108 



Messrs. F. L. Usher and J. H. Priestley. [Apr. 13, 



flimsy metal part, and to diminish to some extent the loss of heat from one 

 surface by radiation and convection. The glass tubes into which the strips 

 fitted were about 15 cm. long and 1*5 cm. in diameter. When an experi- 

 ment was in progress the tubes were closed at their lower ends by rubber 

 stoppers, on which rested two short tubes (d, d'), one containing a solution 

 of carbon dioxide in water ("soda-water"), and the other a solution of 

 potassium hydroxide ; in this way the films were immersed either in an 

 atmosphere containing moist carbon dioxide or in one devoid of it. The 

 thermo-couples were connected differentially through a reversing key K and 

 a suspended- coil galvanometer G. Any error due to lack of symmetry in the 



thermo-electric behaviour of the system was eliminated by taking always 

 the mean of readings obtained before and after the key was reversed, and 

 differences of temperature due to slight inequalities in the films themselves 

 were allowed for by interchanging the solutions in the tubes d, d' . The 

 thermocouples were calibrated directly against two mercury thermometers, 

 so as to obtain the number of galvanometer scale divisions per degree 

 difference of temperature. 



Some preliminary experiments were carried out in September, 1908, 

 with a slightly different form of apparatus, and one series of observations 

 is given below; the figures do not definitely settle the question at issue, 



