AT THE BOUNDARY OF A LIQUID IN MOTION. 
581 
the observations marked (4) the bulb was suspended in the silvering solution for about 
an hour, and then removed. It was then found that the top hemisphere was covered 
with a black sediment from the solution on top of the silver, while the under hemi¬ 
sphere was silvered as usual. Nevertheless a series of three determinations of the 
logarithmic decrement was taken. The mean of these shows an increase in both the 
logarithmic decrement and in the time of swing, even though the temperature was 
actually higher. It was thought that this might be due to the black sediment, which 
could not be removed alone ; so the whole deposit was dissolved off and the bulb 
resilvered, the solution being kept stirred and frequently changed. This time the 
deposit was bright and uniform. The observations were, however, identical with the 
last. The only explanation of this (unless we suppose that the friction is about 3 per 
cent, greater for silver, instead of 4 per cent, less as Piotrowski deduced) is to suppose 
that a change had occurred in the suspension. To test this, a series of observations 
(6) to (9) were taken with the silver on to get the temperature correction, and 
immediately after (9) the beaker of water was removed from under the bulb, and one 
of nitric acid of the same temperature put in its place wdthout disturbing the suspen¬ 
sion. As soon as the silver was dissolved the bulb was washed, and the beaker of 
water replaced. 
A series of observations (10) were at once taken with the glass surface. Thus by 
comparing (9) with (10) we get a comparison of the friction on glass with the friction 
on silver, free from all possible errors due to change of suspension, and at tempera¬ 
tures whose difference is only 0°‘25. The means are— 
Temperature. 
Time of swing. 
Logaritlimic 
decrement. 
(9) Silvered. 
9-75 
9-898 
-20718 
(10) Unsilvered. 
10-00 
9-938 
-20681 
The temperature correction for the logarithmic decrement is by (8) and (9) '002815 
for 1° C. or -00070 for 0°-25. 
Therefore the logarithmic decrement for the glass surface corrected to 9°-75 is 
•20751. 
The ratio of the frictions is n ofw ^ ^ — 1‘00564 : 1. 
9oyo X -.iU/io 
Thus the effect is, if it exists at all, less than O'G per cent, instead of 4 per cent. 
Another independent comparison can be taken between the series marked (6) and 
(7) and the series marked (11). The means are— 
