between the viscosity oe liquids and their chemical nature. 5^9 
obtaining would alter the value of n by more than 60 per cent. It is worth noting, 
however, that in spite of this uncertainty, which is, of course, greatest, in the case of 
mobile liquids having low boiling-points, the value of n varies within comparatively 
narrow limits for all the liquids ; if the alcohols investigated be excepted, the sixty- 
two remaining liquids give a mean value for n of 1‘766. The maximum value 2'234 
occurs in the case of isohexane, which has an extremely small value for h, and the 
minimum 1'408 in the case of bromine. Aldehyde gives the large value of 2755 for n, 
but here the value of h is smaller than for any other substance ; methyl alcohol gives a 
value similar to that given by aldehyde ; but in the case of ethyl alcohol n is as high 
as 4 ' 373 . For the higher alcohols two or three formulae are used in each case, but 
still the value of n remains in general high, its average value for the alcohols being 
3‘25. In the case of triniethyl carbinol the values of n are small; but here the 
values of h are enormous as compared with those given by the majority of the liquids. 
The alcohols thus differ from all the other liquids in giving larger values of n or of h, 
or of both these constants; they are thus characterised by the marked effect exerted 
by temperature on their viscosities. 
From the fact that the values of n and h are small for the great majority of the 
substances, it became a matter of interest to ascertain if in the formula = C/( 1 + 6 ^)" 
a few terms in the expansion of (I -f- ht)”- would not suffice to denote the effect of 
temperature upon viscosity. For inasmuch as the coefficients of t, &c., in the 
expansion involve both n and b, it might be expected that the magnitudes of these 
coefficients wmuld be related to the chemical nature of the substances. 
The data given in the following table serve to test the above points wlien three 
terms in the expansion are employed. 
In the columns headed ^ and y are given, for all the liquids but the alcohols and 
water, the values of these coefficients in the expression 
■n = C/(l + nht + = C/(l +l^t+ 
The remaining columns serve to give an idea of the accuracy with which this 
modified formula reproduces the observed values of rj. Since the terms omitted in 
the formula are most important at high temperatures, the differences between 
observed and calculated values will be greatest at the highest temperatures. Under 
t are given temperatures as near as possible to the boiling points of the liquids and 
satisfying also the condition, which simplifies the calculations, that each is an integral 
product of 10 . Under rj (calculated) are given the values of at the temperature t, 
as deduced from the above modified formula, and under 17 (observed) the values of if at 
t as read off from the curves. The differences obtained may be taken as the maxiraiun 
differences between the observed and calculated values given at any temperature. 
4 E 2 
