190 G. H. KNIBBS. 
The recorded efflux times, observed to 0-1 second, ranged from 
about 216 to about 681 seconds, and the determinations for each 
tabulated temperature were from two to eight in number. 
Thorpe and Rodger 1894.—Thorpe and Rodger’s measurements 
of the viscosity of water, made in connection with their important 
contribution to chemical physics hereinbefore referred to, give the 
results of no less than 13 double observations of the viscosity of - 
between 0° and 8:01° C., and also those of double observations 
at about every 8:5 from about 4°5 to about 99-7 C. Their pub- 
lished data are incomplete, and it is not therefore possible to 
undertake a thoroughly independent reduction ; but as they have 
given the applied corrections, in which they assumed m to be unity 
instead of 1°12, I have further reduced their viscosity values by 
subtracting 0-12 times the correction, and from these corrected 
values have obtained the relative fluidities, ascertaining them for 
every 0:5 between 0° and 8° by a system of parabolic interpola- 
tions, and similarly for every 5° from 0° to 100°. 
They mention a formula deduced for them by Riicker, which is 
identical with formula (18) page 112 of my former paper. The 
method employed to determine the axes of the elliptical section 
of the tube is well worth noting. The ratio of the axes was 
observed optically, and the semi-axes were found from the volume 
and the ratio, on the assumption that the latter was constant 
throughout. This method, however is, as I have previously 
shewn, satisfactory only when both the ratio and the area of the 
section are actually constant for the whole length of the tube. 
For absolute values of the viscosity the data are therefore not 
satisfactory, in fact Poiseuille’s tubes are the only ones that appear, 
so far, to have been thoroughly measured. 
The imperfect knowledge of the effective radius of the tube, 
however, in no way prejudices the relative values of the fluidity 
for various temperatures ; it can only affect the abso/ute value of 
the viscosity constant. The time of efflux ranges from about 160 
to about 1000 seconds. The former time is rather short for a 
thoroughly satisfactory determination, The general disposition 
