118 PROCEEDINGS OF THE AMERICAN ACADEMY 



of which I desired to study. At each trial the type-metal was moved 

 so as to expose a new cutting surface. The type-metal was insu- 

 lated from the rest of the apparatus, but connected with the outer 

 coating of the Leyden jar ; first both terminals of the Holtz machine 

 were thrown off, and immediately after the cutting tool, ploughing its 

 way through the type-metal, placed the' outer coating of the Leyden 

 jar in circuit with one of the two parallel wires leading to the ter- 

 minals of the spark. The other wire was permanently in connection 

 with the inner coating of the jar. 



Beside the short lead wires above described, the discharging circuit 

 consisted of two parallel wires 30 cm. apart and 510 cm. long. These 

 were the only portions of the apparatus changed during the experi- 

 ment, and they were replaced by wires of different material and of 

 different size. The other conditions — length of spark, lead wires, 

 and the copper cross wire connecting the outer end of the long parallel 

 wires — remained undisturbed throughout the experiment. 



The Leyden jar was charged each time as nearly as possible to the 

 same potential, judging by the number of turns given the Holtz ma- 

 chine. It is unfortunate that no more accurate means of measuring it 

 were at hand, although the different negatives showed but slight varia- 

 tion. The capacity of the jar to alternations of this period was 5060 

 electrostatic units. 



I describe the discharging portion of the apparatus minutely, for 

 the success of an investigation of this na'ture depends upon the sup- 

 pression of all sparks save that which one wishes to observe ; and the 

 method surely and completely accomplished this. The photograph of 

 the spark could thus be made to fall very accurately on the sensitive 

 plate. When one considers that the image of the spark was flying 

 through the air on a circle of a radius of ten feet with a velocity of a 

 mile a second, it will be seen that an extremely small deviation in the 

 point of contact between the cutting tool and the type-metal would 

 have thrown the image entirely off the sensitive plate, A singular 

 phenomenon was noticed in this connection. When a comparatively 

 low potential was used, such as that afforded by the air condenser used 

 in our previous investigation, the cutting tool ploughed two or three 

 millimeters along the surface of the type-metal before a spark passed 

 at the point in the circuit where it was de>ired. With higher poten- 

 tials this phenomenon was also observed, but the extent of cutting 

 was diminished. 



It is possible that that the insulating wax may have melted under 

 the sudden blow of the cutting tool, and, flowmg around it, prevented 



