106 G. H. KNIBBS. 
of the tube itself. Fig. 3 illustrates the method applied to 
Poiseuille’s series B and F, the former giving a negative and the — 
latter a positive value for n, viz. about — 5°2 and + 11-2, these q 
however, especially the latter, being subject to considerable 
uncertainty, as is manifest from the figure. The opposition of 
the sign of mis very evident in the following values for the | 
viscosity (7) computed without regard to the corrective term a2. : 
R . 
Tube. . xX 10° (1) Tube. : x 10° (7) 
B 56 013202—« CF t=«<HCité«“ OA 
B 75 3134 F, 163 3065 
B, 115 3070 F, 326 3249 
B, 240 3002 oe 646 3967 
BD, 630 2742 ¥, 1254 4891 
Bb, 1455 2193 F; 3034 4851 
These results challenge the propriety of Couette’s statement that | 
1 may be always regarded as positive and taken as nearly three 
times the diameter of the tube.2 His deduction seems to 
entirely based upon the fact that the indications of Poiseuilles . 
experiments with tubes A, and A,, happen to be in fortuitous 
agreement with the results of his own ingenious method of simul | 
taneous and equal flow through two tubes of equal radii, but of 
different lengths. In order to adequately test the question, we 
indicated hereunder, reducing them rigorously by the formul® 
given hereinafter for flow in tubes elliptical in section and conical 
longitudinally. Whenever possible the values of 17’, were found 
taken as 1-12 for the correction, and these instances are indical 
by an asterisk in the table hereunder. The results, arrang? 
Dividing them therefore in parcels of eight, we obtain 
identical mean values for the viscosity, whereas if Cou 
So 
1 Rejecting Fs: see the figure. 
2 Ibid p. 504-505, 
