OF ARTS AND SCIENCES. 41 



Notwithstanding the great similarity of cast-iron and cast nickel 

 in many respects, they make a powerful thermo-electric couple, so 

 that nickel appeared to he of all substances the one best fitted for 

 my purpose. The hole bored in the iron to receive the nickel 

 cylinder was about 0.65 cm. in diameter. The cylinder itself was 

 about 0.025 cm. less in diameter, so that when wrapped with a sin- 

 gle layer of ordinary printing paper it would fit rather snugly in 

 the hole. The cast-iron plug with its cap in place was about 2.2 cm. 

 long, and the nickel cylinder projected very slightly beyond the 

 outer end of the plug. 



The inner end of the iron plug, the inner end of the nickel 

 core, and that surface of the iron cap which was to press against 

 them, were all worked with great care to a plane polished surface. 

 The cap and the nickel core were then soldered together in the 

 following manner. That part of the cap surface which was to meet 

 the nickel was thinly tinned with ordinary solder and then the 

 cap was screwed firmly into place upon the plug. That end of the 

 nickel which was to meet the iron was also tinned, and then this 

 core, wrapped, save at the ends, by a single layer of paper, was 

 thrust into the hole in the plug. The plug with its contents was 

 then heated very hot, and while hot was placed in a vise between 

 blocks of wood, which by the action of the vise pressed the nickel 

 core and the iron cap so close together that the layer of solder left 

 between them was, according to my measurements, less than 

 0.002 cm. thick. Then, after cooling, the screws holding, the cap 

 in place were drawn out, and cap and core were carefully removed 

 from the plug. The excess of solder which had been pressed from 

 between them, and which had gathered about the base of (lie core, 

 was then carefully turned off in a lathe with a brass tool. The 

 core was then wrapped with fresh paper and put back in place, 

 the cap being carefully screwed on. Red lead was used around the 

 screws holding the cap, but I depended mainly upon the perfec- 

 tion of the contact between cap and [dug to prevenl leakage at 

 this joint.* 



* In some of the experiments this joint was not perfectly tight, for water 

 was sometimes seen to come out past the core, N, before the plug became ln>t. 

 Generally upon such occasions the leaking appeared to cease when the plug 

 became hot. The fact probably is that the joint continued to leak Blightly, but 

 leaked dry steam, the temperature at the outer end of the plug being about 

 110° C, and at the inner end considerably higher. Both experiment and reason 

 indicate that the effect of such leakage upon the temperatures observed was 

 slight. 



