OF ARTS AND SCIENCES. 



30 



For this purpose I naturally had recourse to a thermo-electric 

 device.* 



The engine belonging to the Harvard Physical Laboratory, with 

 which my experiments were made, is a Kendall and Roberta of 

 10-inch cylinder and 15-inch stroke, provided with an ordinary 

 Myer's valve and cut-off. It is jacketed by an air space about 

 2 cm. wide. 



/£- 



\ 



^a 



\yt 



r^ 



1 



Fig. 1. 



Figure 1 shows a horizontal longitudinal section of the cylinder, 

 with the piston at the beginning of its forward stroke. The cylin- 

 der wall, about 2 cm. in thickness, is pierced at o for making 

 connection with an indicator. Into this hole, which is about 

 2 cm. wide, I screwed a plug represented accurately enough by 

 Figure 2. On the inner end of this plug was soldered a cover of 



M 



Fif.2. 



thin steel, S. Through two holes in the plug, and insulated from 

 it by glass tubes, two rods extended to touch the steel cover, one, 

 A, of antimony, the other, B, of bismuth. T> the outer cud of .1. 

 at a, was soldered a copper wire ; to the cuter end of 11 &\ b, was 

 soldered a similar wire. These two wires lead i<> a ballistic gal- 

 vanometer, but at one point the electric circuit was broken, except 



* A letter received from Mr. Bryan Donkin, Jr., of London, the well known 

 engine builder, after the first account of my experiments was published, -li- 

 me that he had been before me in making experiments in the Bame general man 

 ner, but with small bulb thermometers in mercury cisterns instead of thermo- 

 piles. The Bulletin <1< Sociele" de Mulhouse for 1890 contain- - i mint of 

 his work. My results appear to be not discordant with those which lie obtaii 



