104 Allen and Clement — Role of Water in Tremolile. 



only heating in a current of dry instead of moist air.* The 

 air was dried by concentrated sulphuric acid. The apparatus 

 in* which the dehydration was carried on is shown in fig. 1. 

 It is a cylinder of Berlin porcelain, closed at one end and 

 glazed on the inside to make it impervious to gases. The 

 upper end is molded so as to form a groove, into which fits 

 the iron cover. There are also two inner covers of porcelain, 

 each l cm thick, which rest on lugs baked on to the inner wall 

 of the cylinder. These keep the temperature more nearly 

 constant and protect the iron cover from excessive heat. All 

 three covers are perforated near the edge to allow the passage 

 of two porcelain tubes. One of these, which is closed at the 

 bottom, reaches down to the charge and carries the thermo- 

 element. The other is open at both ends, reaches nearly to the 

 bottom of the cylinder, and admits the current of dry air. One 

 of the tubes may be fitted closely to the iron cover by a " fibre " 

 ring, the other may be left loose enough to allow the escape of 

 the air. If it is desired to use some other atmosphere in the 

 furnace, mercury may be poured into the groove to more 

 effectually prevent any air from leaking in. The crucible 

 which holds the tremolite rests on a platinum triangle which is 

 supported by a hollow cylinder of fine white clay. This whole 

 apparatus in an upright position is then slipped into a platinum 

 resistance coil furnace which envelopes it to within 2 cm of the 

 top. The furnace is heated by a storage battery and the tem- 

 perature can be regulated very closely indeed for a period of 

 many hours. 



Conditions of Experiment. — There was no difficulty in 

 maintaining the temperature within 5° without much atten- 

 tion, except when the same battery was required intermit- 

 tently for other work. In such cases there were sometimes 

 aberrations of 10° or more. As the time of experiment is not 

 essential, a fall in temperature is of no importance ; a half 

 hour or more at a temperature as much as 5° below the 

 point aimed at was not counted. On the other hand, if the 

 temperature ran 5° or more above the point, the work was 

 rejected unless the charge underwent further loss at the same 

 temperature on the following day. The loss of water was so 

 very slow that an aberration of this kind scarcely ever affected 

 the results. Besides, our object was only to determine the 

 form of the curve w 7 ith sufficient accuracy to settle the ques- 

 tion whether the loss of water was continuous or discontinuous. 



Regarding the conduct of an experiment only a few words 



of explanation are needed. The mineral was generally in the 



form of a rather coarse powder, ground only fine enough to 



pass a screen of 100-120 meshes to the linear inch, because 



*The question of equilibrium is raised on p. 117. 



