HALL. — THERMO-ELECTRIC HETEROGENEITY IN ALLOYS. 567 



these tubes with shields of asbestos and cotton, thus leaving it to be 

 heated from the tops of the pots. The temperature attained in the 

 middle of this air space was perhaps 100°, not so high as we wished it to 

 be, and we were far from getting a uniform temperature gradient in the 

 wires leading from one pot to the other. At any rate, the performance 

 of the couples in this trial was very little if any more harmonious than 

 it had been on September 29, when no care had been taken to keep the 

 parts in question warm. On the other hand, the mean electromotive 

 force found on October 28 was about § per cent less than it had usually 

 been with the same interval of terminal temperatures. 



Having now come to the conclusion that the arched 3 glass tubes, 

 containing the german-silver wires between the pots, were an obstacle 

 to the attainment of the temperature condition desired in this part of 

 the apparatus, we now cut out these connecting parts of the glass tubes, 

 thereby returning to an arrangement very like that shown in Figure 1. 

 Placing the pots as near together as they could well be, we surrounded 

 their tops with an elliptical wall of asbestos board, within which we placed 

 loose asbestos fibre. Within this bed of asbestos ran the german-silver 

 wires from pot to pot, each wire being covered to a depth of some centi- 

 meters. As the space from pot to pot was shorter for couples B 3 and C 3 

 than for A 3 and D 3 , the junctions of B 3 and C 3 were let down into the 

 pots two or three centimeters farther than those of the other couples. 



With this arrangement, calibration tests were made on December 16 and 

 20 with a new interval, 180° to 218°, approximately. The high mean 

 temperature of the junctions and the comparatively small interval of tem- 

 perature would naturally have given still larger values of (p — c) than 

 we had commonly found, even with the 140° to 180° interval, if we had 

 not taken care to shield and warm the middle parts of the german-silver 

 wires. More care in this regard was taken on December 20 than on 

 December 16. The term (p — c) is not strictly applicable to the tests of 

 these two days; for, both pots containing naphthalin, the reversal of 

 conditions was effected not by changing from " parallel " to " crossed " 

 or vice versa, but by making first the north pot and then the south 

 pot the hotter, or vice versa. We shall therefore use the term (n — s) 

 instead of (p — c). The summaries of galvanometer deflections for the 

 two days are as follows: 



3 We should probably have done well to retain these connecting parts of the 

 tubes, reducing the arch as much as possible so as to bring the wires down near 

 the tops of the pots. 



