618 Mr. N. Campbell on the Effect of Change of 



independent of the gas in the vessel : air, hydrogen, carbon 

 dioxide, and coal-gas were employed. 



11. Several methods of heating the vessel were tried — 

 a small luminous gas-jet burnt from a fine glass tube, a 

 Bunsen burner, a steam-jet, a metal box through which steam 

 was passed, an oil-bath and a coil o£ wire wrapped round 

 the vessel and carrying a current. The magnitude of the 

 changes do not depend directly on the temperature of the 

 source of heat, for all the above sources could be made to 

 give similar effects : it seems rather to be determined by 

 the rate at which heat is communicated to the vessel. No 

 actual measurements of this quantity were taken, since they 

 would haA'e required great experimental elaboration and did 

 not seem likely, in the light of the only explanations of the 

 matter that could be devised, to lead to any valuable results. 

 In any given vessel the magnitude of the changes could not 

 be increased beyond a certain limit whatever the source of 

 heat employed. But by employing less powerful sources the 

 changes might be diminished without limit. Thus in the 

 case of the vessel described in § 2, the value of Q was 

 approximately the same whether a luminous jet 2 cms. long 

 or a large Bunsen burner was used; but if a jet only 1 cm. 

 was used, Q was too small to be measured with any certainty. 

 With the big vessel used in the experiments described in the 

 former paper, no source of heat less powerful than a Bunsen 

 burner produced the maximum effect. 



If the vessel is turned upside down and heated on the top, 

 the changes are very much smaller. Indeed, it was only by 

 playing a large blowpipe-flame on the top of the vessel, at 

 the risk of melting the joints, that changes were obtained 

 that could be detected witli certainty. 



It should be remarked that, since all methods of heating 

 produce the changes., they cannot bo due to the increase of the 

 ionization in the air outside the vessel owing to the presence 

 of the flame gases. 



The effect of cooling the top of the vessel by placing on it 

 ice or liquid air was also tried. When the cooling agent was 

 applied a temporary increase in the leak was noted, and when 

 it was removed a temporary decrease. Hence it seems that 

 the sign of the change is not determined by the direction of 

 the change of temperature, but that an increase in the leak is 

 the accompaniment of a disturbance of the normal thermal 

 equilibrium, and a decrease of a return to that condition. 



12. Q increases with the ionization in the vessel, but 

 not so rapidly as the ionization. When the latter was increased 

 twenty-fold by placing uranium oxide in the vessel, Q was 



