132 
MR. W. R. BOUSFIELD: IONIC SIZE IN RELATION TO 
Table XXI.—Viscosity of Mixtures of Normal KC1 and NaCl Solutions, calculated 
from the Hadions. 
CL 
y] Observed. 
r/ Calculated. 
Difference. 
1 
0-01037 
0-01037 
+ 
0-75 
0-01063 
0-01063 
+ 
0-5 
0-01088 
0-01089 
- 1 
0-25 
0-01116 
0-01115 
+ 1 
o-o 
0-01143 
0-01141 
+ 2 
The agreement between observed and calculated values m Tables XIX., XX., 
and XXI., is sufficient to show that the chief element which determines the viscosity 
of the solutions is the product m(r 1 + ?’ 2 ). But the expressions (14), (15), and (16), 
are not put forward as true viscosity formulae for KC1 and NaCl. They are only 
intended to show that the viscosity of the solutions is mainly dependent on the size 
of the radions. The results tend strongly to confirm the fundamental hypothesis 
as to the character of the magnitudes which we have called “ radions,” and which 
we derived from the conductivities by the application of the Stokes theorem and 
the Van’t Hoff law. 
It is surprising that the agreement with the observed values in Tables XIX. 
and XX. should be so close when we consider that at normal concentration the 
ionisation is less than 07. This may be explained on our theory in the following 
way. As indicated by the agreement upon the density formulae (see Tables XVII. 
and XVIII.), there appears to he no break in the hydration law as between ionised and 
un-ionised molecules. Thus, if a represents the condition of things when the molecule 
is just sufficiently hydrated for ionisation, and ( b ) the condition of things shortly 
before ionisation, the same functions approximately express the 
ionic radii before and after ionisation. A little consideration 
will show that the expression m (r, + r 2 ) takes the radion of 
the un-ionised molecule as r x - yr 2 . For if cl be the ionisation, 
this would give us for the separate ions the products amr 1 
and amr 2 , and for the un-ionised molecule the product (1—a) m (r! + r 2 ). The sum of 
the three products is simply w(?'! + r 2 ). This seems to explain why the linear relation, 
which might be expected to obtain between m ( i \ + r 2 ) and y for dilute solutions, does 
in fact approximately persist up to normal solutions. 
(b) References to Previous Theories. —-The results obtained in the last section 
suggest considerations which may tend to a great simplification of our ideas in 
reference to the internal friction of fluids. Hitherto, very diverse theories have been 
advanced in order to account for the apparent anomalies in the viscosity of liquids, 
and various general formulae have been proposed in order to express the experimental 
a. 
