132 
When the cocks A and C are open the mean velocity becomes 
548.1, with B and D open 553.6. The halved sum of these numbers 
is 550.8. In observing the interference fringes we did not measure 
the displacement from a zero position, but determined the total 
displacement with reversal of the current, which therefore must be 
proportional to 550.8. 
In my former paper I accepted Una, = 553.6 cm./sec. We now 
obtain a value for the velocity Une, differing only by '/, percent 
from the value used in Communication Il and giving excellent 
agreement between the experiments and the formula with the dis- 
persion term. The difference between the two expressions under 
consideration amounts for the wavelength 4500 A. U. to quite 5 
percent. We therefore conclude that also as to the absolute phase- 
difference the results of Communication Il remain largely in favour 
of the LORENTZ expression. 
The comparison between theory and observation now has become 
very simple, Ynce being measured directly. A separate determination 
Ob. pel, YY is avoided. Finally bowever, the value of this ratio 
mar be 
may be calculated from the results obtained at the pressure of 2.14 
K.¢./em?, v, = 465 and woar — 550.8. This gives for the ratio 0.844. 
This mean number is not, however, a general physical constant but 
a constant of the apparatus. The course of the curves in Fig. 2 and 
Fig. 3, suggests for a long tube a final value of g perhaps 1 or 2 
percent lower than 0.844. *) 
The formula for the displacement of the interference fringes must 
l 
Ld 
heneeforth be written with a factor runs ‚dl instead of the simple 
0 
product Unar-/3 Vmax being a function of the distance to the origin 
of the moving fluid column. 
1) [Only after finishing my investigation, | became acquainted with the important 
memoir on fluid motion in pipes by Drs. Sranron and Panne xt. (Phil. Trans. vol. 
214. 1914). From their data the often mentioned ratio appears to be 0.82 for my 
case. There is no conflict between the two cases, as their observations were 
made after the passage of a length of pipe varying from 90 to 140 diameters. This 
length is sufficient to enable any irregularities in the distribution of velocities to 
die away. In my apparatus this ideal is largely departed from]. (Note to the 
translation). 
