72 Mr. A. Stansfield on some Improvements in 



both. The rhodium used in the composition of No. 15 was 

 part of a commercial sample and was found to contain no 

 less than 30 per cent, of iridium, so that the alloy consisted 

 of 90 per cent, platinum, 7 per cent, rhodium, and 3 per 

 cent, iridium ; No. 16, on the other hand, contained 10 per 

 cent, of pure rhodium. Observations made with these couples 

 reveal the unexpected fact that, although the 10 per cent, 

 iridium alloy is rather more powerful than the rhodium alloy, 

 the partial substitution of iridium for rhodium in the latter 

 alloy very materiallv lowers its thermo-electric power when 

 compared with platinum. This result is of interest, as it 

 suggests that the change in the thermo-electric power of a 

 metal depends on the extent to which it is saturated with the 

 alloying metal ; 10 per cent, of either rhodium or iridium 

 would more completely saturate the platinum than would 

 10 per cent, of a mixture of the two metals, in the same way 

 that a single salt more completely saturates a liquid than the 

 same weight of a mixture of two salts of equal solubility. 

 The curve marked " Barus, 10 per cent, iridium " was obtained 

 by adding the mean results obtained by him with couples 

 19 and 20, consisting of "hard " platinum and platinum with 

 5 per cent, of iridium, to those obtained with couple 27, 

 consisting of platinum with 5 per cent, of indium and pla- 

 tinum with 10 per cent, of iridium. The use of " hard/' i. e. 

 impure platinum in these alloys probably accounts in part for 

 the smallness of the observed thermo-electric power. A com- 

 parison of his couples 22 and 35 shows that the E.M.F. at 

 930° would have been about 1200 microvolts higher if" soft" 

 platinum had been 'used : this, however, does not materially 

 reduce the difference between his results and those obtained 

 by the author with couple 13, and it appears probable that 

 the iridium he employed contained a considerable percentage 

 of rhodium or other metal of the platinum group. 



Although the curves in fig. 5 differ so widely from one 

 another in direction, there is a general agreement as regards 

 their form. All the curves are convex towards the scale of 

 temperature and their curvature decreases as the temperature 

 rises. The 10 per cent, iridium curves are decidedly straighter 

 in their lower parts than the 10 per cent, rhodium curves, so 

 that their indications are more nearly proportional to the 

 temperature. 



The usual method of presenting thermo-couple observations 

 has been adopted in plotting the curves in fig. 5, but it is 

 evident that a more critical method is necessary; for although 

 the uncertainties in the actual temperatures of points would 

 be quite visible on these curves, the observations themselves 



