THE LAW OF RESISTANCE IN PARALLEL CHANNELS. 
947 
This last result was brought out in the most striking manner on reducing the results 
by the graphic method of logarithmic homologues as described in my paper on 
Thermal Transpiration. Calling the resistance per unit of length as measured in the 
weight of cubic units of water i, and the velocity v, log i is taken for abscissa, and 
log v for ordinate, and the curve plotted. 
In this way the experimental results for each tube are represented as a curve ; 
these curves, which are shown as far as the small scale will admit in fig. 9, present 
exactly the same shape, and only differ in position. 
Pipe. 
No. 4, Lead . 
ti 
5 ) 55 • 
A, Glass . . 
B, Cast iron 
D, 
C, Varnish . 
Diameter. 
m. 
0-00615 
0-0127 
0-0496 
0-188 
0-5 
0-196 
Either of the curves may be brought into exact coincidence with the other by a 
rectangular shift, and the horizontal shifts are given by the difference of the 
logarithm of 
5! 
o 
/* 
for the two tubes, the vertical shifts being the difference of the logarithms of 
D 
n 
The temperatures at which the experiment had been made were nearly the same, but 
not quite, so that the effect of the variations of /x showed themselves. 
15. Comparison with Daecy’s experiments. — The definiteness of these results, their 
agreement with Poiseuille’s law, and the new form which they more than indicated 
for the law of resistance above the critical velocities, led me to compare them with 
