404 PROCEEDINGS OF THE AMERICAN ACADEMY. 



explanation of the irregularities apparent in the distribution of the 

 images in the actual photographs. The similarity of their distribution 

 at the beginning of a series and their distribution at the end of a series 

 shows that the accumulated effect of a number of discharges does not 

 render the bulb conducting so as to weaken succeeding discharges. 



By the use of a small Leyden jar as capacity and a charging potential 

 of 15,000 volts, I have been able to obtain over 200 complete discharges, 

 each with its series of oscillations, during one-half cycle ( T ^o second) 

 of the charging transformer. These complete discharges, comprising 

 each many oscillations, were separated by an interval of time of about 

 tt> o h o o °f a second, yet every discharge was sharp, definite, and regular, 

 and showed that even after a long operation of the interrupter at this 

 frequency of charging the bulb did not become filled with conducting 

 vapor or conducting ioris so as to lower materially the potential of suc- 

 ceeding discharges. This seems to me to be a very important part of the 

 whole advantage that the Cooper-Hewitt mercury interrupter has over 

 the spark in air. 



Figure 9, Plate III, is a typical example of the behavior of a spark in 

 air when produced by a high potential. This is a picture of a spark 

 between zinc terminals taken with a rotating film. Each spot is the 

 image of a complete discharge with its series of oscillations. It is seen 

 that these discharges at the beginning of a cycle are strong, but through- 

 out the cycle become spasmodically strong and weak, showing that the 

 spark-gap often retains its conducting character long enough to prevent 

 the proper subsequent charging of the condenser. This discussion may 

 apply only to the case in which the condenser is charged by some per- 

 sistant source of current like the transformer. I have not ascertained 

 whether a similar result is to be found when the condensers are charged 

 by an induction coil or a static electric machine. 



VI. The Resistance of the Mercury Interrupter. 



In measuring the mean resistance of the mercury interrupter, I have 

 made use of a calorimetric method similar to that employed by Battelli 

 and Magri* in their determination of the resistance of a spark-gap in 

 a series of measurements on condenser discharges. A calorimeter con- 

 taining the mercury interrupter was put in series with a calorimeter 

 containing a known resistance. The discharge from the condenser 

 while connected to the transformer was allowed to pass for a suitable 



* Phys. Zeit, 3, 539 (1901-1902). 



