 

 which was in rapid rotation while the other was fed in slowly 

 by means of a screw. This arc, including the rotating electrode,, 

 was placed in a brass hood which was made of two parts, not 

 unlike the Magdeburg hemispheres, except that these two 

 hemispheres, instead of having flat faces, were threaded with 

 a screw and provided with accurately turned flanges, so that, 

 when closed, the hood was gas-tight. 



Into one of these hemispheres were fitted the two elec- 

 trodes, one by means of a gas-tight bearing, the other by 

 means of a gas-tight nut-and-screw. It was necessary, of 

 course, to insulate at least one of these electrodes from the 

 brass hood ; this was done by placing the nut-and-screw 

 on a slate plate which, in turn, was screwed on to the brass 

 hood. 



The shaft which carried the rotating electrode and the 

 current could, therefore, touch the brass hood with impunity ; 

 but, as a matter of fact, this was not the case. 



On the contrary, the shaft fitted loosely into a bearing 

 which was packed with clean asbestos. This packing served 

 not only the purpose of insulation, but also to prevent any 

 oil reaching the interior of the hood. Consequently no 

 hydrocarbons were formed at the arc unless from carbon 

 impurities in the electrodes themselves. 



On the side opposite the arc, the hood was provided with 

 an opening into which screwed a brass tube about one foot 

 long. At its outer end, this tube carried an image-lens of 



* Kayser, " Bogen spectren der Elementen der Platin-gruppe," Abh. 

 Berl. Akad. 1897, p. 42. 



+ The entire apparatus described below and the experiments depending 

 upon it were made possible only through the generous consideration of 

 the Committee of the American Academy of Science in charge of the 

 Rumford Fund. 



