456 Sei'viss — Internal Temperature Gradient of Metals. 



the under side of the ring at the top. The suspended ring had 

 a socket which held the lower cylinder without slipping. The 

 threads were set vertically on a circle slightly larger than the 

 circumference of the thermopile. The junctions were thus 

 easily put in position symmetrically and could not be displaced 

 by any jar during the experiment. The weight of the upper 

 cylinder was relied upon to hold it in place. On opening the 

 thermostat nothing was ever found disarranged or wet. 



The thermostat was clamped to a beveled ring, which was 

 firmly supported by three lengths of maple dowelling rod. 

 The ice vat was -±2 C[n in diameter and T2 em tall at the center. 

 A one-inch pipe was inserted in the bottom to insure com- 



For convenience in draughting the three suspending threads and the 

 three legs are shown as two. 



plete drainage. The whole was set in a barrel and packed 

 loosely with sawdust. 



The Ice Bath. — As an ice bath is often used to produce a 

 constant temperature, I here give a detailed account of the 

 various schemes tried, hoping that my experience may be of 

 service to other experimenters. I may state here that this was 

 the most troublesome part of the whole investigation. 



Dr. Thwing mentions a noticeable rise of temperature 

 (0°-00001 to b -00002) caused by the melting of the ice from 

 the bottom of the thermostat during the twelve-hour intervals 

 between which he packed the ice down. In other words, 



