Air in Closed Vessels and its Causes. 571 



thus be drawn from the first vessel to the second at any 

 desired speed. The air as it entered the vessels was freed 

 from dust by passing through a plug of cotton-wool, and 

 dried by passing through another tube containing calcium 

 chloride. A steady current of this dry. dust-free' air was 

 now drawn through the two vessels, and the ionization in 

 each measured. The first was then heated, and measure- 

 ments taken in the second while the heating was in progress, 

 and in both when the heating had been stopped. As tin 

 had given the most marked effect in the preliminary ex- 

 periments, this metal was used for the first vessel, and as 

 zinc bad given the smallest normal leak it was chosen for the 

 second as being likely to make any small increase in the leak 

 in the second vessel easy of detection. It was hoped by this 

 method to ascertain whether the increase of the leak noticed 

 on heating a vessel was due to something of the nature of an 

 emanation which could be drawn from one vessel to another. 

 The experiment proved conclusively that such was indeed the 

 case. When heat was applied to the tin vessel, the ionization 

 increased not only in this vessel but also in the zinc one, 

 showing that whatever was causing the increased leak in the 

 first vessel was capable of being carried over in a current of 

 air from one vessel to another. So far the source of heat 

 used was a Bunsen burner, but this was found to be rather 

 inconvenient, as it was impossible to be sure of reproducing 

 the same temperature. This defect was obviated by wrapping 

 a thin sheet of micanite round the vessel and winding on this 

 a resistance of german-silver. By passing a current of about 

 2 amperes through this, the vessel could be heated to about 

 120° C. The insulation was thoroughly tested up to this 

 temperature, and found to act perfectly. The foregoing 

 experiments were now repeated with the apparatus thus 

 improved, and similar results were obtained. They were 

 accompanied, however, by a suspicious circumstance which 

 at once threw doubt on their genuineness and led to the 

 detection of a source of error previously unsuspected. The 

 leak at first increased when heat was applied. If now the 

 temperature was kept constant, the leak fell off again to its 

 initial value, and remained constant at that value until the 

 source of heat was removed. While the vessel was cooling, 

 on the other hand, the leak fell to an extremely low value 

 and in some cases it was even reversed in sign. It was this 

 manifest absurdity of a leak against the field which suggested 

 that the results were vitiated by some source of error. This 

 source was by no means easy to find, but the effect was 

 ultimately traced to the insulation. An enlarged diagram 



2 P2 



