298 PROCEEDINGS OF THE AMERICAN ACADEMY. 



element in the shift, for sets 3 and 4, in which nearly the same power 

 (928 and 746 watts) was used show a large difference in shift. This is 

 due to self-induction, which causes less disruptiveness in the discharge. 



Ohmic resistance (sets 9 and 10) gives the same effect. 



The influence of a low percentage alloy appears in sets 5 and 6, and 

 the difference between the centre and end regions of the spark is shown 

 in sets 7 and 8. 



It may be remarked that set 3 (power 928 watts) shows less shift, 

 0.033, than set 2 (power 388 watts), 0.037; but the difference is very 

 small, and moreover the disruptiveness in the latter case may have been 

 greater. 



Summation of Results. 



In conclusion it may be said that the writer's former results have 

 been qualitatively and, to a certain degree, quantitatively confirmed. 

 It is difficult to see how we can avoid the following deduction : Under 

 certain conditions, namely, when self-induction and ohmic resistance are 

 absent, large capacity or long secondary gap are used, and the terminals 

 are of a high percentage alloy, the part of the spark which lies near the 

 terminal gives a spectrum, the wave-lengths of the lines of which are 

 greater than those of the arc spectrum. The shifts occur when the 

 energy involved in each discharge is great, and the discharge is disrup- 

 tive in nature. These conditions mean the creation of a large amount 

 of metallic vapor in a short time — a highly explosive phenomenon. 

 That an increase of wave-length should result is rather to be expected. 

 This may be rendered more evident from the following discussion : 



In addition to those theoretical considerations (mentioned in the 

 writer's former paper) * which would lead us to expect an increase in 

 wave-length, namely those dealing with the pressure, "p" in a gaseous 

 medium as a function of the number of particles per c.c, u n," the 

 mass of each, "»?," and the mean squared velocity, "?< 2 ," postulating 

 the truth of the equation p = I mtin 2 , it may be stated that such an 

 increase in pressure may come from the fact that, at the beginning of 

 the discharge, the air acts as an incompressible medium (compare the 

 downward explosive effect of dynamite placed on the surface of the 

 earth). The # development of the metallic vapor takes place so quickly 

 that the air has no time to move. The lessened shift in the centre of 

 the spark can then be easily explained, for it is known that some time 



* Astrophysical Journal, 17, 296 (1903). 



