HALL. — THERMO-ELECTRIC HETEROGENEITY IN ALLOYS. 551 



couples are to be seen, the extreme difference, in the " crossed " group, 

 being more than 2 per cent of the mean deflection. These differences 

 are not to be accounted for by differences of resistance of the different 

 couples, for each had a resistance very near 10 ohms, while the resist- 

 ance of the rest of the circuit, into which each couple was introduced 

 in turn, was more than 1000 ohms. Nor are the differences in 

 question to be regarded as purely accidental, in the sense of having no 

 tendency to recurrence. Taking the observations of the five days on 

 which the four couples here considered were tested with temperatures 

 near 140° and 180° we get, as means : 



A 2 B 2 C 2 D 2 



"Parallel" 19.82 19.64 19.93 19.78 



>0.03 >0.30 >0.36 >0.05 



"Crossed" 19.79 19.94 19.57 19.83 



19.81 19.79 19.75 19.81 



Here we have as the order according to magnitude of deflections: 

 " parallel," C > A > D > B, " crossed," B > D > A > C, just as on 

 July 21, the case cited in detail above. The mean range of variation 

 through the five days now considered was about 1.6 per cent for the 

 " parallel " and about 2 per cent for the " crossed " arrangement. 



Third, that a couple which gives a deflection larger than the mean 

 with the " parallel " arrangement is likely to give a deflection smaller 

 than the mean with the " crossed " arrangement, and vice versa, so that 

 differences of the kind mentioned in the preceding paragraph are very 

 small among the individual means, of " parallel " and " crossed " taken 

 together, yielded by the various couples. The greatest difference on 

 July 21 was less than \ per cent of the mean deflection, and the great- 

 est difference in the five-day means given above is the same, 6 parts in 

 nearly 2000. 



Fourth, that the mean "parallel" deflection for July 21, which is 

 19.84, exceeds the mean " crossed " deflection for the same day by 

 nearly 1 per cent of the mean deflection. It is to be observed, how- 

 ever, that the mean " parallel " deflection and the mean " crossed " 

 deflection of the five-day summary are almost exactly equal, being 

 respectively 19.79 and 19.78. 



As an example of the daily summaries obtained when the junction 

 temperatures were 100° and 218,' approximately, the following, for 

 July 25, will serve : 



