The Chemistry of Globulin. 



131 



required to just dissolve the whole of G-. Denote g/G by p, and c/C by q, 

 then Mellanby's results are expressed by the equation 



p{l-q) = Aq(l-p), (1) 



in which A is a parameter having characteristic values for each type of 

 neutral salt. To illustrate the applicability of this equation, the following 

 comparison is made of the values of p (1 — g) / q (1 —p) for the different values 

 of c and p given by Mellanby for his first set of data for NaCl. C was not 

 actually measured by Mellanby, but is obtainable from his first graph with 

 only a small extrapolation as 0*0052. Hence, from Mellanby's tabulated 

 values of c, we obtain the values of q, given in the table along with his 

 values of p. The last row contains the ratio p(l — q)/q(l—p). 



Table I. 



10 4 c 



I0r\ 



l&p 



1Q2 J ? ( 1 ~g) 



2 (1 -j0 



50 



45 



40 



35 



30 



25 



962 



865 



769 



673 



577 



481 



910 



830 



665 



555 



450 



356 



40 



76 



60 



61 



60 



60 



The first two values are aberrant, because 1 — q and 1—p being 

 •small are powerfully affected by experimental uncertainty. If we take 

 j9 (1 — 2)/*2 0-"~P) = 0*60 = A, and use the equation to calculate p from q, we 

 find for p in these first two cases 0*938 and 0*794 in place of 0*91 and 0*83, 

 the differences being within the limits of experimental error. 



Mellanby shows that the graphs for NaCl, KC1, and NH 4 C1 coincide when 

 € expressed as gramme equivalents is made abscissa and p ordinate. Hence 

 the value of A just found for NaCl, namely 0*60 holds for KC1 and NH 4 C1. 

 Mellanby generalises this result for all neutral salts composed of two 

 monovalent ions, but in the appendix to his paper he points out that the 

 graphs for KA, KBr, and KI do not coincide with that for KC1, a result 

 previously found by T. B. Osborne, and given also by Hardy. Mellanby 

 iinds that a single graph will suffice for Na 2 S0 4 , K 2 S0 4 (NH 4 ) 2 S0 4 , MgCl 2 , 

 BaCl 2 , and CaCl 2 . For this type, namely, a divalent and two monovalent 

 ions, I find A= 0*613, say 0*61, and for the type MgS0 4 A = 0*66. 



One of Mellanby's discoveries leads towards a useful generalisation in 

 connection with A. He measures the average solvent power of any salt for 

 globulin up to the point when the whole of a suspension is dissolved, and 

 finds this proportional to the sum of the squares of the valencies of its ions. 

 For example, if the solvent power of Na is l 2 and of CI is l 2 , that of NaCl 

 will be 2, that of MgCl 2 will be 2 2 + 2 or 6, and that of MgS0 4 2 2 + 2 2 or 8. 

 It seems simpler to measure solvent power by the reciprocal of the con- 



