OF ARTS AND SCIENCES. 193 



charged plates, assuming that the charge on each atom is that deduced 

 from electrolytic considerations." 



The unusual dispersion afforded by a Rowland concave grating led 

 me to test this hypothesis, in so far as it relates to the question, 

 Are the molecules of metals carried with the oscillations of electricity 

 between terminals between which the oscillations take place ? A 

 circuit of wire giving a suitable value of self-induction was arranged 

 iu connection with a series of Leyden jars. The time of oscillation 

 was calculated from the well known formula, t =: 2 ir VLC; iu which 

 L is the value of the self-induction of the circuit; C, the capacity of 

 the Leyden jars. Preliminary examination of the electric spark 

 taken through this circuit with a revolving mirror showed that the 

 discharge was an oscillatory one. Two different values of self- 

 induction were employed. One gave the duration of a double oscil- 

 lation < = .0000003 of a second; the other gave <= .0000024 of a 

 second. 



If we denote by V the velocity of light, X and A^ wave lengths, 8 the 



speed of approach of the atom, we shall have X^ = x( ) . 



The distance across which the oscillations took place was six millime- 

 ters. Calculation shows that if the iron atoms were conveyed to and 

 fro between the terminals, a broadening of the iron lines in the spec- 

 trum would result, which could be readily detected. The broadening 

 might amount to a space equivalent to a whole tenth meter. 



The oscillatory spark passed between tw^o iron terminals. One of 

 these terminals was hollow. The hollow terminal was placed in a 

 line perpendicular to the slit of the spectroscope, so that the oscilla- 

 tion of the spark should be toward and away from the slit. If, there- 

 fore, the iron atoms moved to and fro with the oscillations of electricity 

 across the air gap, a displacement of the iron lines in the spectrum 

 of the metal would result. There would be both a displacement 

 toward the less refrangible, caused by the recession of the atom, and 

 one toward the most refrangible end of the spectrum, caused by the 

 approach of the atom. The great amount of dispersion afforded by 

 a concave grating of 20,000 lines to the inch enabled me easily to 

 detect a movement of y^^ of a wave length. I accordingly took a 

 photograph of the iron lines with the terminals in the position I have 

 described, and on the same plate, immediately above this photograph, 

 a comparison photograph was taken with the terminals parallel to the 

 slit. In this case the iron atoms did not make their supposed ex- 



VOL. XXV. (N. 8. XVII.) 13 



