180 



Of the measurements recorded in the above tables, those 

 for acetylene, ethylene, carbon dioxide, and nitrons oxide were 

 made some time ago. But they are probably quite correct 

 enough to rank with the rest, which have for the most part 

 been made recently, since they are not affected by tempera- 

 ture and initial recombination difficulties. The measurements 

 most likely to contain error are those of the alcohols and 

 methyl iodide, the latter because I have been unable from 

 lack of material to repeat the one somewhat ancient determi- 

 nation, the former because for some reason the alcohols are 

 very difficult to manage in my apparatus. They are apt to 

 cause — particularly methyl alcohol — very large normal leaks, 

 though other vapours, such as benzene, have no such effect. 

 I believe the cause to be connected with the presence of min- 

 ute particles of fluff, which bridge across the walls of the 

 ionisation chamber, being stretched along the lines of force. 

 Although the apparatus is guarded with plugs of glass wool, 

 yet things of this sort seem to find their way into the cham- 

 ber at times, and it is possible that the methyl alcohol sets 

 them free from the sides or base of the apparatus to which 

 they are fastened by traces of grease. I have only once had 

 the apparatus in perfect working order with methvl alcohol ; 

 at that time T had gone over the working parts with a magni- 



