42 PROCEEDINGS OP THE AMERICAN ACADEMY 



Three plugs were thus completed, the caps being respectively 

 0.051 cm., 0.101cm., and 0.203 cm. thick. All the iron parts 

 were from one bar, and all the nickel parts were from one bar. 



From the nickel core of each plug when in use extended a slender 

 bar of nickel about 15 cm. long, made from the same piece as the 

 core, and from the outer end of the iron plug extended a similar 

 bar of iron made from the same piece as the plug. The bars were 

 not rigidly fastened to the nickel core and the iron plug, but were 

 held by friction in holes bored to receive their ends. The outer 

 ends, i and n, Figure 3, extended through a plug of hard rubber, _P, 

 screwed into the thickened wall, WW, of a vessel containing melted 

 paraffin, and copper wires soldered to these ends led back through 

 the hard-rubber plug toward the galvanometer, which was about 

 60 rn. distant. To facilitate adjustments the central hole of the 



hard-rubber plug was made somewhat too large for the nickel con- 

 necting bar reaching through it, and a stuffing-box, b, was depended 

 on to prevent leakage past this bar. The bulb of a thermometer 

 hung near the junctions i and n in the melted paraffin, which was 

 heated by a lamp placed below and was stirred by a mechanism 

 driven by the engine. 



The key by means of which the electric circuit could be closed 

 for a short part of each stroke required some thought in its evolu- 

 tion, and will be described with detail, although no doubt it 

 might be improved. 



At one stage in my experiments I tried making contact by means 

 of a fork-shaped spring of brass, which revolved with the engine 

 crank, and at one part of the stroke bestrode a gap in the circuit, 

 rubbing with each prong upon brass. A certain amount of thermo- 

 electric force was to be expected at each of the two rubbing points 

 of contact, but I had hoped that the two forces would counteract 

 each other sufficiently for my purpose. In this I was disappointed, 

 and I afterwards endeavored to avoid rubbing at the point of 

 closure of the circuit. The key finally used was essentially like 



