JACQXJES LOEB 471 



The effect of the salt on e was depressing and no increase in € is 

 noticeable in the low concentrations of CeCls or CaCl2. It is there- 

 fore obvious that the effect of the salts on e exhibited by our method of 

 experimenting can account for the drop in the curves in Fig, 1 but 

 apparently not for the initial rise or the augmenting effect of low 

 concentrations of CeCla or CaCl2. If Helmholtz's formula holds for 

 these experiments curves representing the product of the values 

 jE X € should therefore show an initial rise, followed by a drop ; more- 

 over, they should show the relative order of the rise as exhibited 

 in the transport curves in Fig. 1. 



In Fig. 2 the curves for £ X e are plotted with the concentration 

 of the salt as abscissae and the value of £ X e as ordinates. The 

 values of e are taken from Table IV and those for E from Table II. 

 The general order of the four curves in Fig. 2 is sufficiently similar 

 to that in Fig. 1 to indicate that our interpretation of the curves 

 in Fig. 1 is approximately correct. Thus the transport curve for 

 Na2S04 in Fig. 1 is flat, and so is the E X e curve for Na2S04 in Fig. 2. 

 The curves for NaCl, CaCl2, and CeCls rise in both figures in the order 

 named and in both curves the rise is followed by a drop. The second 

 rise of the curves in Fig. 1 after a concentration of m/4 is due to the 

 osmotic effect and has no connection with the electrical effect, as is 

 shown by the fact that this rise occurs also in the cane-sugar curve. 

 Hence, as far as the electrical effect in Fig. 1 is concerned, the charac- 

 ter of the curves resembles that of the curves in Fig. 2, as is proven by 

 the fact that if we superpose the curves for £ X e in Fig. 2 over the 

 transport curve for cane-sugar in Fig. 1 we get the curves of the type 

 of Fig. 1 . This is still more approximately the case if we substitute for 

 the purely osmotic transport effect not the curve for cane-sugar but 

 the curves for NaCl and CaCl2 at the isoelectric point of gelatin (Fig. 

 4). Since the curve for NaCl at pH 4.7 (Fig. 4) rises more slowly 

 than the corresponding curve for CaCl2, the lowest point in the 

 NaCl curve in Fig. 1 must be lower than that for CaCl2. 



This explains the empirical rule at which the writer had arrived in 

 his previous papers* on anomalous osmosis; namely, first, that the 

 transport of water through the collodion-gelatin membrane from the 

 side of the water to the side of solution is increased by that ion of the 

 salt which has the same sign of charge as the membrane, and dimin- 



