44 R. HOSKING. 
The above values for C were used in calculating K in 
equation 5, for the four tubes. R/L was also calculated. 
These values are collected in the following table. 
Tube I. Tube II. Tube III. Tube IV. 
K (mean) | ‘005504 | -005463 | -005480 | -005478 
RIL 003407 | 002908 | 003774 | 003172 
It is evident that there is no linear relation between K 
and R/L. When the above values are plotted in the way 
already mentioned, it will be noticed that they lie along a 
straight line parallel to the axis of abscissee; that, there- 
fore, k in equation 5 has zero value, i.e., the value for n is 
zero. See Figure 6. 
A set of readings was taken at 25°C. andalso at 0°C., 
and values for K.; and Ky were found. The results are 
tabulated below, and the corresponding graphs appear in 
Figure 6. They bear out the conclusions arrived at in 
connection with the results at 50°C. 
Tube I. | Tube II. | TubelIII. | Tube IV. 
ee 00896 | 00893 00892  -0089 
ge / :01791. | -01790 01790 | -01791 
R/L 003407 | -002908 | -003774 | 003172 
In the general reduction formula, the most difficult con- 
stant to measure accurately is R, the mean radius of effiux. 
Capillaries are not generally right circular cylinders, nor 
even elliptical cylinders; and as the degree of precision 
with which R must be calculated is always four times as 
great as that required in the deduced viscosity, the examin- 
ation and measurements of the capillaries must be carried 
on with extreme care. 
Tubes I. II. III. and IV. were in the first place selected 
from a large number on account of their uniformity of bore 
