654 
PROFESSORS A. W. REINOLD AND A. W. RUCKER 
In this case, 
and 
dy\+dy, + dy^0, 
7 , 7/7 7 1 + 2 cos 2^ 
da + dp^-cly, . 2sin ^ ~ 
This formula affords a means of deciding whether any slipping of the liquid ring 
takes place. We have carried out the above calculations to the second degree of 
approximation, but, as the numerical work is troublesome and the correction intro¬ 
duced is small, we here neglect the squares of small quantities. 
A vertical millimetre scale was placed by the film box, and the reading on this was 
determined, which corresponded to the middle of the distance between the cups to 
which the films were attached. The diameters were then measured not only in this 
horizontal, but also in planes 1 cm. above and below it. 
The values of Y, dy x , dy 2 , and dy 3 were thus found, and c/a-f-d/3 was determined 
without any assumption as to the nature of the attachments of the films to the cups. 
Since the pressures exerted by the films on the internal air were the same, we have 
T + clT __ 
2Y + clu + d(3~ 2T + dec' + d A’ 
( 18 ) 
whence c/T/T is found. 
It can also be calculated from the measurements of the principal ordinates and from 
the sensitiveness. If the results of the two methods agree the assumption made in 
the earlier part of this paper that the generating curve of the film passes through the 
edges of the upper and lower cups is justified. 
In making an experiment with a thick film as the standard of comparison, it is 
necessary that the same amount of liquid should always be used in flooding, and that 
the measurements should always be made at the same interval after that operation. 
In the experiment now to be described the flooding took place between each pair of 
readings near the top, middle, and bottom of the films. One minute elapsed between 
the flooding and the beginning of the measurements, and three minutes between con¬ 
secutive floodings. The readings were taken in the order: top, bottom, middle, bottom, 
top, &c., so that the measurement of the principal diameter corresponded to the mean of 
the two for the top and bottom, from which it was in point of time equidistant. Any 
error due to slow changes in the position of equilibrium was thus as far as possible 
obviated. 
The following results were obtained when the left film was colourless, and when 
C,= {16, Y(2,9)} 
