upon the Velocity with which it is traversed by Light. 255 
water in the centre of the tubes, therefore, must be greater than 
that of the water near the sides, and consequently also greater 
than the mean of both velocities. 
Now the slits placed before each tube to admit the rays whose 
interference was observed, were situated in the middle of the cir- 
cular ends of the tubes; so that the rays necessarily traversed the 
central zones, where the velocity of the water exceeded the mean 
velocity*. 
The law followed by these variations of velocity in the motion 
of water through tubes not having been determined, it was not 
possible to introduce the necessary corrections. Nevertheless 
analogy indicates that the error resulting therefrom cannot be 
considerable. In fact, this law has been determined in the case 
of water moving through open canals, where the same cause 
produces a similar effect ; the velocity in the middle of the canal 
and near the surface of the water is there also greater than the 
mean velocity, It has been found that, for values of the mean 
velocity included between 1 and 5 metres per second, the maxi- 
mum velocity is obtained by multiplying this mean velocity by a 
certain coefficient which varies from 1:23 to 111. Analogy 
therefore permits us to assume that in our case the correction 
to be introduced would be of the same order of magnitude. 
Now on multiplying w by 1°1, 1°15, and 1:2, and calculating 
the corresponding values of the displacement of the bands, we 
find in place of 0°20 the values 0°22, 0:23, 0°24 respectively ; 
whence it will be seen that in all probability the correction would 
tend to cause still greater agreement between the observed and 
the caleulated results. We may presume, then, that the small 
difference which exists between the two values depends upon a 
small error in estimating the real velocity of the water; which 
error cannot be rectified in a satisfactory manner, in consequence 
of the absence of sufficiently accurate data. 
Thus the displacement of the bands caused by the motion of 
water, as well as the magnitude of this displacement, may be 
explained in a satisfactory manner by means of the theory of 
Fresnel. 
It was before observed that the motion of air causes no per- 
ceptible displacement of the bands produced by the interference 
of two rays which have traversed the moving air in opposite di- 
rections. This fact was established by means of an apparatus 
which I will briefly describe. 
A pair of bellows, loaded with weights and worked by a lever, 
impelled air forcibly through two parallel copper tubes whose 
extremities were closed by glass plates. The diameter of each 
* Each slit was a rectangle 3 millims. by 1°5, and its surface was equal 
to one-fifth that of the tube. 
