462 Proceedings of the Royal Society of Edinburgh. [Sess. 
carbamide, was unable to confirm Rtidorf’s results. In our previous paper 
it was conclusively shown that aqueous solutions of carbamide do not 
exhibit “ negative ” viscosity at 25° C but that at low temperatures ( + 8° C .) 
a dilute solution is less viscous than water. The suggestion, first put 
forward by Arrhenius, was revived by Getman, that the anion and the un- 
dissociated molecule have the same influence on the viscosity of the solvent, 
but that the cation has the opposite effect, the tendency of the cation being 
to diminish the viscosity, that of the others to increase it. Jones and 
Yeazey extended this idea, starting from the observation that the salts 
which diminish the viscosity of water are salts of metals with large atomic 
volumes. Several instances are adduced in support of this view ; thus 
potassium chloride and potassium nitrate solutions are less viscous than 
water, while many other potassium salts are not. In the latter, the increase 
in viscosity due to the anion is supposed to more than counterbalance the 
decrease due to the potassium ion. Potassium has the largest atomic 
volume of all the metals, with the exception of rubidium and caesium, and, 
as anticipated by the hypothesis, the chlorides of those metals diminish the 
viscosity of water to a greater extent than does potassium chloride. In 
connection with potassium chloride, it is further pointed out that of all the 
elementary anions, chlorine has the largest atomic volume, with the sole 
exception of bromine, which is slightly larger ; this unusually large anion 
is supposed to act in the same direction as the cation, or, at least, not to 
counteract its effect. Now, potassium bromide solutions actually do 
possess “ negative ” viscosity, and to a greater extent than equivalent 
solutions of potassium chloride ; but potassium iodide, the atomic volume of 
iodine being practically the same as chlorine, causes a still greater decrease 
in the viscosity of water than either the chloride or bromide. 
It would thus appear that, taken alone, the hypothesis of Jones and 
Veazey cannot account for the occurrence of “negative” viscosity even 
in the case of aqueous solutions of electrolytes. This conclusion is 
confirmed by a consideration of other solutions which have “ negative ” 
viscosity. 
The cases of potassium ferrocyanide and potassium ferricyanide are 
perhaps the most convincing. The anions Fe(CN) 6 //// , and Fe(CN) 6 /// , differ 
mainly in this, that the one has four negative charges while the other has 
only three, and their effect on the viscosity of water can scarcely be attri- 
buted to differences in volume ; this being so, the ferrocyanide with four 
K* should diminish the viscosity of water to a greater extent than the 
three K* of an equi -molecular solution of the ferricyanide. But the facts 
are quite the reverse, as the following data show : — 
