between the viscosity of liquids and their chemical nature. 437 
from the true value of 17, as deduced from Poiseuille’s observations with long tubes 
in which the coiueclions are negligible. 
These differences Colette attributes to the effect of the magnitude L. Let 77^ be 
the constant value obtained for -q by the above process in the case of tube A, its 
value in the case of tube B. On assuming that L has the same value for both tubes, 
seeing that the diameter of each tube is the same, from the preceding equation the 
true value of -q should be given by 
/r+"L = /3T L- 
This was found to be the case; the true value of 77 thus deduced was identical with 
that obtained by Poiseuille in the case of long tubes. On solving for L, its value 
was found to be 2‘868 times the diameter of the tubes. 
From a set of experiments made by himself in which the diameter was seven times 
that of the tubes used by Poiseuille, Couette found that L was 3 ‘2 times the 
diameter. Couette concludes from this evidence that, unless the velocity of efflux be 
considerable, the value of L is constant and about three times the diameter of the 
tube. Probably its magnitude becomes somewhat larger as the diameter increases. 
The conditions of velocity in our observations and the diameter of our tube are 
similar to those for which the above constant value of L seems to hold, and, on 
applying the correction //(Z -h L) to our results, after correcting for kinetic energy, 
they would be diminished by about 1 per cent. We find, however, tliat without 
applying this correction for L, our observations give results for water identical with 
those of Poiseuille. 
It need not follow, however, that our results are in opposition to the idea that the 
source of error denoted by the correction is inoperative. We are rather inclined to 
believe that its effect is made negligible in our case by another source of error which 
acts in the same sense, and which it seems impossible to allow for. The radius of our 
tube was obtained by weighing with mercury, a method which gives low results, due 
to the film of condensed air which invariably i)rtervenes between the mercury and 
the glass. The radius thus obtained will be too small, and seeing that the fourth 
power of the radius is employed in determining 77, the percentage error made in 
observing B will be magnified. 
For this reason, we have neglected the correction for L in deducing the value of 77; 
this was indeed the only course which appeared to be open. The correction L is an 
experimentally determined quantity, which was introduced in order to make observa¬ 
tions taken with short tubes coincide with those taken by the long tubes of Poiseuille ; 
m our case this coincidence exists without the use of the correction. 
The formula used by us to obtain 77 was thus 
77 = 7rBkp/8W — pYlSrrlt, 
