The Gladstone u Laiu " in Physical Optics. 341 



On examining this table we find that the new expression 

 gives results w T hich very fairly agree with those derived from 

 Lorentz's formula. The average difference between the two 

 is about 5 per cent., a difference not greater than that which 

 would be yielded by the expressions of Clausius and Maxwell. 

 So far as this agreement goes we may regard 



as a fair approximation to a volume quantity, which I may 

 be allowed to call " molecular domain." 



The value, as given by either formula, increases with the 

 temperature. The highest increase by the /ul 2 expression is 

 about one per cent, in the case of aniline. Whether such an 

 increase could be attributed to the molecule proper, is an 

 unsolved problem. It is usual to regard the values yielded 

 by the Lorentz formula as more constant than here appears, 

 but the figures are too numerous and too reliable to leave any 

 doubt jibout the magnitude of the change. 



As I have already said, it seems probable that domains will 

 change in the same ratio between the same temperature limits. 

 This can be tested for the new expression by the figures in the 

 last column, which gives the ratio of the two values for 



(v — B) or (v — B) 



found at two temperatures. 



To facilitate comparison, the substances are arranged in 

 groups, marked off by a line. The members of each group are 

 heated between approximately the same temperatures. Thus 

 thymol is heated between 24°*4 and 77°, cinnamic alcohol 

 between 24°*8 and 77°. The domain of the first increases 

 in the ratio 1*0154, that of the second in the ratio 1*0152. 



The members of the large group beginning with heptane 

 and ending with ethyl cinnamate are heated between limits 

 the lower of which lies between 4° and 12°, while the upper 

 limit lies between 87° and 90°. The new domain rises 

 generally in this group by 3 per cent. Out of the ten sub- 

 stances, seven give a ratio of increase very near to 1*030, the 

 other three are near 1*027. 



The next group includes eight substances heated between, 

 say, 12° and 100°. Three give ratios close to 1*033, two yield 

 1*027. The last group is heated from 10° to 20° only, and 

 gives ratios close to 1*004, except in the case of alcohol. 



These results seem to afford a reasonable presumption in 

 favour of the figures being regarded as a measure of domain. 

 There is evidently a tendency for the ratio of increase to be 

 the same for the different liquids, although their densities vary 



