HALL. — THERMO-ELECTRIC HETEROGENEITY IN ALLOYS. 557 



the experiment of raising all of the couples about 20 cm. higher than 

 usual, with the object of getting the bent part of the german-silver 

 wires far enough away from the top of the pots to have them in a 

 region of nearly uniform temperature. With the same object in view 

 we placed shields of asbestos to protect these parts of the wire from 

 the heat given off by the pots and the tubes above them. All this 

 was done in the hope of more uniform performance by the different 

 couples. How little ground there was for this hope may be seen from 

 the following summary of the series of observations made with this ar- 

 rangement, although the observations at the last went so badly that 

 the series was not completed in the usual way: 



A 3 (p — C) B 3 (p — c) C 3 (P — C) D 3 (p — C) Mean (p — C) 



19.69 



19.21 



19.12 



19.20 



19.31 



These observations of September 

 27, taken in connection with what 

 had gone before, told strongly against 

 the theory which attached great im- 

 portance to the bent parts of the 

 german-silver as a source of variation 

 and irregularity, and after the exper- 

 iments of September 28 and 29 had 

 shown that moisture had probably 

 exerted no appreciable influence on 

 the performance of the couples, it 

 seemed probable that the cause of the 

 irregularities noted must be looked 

 for in differences of condition in the 

 straight parts of the german-silver 

 wires. Accordingly we decided to 

 look for such differences by direct 

 experiment; but we thought it best 

 to make, before this study was under- 

 taken, a direct experiment on the 

 change which moderately sharp bend- 

 ing makes in the thermo-electric condition of german-silver wire. 



Figure 



