398 PROCEEDINGS OF THE AMERICAN ACADEMY. 



etc., were tried. But in each case, as soon as the region inside got hot 

 enough to affect the character of the spark and render it quiet, we found 

 (as, indeed, ought to have been anticipated) that the walls of the vessel 

 became conducting. 



We tried next to get a gradual variation of temperature by moving the 

 spark gap slowly from the centre to the edge of an ordinary carbon arc, 

 knowing that, at the centre of the arc, the spark was quiet and non- 

 luminous, while just outside the arc it became noisy and brilliant. But 

 in carrying the spark electrodes from one of these positions to the other, 

 we encountered a peculiar discontinuity, i. e. a position at which the 

 spark instantly changed character. 



When the spark was passed through the " horsetail " above the hori- 

 zontal arc at a distance of from ^ to 2 centimeters from the arc, the 

 quiet discharge mentioned above was still obtained and a spectrum of 

 feeble intensity could be observed. When, however, the terminals were 

 removed slightly farther above the arc, a point was reached at which the 

 discharge instantly assumed the ragged character of the ordinary cold 

 spark ; and when the spark was then moved back toward the arc it did 

 not resume its quiet character, but blew the " horsetail " away, and in 

 most cases put out the arc. It did not seem possible to obtain any inter- 

 mediate stages. The instability was very marked. The spark was 

 liable at any time to break down into the ragged character, and when it 

 had once done so it retained that character until the circuit was broken. 



Apparatus and Method. 



Accordingly we had recourse to soft-cored carbons worked in air, 

 using the same electrodes for both arc and spark ; in other words, we used 

 the hot region between the poles of an ordinary carbon arc as the heated 

 medium in which to study the slowly developing spark. 



The next step consisted in isolating the particular phase of the 

 development which we wished to examine. 



This was accomplished by means of a device (designed with the gener- 

 ous aid of Professor Basquin) which performs automatically the following 

 cycle of operations : — 



1. Closes the arc circuit and lights the arc, thus heating the carbon 

 electrodes and the region between them to a very high temperature. 



2. After an interval of a few seconds, sufficient for the carbons to 

 become thoroughly heated, interrupts the arc circuit. 



3. After an interval which is les3 than one-tenth of a second, closes 

 the spark circuit. 



