148 BRIDGMAN. 



pressure obtained from a large compressed air bottle, and was regulated 

 to any desired value with a safety valve of special construction. 



In addition to the gold and silver leaf, I made attempts to detect 

 the effect in manganin wire 0.001 inch thick rolled flat, and with 

 Wollaston wire of platinum about 0.00006 inch thick. The attempt 

 with manganin failed because the heating effects were too large, due 

 to the dimension of the specimen. The attempt with platinum failed 

 because of mechanical difficulties in mounting the wire and su)>jecting 

 it to a stream of water. It is possible that with more pains it might 

 be feasible to obtain results with platinum in this way. 



The thickness of the leaf metals used in this experiment was more 

 than sufficiently small to ensure conduction of the Joulean heat 

 developed by the heavy current without excessive rise of temperature. 

 To illustrate the order of magnitudes involved let us consider an 

 example. One specimen of silver that gave good results had the fol- 

 lowing dimensions: Length 0.536 mm., width 0.072 mm., and thick- 

 ness 2 X lO"'' cm. The maximum current before the specimen burned 

 out was 0.745 amp., and the initial resistance was 1.30 ohms. The 

 heat input into this specimen was therefore: 



rr . (.745)2X1.30 , 



Heat input = -tTo S^^ cal/sec. 



4.18 



= 0.173 gm cal/sec. 



This heat flows out through the area of one face, which is 0.0536 X 

 0.0072 = 3.87 X 10-* cm^. The heat outflow per unit area is therefore 

 (0.173)7(3.87 X 10^) = 4.5 X 10- cal/sec cnr. Since the thermal 

 conductivity of silver is approximately unity, the temperature gradi- 

 ent required to drive this thermal stream is 4.5 X 10- degrees per cm. 

 But the total thickness of the film is 2 X 10"^, so that the extreme 

 temperatiu'e difference in the specimen between front and back face 

 is 4.5 X 10-' X 2 X 10-» = 0.009°. 



It is of interest to compare the heat input with the heat capacity of 

 the specimen. Its volume is 3.87 X IQ-^ X 2 X 10"^ = 7.8 X IQ-^cml 

 Taking for the specific heat of silver 0.056, and the density as 

 10.5, we find the heat capacity to be 10.5 X 0.056 X 7.8 X 10-^ = 

 4.6 X 10"^. If there were no heat outflow the temperature would rise 

 at the rate of (0.173)/(4.6 X 10-^) = 0.038 X 10^ = 38,000,000 

 degrees per second. 



The magnitude of the steady temperature rise actually observed in 

 this specimen was about 50°, or 5000 times more than the mean rise 

 of temperature required to procure conduction of the heat input out of 

 the metal itself. It is ol)vious, therefore, that practically all the 



