AT THE BOUNDARY OF A LIQUID IN MOTION. 
5G5 
Time for iinsilvered tube, corrected for changes in 
temperature and radius. 2 13'59 
Time observed for silvered tube. 2 12'82 
A difference of — 0'6 per cent. 
Another series was made with the same tube which gave as the mean of ten 
observations for each state ;— 
t // 
Unsilvered. 2 18'78 
,, corrected. 2 15‘92 
Silvered, observed. 2 16'51 
A difference of + 0'4 per cent. The silver was very thin, the change in being 
OTO per cent. 
Thus, as the result of four series of observations with three different tubes, we have 
that the difference in the times of flow for the silvered and unsilvered tubes is never 
greater than 0’7 per cent. With both the first and second tubes, and in one series of 
observations with the third tube, the time of flow is slightly greater (by 0'2, 0’7, and 
0‘4 per cent.) for the silver surface, while in one case—the first series of observations 
with the third tube—the time is slightly less (by 0‘6 per cent.). 
These differences are all within the limits of experimental error, found by compar¬ 
ing the times of flow for the same tube in the same state on different occasions. 
On the whole there is some evidence that the time is a little greater for the silvered 
surface, as would, of course, be the case if the deposit was not quite strictly uniform. 
In the one case, when the time was less, the temperature difference was largest, and 
errors likely to be most important. 
These experiments may at any rate be considered conclusive against the existence 
of the large effect, for the existence of which I was searching. 
A new series of observations was then undertaken to determine whether any 
slipping occurred in a silvered tube when the velocity of the water was greater than 
before, and the gradient of velocity was pushed near the limit beyond which the 
motion ceased to be linear. 
This limit was calculated for each tube by means of a formula given by Professor 
Osborne Peynolds,* who found by experiment that in order to insure linear 
motion, 
Dt’yo/p, must be < 1400, 
where D is the diameter of the tube, /r the coefficient of viscosity, p the density, and 
* ‘ Phil. Trans.,’ 1886. 
