On a Practical Constant-Volume Air- Thermometer. 149 



to the acid, we have succeeded in obtaining such an upper 

 limit. The discovery of an upper limit makes some of our 

 experiments of less value, for where we have stated that no 

 film has been obtained, the non-success may be due to failure 

 in fixing on the right limits of current-density. It also 

 explains the peculiar jumps in the current which we have at 

 times observed in using platinum electrodes in strong H 2 S0 4 . 

 These jumps, which were of a similar kind to those which 

 we obtained with gold electrodes, are probably due to the 

 current-density used being near the upper limit when small 

 causes, such as the heating of the liquid, would involve the 

 disappearance of the film*. 



This provisional hypothesis likewise accounts for the non- 

 formation of the film in dilute H 2 S0 4 f. Here the rate of 

 dilution (even if the electrolyte be made viscous by the 

 addition of glycerine) of the concentrating layer will be more 

 rapid, and hence a greater length of time would be required 

 to produce a thoroughly concentrated layer. But the longer 

 the time that elapses before this is attained, so much the 

 greater must be the quantity of oxygen produced. Thus, just 

 as in the case of current-densities above the upper limit in 

 strong H 2 S0 4 , the formation of film will be prevented. 



The less rapid increase of the specific resistance with the 

 degree of concentration in the cases of H 3 P0 4 and KHO 

 probably explains the difficulty found in obtaining such a 

 high degree of insulation as we have observed with H 2 S0 4 . 



Some other aspects of transition resistance, evidenced by 

 some experiments not included here, will be considered in a 

 subsequent paper. 



XVII. On a Practical Constant- Volume Air- Thermometer. 

 By J. T. Bottomley, M.A., F.R.S.l 



IN the fourth Memoire of his celebrated Relation des Ex- 

 periences, published in 1847, Regnault gives cogent 

 reasons for preferring the air- thermometer before any other 

 as the instrument by means of which temperature may bo 



* "\Ye are not disposed to think that the jumps obtained with gold are 

 due to the same cause. Gold really comes under the head of attackable 

 electrodes, and the phenomena with it are, therefore, probably of a more 

 complicated character than with either platinum or carbon. Palladium, 

 in this respect, must be classed with gold, for when used as anode in 

 moderately strong sulphuric acid, it dissolves. 



t The results obtained by Christiani, we believe, may also be explained 

 by this hypothesis. 



X Communicated by the Author, being a paper read before the Royal 

 Society of Edinburgh, Jon. 6, 1888, with an addition of June 5, 1888. 



