MancJiester Memoirs, Vol. xlviii. (1903), No. J{. 5 



of liquid air changes at a rate which is considerable in the 

 case of a small quantity, and it seemed very unlikely 

 that the fluctuations of temperature of the metal would 

 follow those of the liquid air at all closely. Lastly, the 

 rise of temperature of a cold dry body during the 

 transference to the calorimeter is considerably greater 

 than that of a body wet with liquid air. 



I shall not again refer to the method of cooling by 

 protected immersion, but I should like to suggest that 

 this method would probably repay investigation if liquid 

 oxygen or a large supply of liquid air were available. 



The Transference from Cooler to Calorimeter. 



In a series of experiments by the method of 

 mixtures on the heating of a copper cylinder during its 

 transference from a bath of liquid air to a calorimeter, I 

 found that the rise of temperature per second was less 

 than •001 of the whole temperature difference. From 

 these experiments and from observations on other speci- 

 mens, I concluded that, for a transference lasting not 

 more than five seconds, the heating effect of the atmos- 

 phere might be neglected, being approximately neutralised 

 by the calorimetric effect of the small amount of liquid air 

 still adhering to the object at the moment of immersion.* 



In the present method the adhering liquid air has 

 practically no effect on the weight of ice formed, that is, 

 it has no compensating calorimetric effect. I found it 

 convenient, however, to use a time of transference of three 

 or five seconds ; various errors, which will be considered 

 later, made it superfluous to apply any correction for 

 heating during transference. 



The following numbers, the results of some early 

 experiments by the present method, illustrate the effect 



* Proc, Roy. Soc, vol. 72, pp. i8o — r86. 



