184 



assumption that they fell into line with the same equation. 

 This is justifiable, since my immediate object is to show a re- 

 lationship between ks and the atomic volume in combination. 

 As a matter of fact, the molecular volume of O., pe?- se has 

 been found by Dewar to be 27 "4 (Chem. JSTews, June, 1898). 

 This is close to the value in the table, viz., 24'4. But Dewar 

 also finds Hj to be 28, which is much larger than the value 

 used in the ordinary formula. 



10». 



The value of k for H, is set down as 100. This is only 

 approximate, and is probably too high. Its accurate deter- 

 mination will require the construction of special apparatus. 



The agreement between the ratios v / ks in the fifth 

 column is not such as to show that v and ks are directly pro- 

 portional; but it is good enough to suggest strongly that 

 they both rest immediately on some more fundamental pro- 

 perty. The case is even a little stronger than appears at 

 first sioht, since it is clear that Ho contributes an abnormal 

 amount to the molecular volumes; the ratio v/ks is high 

 whenever H preponderates in the molecule. Moreover, the 

 molecular refractions also run closely parallel, as is well 

 known, with the molecular volumes, and in general the con- 

 nection between the various physical properties of the mole- 

 cule and its volume is more obvious than any connection with 

 its molecular weight. Consequently the quantity ks is closely 

 related to most of the physical properties of the molecule. 

 As a second instance, I have put in the sixth column of the 

 above table the respective values of Sutherland's molecular 

 volume B (Phil. Mag., Jan., 1895), and shown in the last 



