PEIRCE AND VVILLSON. — THERMAL CONDUCTIVITIES. 43 



speed of A' to one revolution in 20 seconds, though only two wheels are 

 shown in the drawings. The tripod slides in guides which revolve with 

 A', and a swivel at the top prevents the cord from twisting. 



The rotation of A' and of the inside ice holder, Q, which is connected 

 with K by a thin yoke, are matters of much importance. The continual 

 rubbing of the ice over the flat surface of the casting seems to be neces- 

 sary if the latter is to be kept at a uniform constant temperature for 

 hours. The energy used in rotating Q is so little as to be quite negligible, 

 as we shall show further on. The ice in K is piled up so as to cover 

 P completely, and we have been unable to detect any difference be- 

 tween the temperatures within and without P by fine, properly protected 

 thermal junctions introduced for the purpose. If, while A' revolves, Q is 

 kept still, the amount of ice melted in Q becomes irregular, though 

 the whole amount of drip in two or three hours is not very different 

 from the amount of steady drip in an equal time when Q is rotating. 

 Only selected lumps of ice are put into Q. The ice to be used is first 

 broken up into pieces weighing something like 15 grams each, by means 

 of an ice-cracking machine, and these pieces are then put into ice 

 water so that their sharp edges may become slightly rounded. They 

 are then drained and dropped into Q. In this way a slight amount of 

 water attached to the ice is introduced into Q^ but the error due to this 

 cause appears to be of slight importance. In some of our experiments 

 the ice to be used was carefully dried in cold blotting paper, but this 

 precaution does not seem to be necessary, though the use of small bits 

 of ice with sharp edges is to be avoided, ^'s capacity is about 2,000 

 cubic centimeters. After Q has been freshly filled in the course of any 

 experiment while K is rotating, no record is kept for some time, perhaps 

 fifteen minutes, of the amount of drip. Before the expiration of this 

 interval the extra water introduced into Q with the ice has drained off", 

 and the indications have become steady. After this the apparatus is 

 allowed to run for about two hours until .jOO grams of ice or less has 

 been melted, and then Q is refilled. The drip tube always contains a 

 few drops of water, but this amount remains sensibly constant during the 

 progress of our experiment. The drip is collected in a graduated vessel, 

 and the approximate amount is noted from time to time to see whether 

 the flow is steady. The whole is then more accurately determined by 

 weighing, at longer intervals. 



The regularity of drip is a far more sensitive test of the approximate 

 attainment of the final state of the body experimented on and its sur- 

 roundings than is a sensil)Iy constant tenqxM-ature gradient on the axis. 



