290 PROCEEDINGS OF THE AMERICAN ACADEMY. 



radius 4-5-6-13, greater along the latter than along 7-8-9-13, greater 

 along the last than along 10-11—12-13. The mean difference of tem- 

 perature along 1-13 apparently exceeded that along 10-13 about 6 per 



Figure A. 



cent. A partial explanation of this difference probably is that, on one 

 face or the other of the disk, the flow of water was freer in the region 

 1-13-4 than in the region 7-13-10. This inequality of flow might be 

 caused by a slight tilting of HH ox of NH' (Fig. 2). 



Two short sets of observations were made to compare the deflections 

 given by all the pairs of junctions in multiple with the mean of the 

 deflections given by the pairs used separately. The pair 5-5' was now 

 found defective, and was omitted from the comparison. The results, 

 after allowance for the smaller resistance of the multiple arc, were as 

 follows : — 



Nov. 



The test of November 20 was the more careful of the two; but even 

 in this test the difference between the ratio found, 1.014, and unity, was 

 within the possible limits of error of observation. 



The tests which have been described were considered to justify using 

 the junctions in multiple and treating the current obtained from the 

 combination as representing the mean thermo-electromotive force of the 



