50 PROCEEDINGS OF THE AMERICAN ACADEMY. 



Z, Z', Y' and X'. One junction of each of these couples was attached, 

 with a thin layer of sheet rubber between, to one of the guard-ring bars. 

 Thus, one junction in the case of couple Z (see Figure 4) was at the middle 

 of bar 3 ; Y was on the same bar, 4.25 cm. from the middle toward the 

 left end; Jon the same bar 4.25 cm. beyond Y. Z was on bar 3 at 

 the middle ; Y was on the same bar 4.25 cm. from the middle toward 

 the right end ; X' on the same bar 4.25 cm. beyond Y. * The placing of 

 these junctions was not very exact. The outer junction of each couple 

 was placed in oil in a glass tube about 10 cm. long and perhaps 0.4 cm. 

 in diameter of bore. These tubes were placed in a mixture of ice and 

 water. The couples X, Y, Z, etc., had not been directly calibrated ; but 

 similar junctions, A', B', C, and D', made at the same time from the 

 same wire treated in the same way, had been tested for the purpose of 

 getting the reduction factors to be used with X, Y, Z, etc. The use 

 of these factors, applied to the observations made December 8, gave 

 as the temperature of the middle points of the guard-ring bars, 95.1° C. ; 

 as the temperature of points 4.25 cm. nearer the ends, 96.0° ; as the 

 temperature of points 4.25 cm. nearer still to the ends, 98.3°. It 

 was evident, however, that no great confidence could be placed in 

 the exactness of these numbers ; for, with nearly a hundred degrees 

 difference of temperature between the junctions of a couple, an error of 

 one per cent in the reduction factor would make an error of nearly 

 one degree in the estimation of the temperature of the hot junction. 

 Accordingly, on December 22 a new set of observations was made with 

 couples X, Y, Z, etc., the oil tubes containing the outer junctions being in 

 this case placed in steam over boiling water. The temperature of this 

 steam, as. found from the barometric pressure, was 100.3° C. The 

 temperature found for the X point of the guard-ring bar 8 was 97.8° ; 

 for the Y point of the same bar, 95.4° ; for the Z point, 94.2° . For 

 the X' point of the bar 3 the temperature found was 96.5° ; for the Y' 

 point, 94.7° ; for the Z f point, 93.5°. Thus, the mean temperature indi- 

 cated for bar 3 was about 0.9° lower than the mean temperature indicated 

 for bar 8. This is hardly surprising, as these two bars lie a quarter 

 circle apart in the guard-ring, and the region of the lower one, No. 8, 

 is somewhat more shielded than the region of the upper bar. More- 



* The intention was that junction X should be just opposite, just as far from 

 the adjacent end vessel as the mid points of the couples A' A" and NN', and that 

 X' should have a like relation to MM' and P P'. All the junctions on a and (3 

 were entirely hidden when A", "5', Z, ete., were placed, and the mid points of KK' 

 and M M' were supposed to be 17 cm. apart instead of 16 cm. 



