238 SCIENCE PROGRESS. 



divided into three parts, the intensity of the outer portions 

 being the same in all positions of the analyser, and of course 

 equal to that of the middle portion at the zero position. 

 Greater accuracy is obtained with this arrangement than 

 when the field is only divided into two parts. 



Dielectric constant. — An investigation of a property little 

 studied from a physico-chemical point of view, namely, the 

 dielectric constant or specific inductive capacity, has been 

 carried out by Thwing (50). He uses a method, suggested 

 by Hertz, based upon the resonance of electrical oscillations. 

 Most of the observations were made at 1 5 , but in the case 

 of water they extend from o° to 90", and exhibit a well- 

 marked maximum at 4 . If K, M, and D be respectively 

 the dielectric constant, the molecular weight, and the 

 density, the results show that MK/D may be calculated 

 from definite partial values corresponding with the atoms 

 and groups of atoms composing the molecule ; the agree- 

 ment between observed and calculated values being, in 

 general, satisfactory. 



A remarkable relationship, which held for more than 

 twenty hydrocarbons and solid substances, is that K is 2*6 

 times the density. It is somewhat surprising that more 

 definite relationships appear to be obtained by the use of 

 MK/D than have hitherto been obtained by the expression 

 M(K-i)/(K + 2)D( 5 i). 



The above remarkable results will probably serve to 

 direct attention to this property, several points in connection 

 with which require reinvestigation. According to Maxwell's 

 theory, VK should be equal to the refractive index for rays 

 of infinite wave-length. Usino- for the refractive index 

 the first term in Cauchy's dispersion formula, Jahn and 

 M oiler (52) have shown that for several substances this is 

 not the case. Again, according to the Lorentz and Lorenz 

 formula (K— i)/(K + 2) for a given substance should be 

 proportional to the density. Rontgen (53) finds however 

 that K for water or alcohol is practically unaltered even at 

 pressures of 300-500 atmospheres. There are now several 

 methods of measuring K, and a new one has just been de- 

 scribed by Nernst (54). There is, therefore, every oppor- 

 tunity for attacking these questions afresh. 



