C. L. Speyers — Molecular Weights of Liquids. 431 



value is not large and does not seem to change rapidly. The 

 other quantity we find is positive when dw 1 , /at^>0, for then by 

 the nature of the case <£N" 2 '/^<0, and negative when dw/dt<^0 

 for then <ZN//^>0. Since w{+w" —w x a constant, we have 

 both cases so long as the two coexistant phases are present. For 

 the more dilute phase with respect to 1 dw;/dt>0, for the more 

 concentrated phase dw" /dt <0. For this one we write similarly 

 to (4) 

 da," 1 1 r w ' p,dp,"—p,"dp t , p," 



[ 



<% P~P"' XN/Li?,-/?/ dt N 2 " 



From 1, 



dt 



]• (5) 



m. 



P-P"~ l P* 

 which substituted in (5) gives for the bracketed part 

 m N/ r p.dpj'-pj'dp, (ply £ N/rf<-<rfN/ -i 

 *!>/ L ''";■<*'"' VN// 'm,' ^ J 



The solution now being concentrated, we cannot predict much 

 about the first term from a study of the vapor tension of the 

 pure solvent. I think, however, we may assume that the 

 numerical value of this term will not change rapidly compared 

 with the other term. The second term changes rapidly because 

 it is the arithmetic sum of two quantities, not the difference, and 

 because it has in addition a factor which is the square of a 

 fraction whose numerator and denominator both increase as t 

 increases. Consequently, for quite a wide range in concentra- 

 tion we may look for a negative value for the quantity in 

 brackets and therefore expect 



while for the less concentrated solution 



At the critical temperature,' dw/ / dt, dw"/ ' dt, dJS//dt, 

 dN/'/dt all reduce to zero, and so from (4) and (5), 

 da; __ da;' = da = w x 1 P,dp/-p/dp 2 



dt dt dt m x N 9 ' (p^ — p'rf' dt • w 



The solutions being concentrated we cannot predict at all con- 

 cerning the sign of da/dt from the vapor tension curve accord- 

 ing to temperature of the pure solvent, and the data at hand 

 for the vapor tensions of the solutions are limited to a few 



