6GG PROCEEDINGS OP THE AMERICAN ACADEMY. 



In place of a glass spiral to contain the expanding liquid, a metal one 

 was substituted. This was made from several lengths of thin-walled 

 steel tubing, 1 centimetre in diameter. Its total length was about 30 

 feet and its volume about 0.45 litre. The joints were brazed and re-in- 

 forced, and the whole thoroughly galvanized and painted. The change 

 introduced the inconvenience of joining the glass cut-ofE arrangement to 

 the steel tube. This, however, was accomplished without difficulty, by 

 means of a commercial cement called de Kolinsky Laboratory Cement. 

 The spiral responded with great rapidity to temperature changes. To 

 prevent leakage through the glass stopcock * we bent tliat end of the 

 steel tube into a U and filled it with mercury. Evidently a much less 

 efficient stopcock is required with mercury than if benzol (the expanding 

 liquid) were in direct contact with it. Further, bulbs were provided in 

 both arms of the U-tube of the glass cut-off, to afford greater latitude of 

 temperature changes. An additional inlet provided with a stopcock 

 was inserted to admit when desirable large amounts of hot water. This 

 proved of convenience in heating the thermostat rapidly to about the 

 right temperature. 



Ether V7as first employed as an expanding liquid. In many respects 

 it is an ideal substance for such a purpose. Its coefficient of cubical 

 expansion and its relative conductivity are greater than those of any of 

 the commoner liquids. Its specific heat, too, is very small. Unfortu- 

 nately, not only is its boiling point inconveniently low, but, in standing, 

 bubbles invariably appeared in the spiral tube, even when the method 

 of filling precluded the possibility of any bubbles having been initially 

 present. After many trials, therefore, ether was abandoned, in spite of 

 its exceptional advantages. After ether there is little choice between a 

 number of organic liquids, such as benzol, toluol, chloroform, and ace- 

 tone, as the following table shows. The alcohols were quite uusuited 

 to our use, for beside having rather high specific heats they exerted a 

 strong solvent action on the cement. Benzol was chosen. 



To fill the spiral, benzol was poured in at one end ; by proper tilting 

 most of "the air was forced out at the other. The remainder was 

 removed by boiling the liquid under reduced pressure. This was car- 

 ried out by attaching the end of the spiral to a glass tube, which reached 

 nearly to the bottom of a filtering flask, which in turn was connected 



* This glass stopcock is not sliown in tlie diagram ; it was at tlie opiiosite end 

 of tlie steel coil from the cut-off arrangement, and served to admit the proper 

 amount of liquid to the spiral. In order to understand the details of the apparatus, 

 Bradley's pai)er should be consulted, for only the differences are emphasized here. 



