256 Proceedings of the Royal Society 
angular rate is kept up for some time, the rate appears to the 
experimenter to be gradually diminishing ; if the rotation be then 
stopped, he experiences the sensation of rotation about the same 
axis in the opposite direction. If the position of the head he 
changed after the prolonged rotation has been made, the position 
of the axis of the apparent rotation is changed, remaining always 
parallel to a line in the head which was parallel to the axis of 
the real rotation. The readiness with which this complementary 
apparent rotation is produced is not the same for each axis. In 
such experiments, as long as the eyes are shut, and the axis of 
rotation kept vertical, a sensation of giddiness is not experienced. 
That sensation appears to be caused by the discordance between 
the testimony of the sense of sight and that of the sense of rotation. 
It is obvious that this sense must have a peripheral organ physi- 
cally constituted so as to he affected by rotation, and that it must 
be such as to receive different impressions when the axis, direc- 
tion, or rate of rotation is changed. These impressions must be 
transferred to the ends of afferent nerves, and by these nerves 
conducted to a central organ. 
The semicircular canals of the internal ear are eminently fitted 
by their form and arrangement to act as the peripheral organ of 
this sense. I shall consider first the action of one semicircular 
canal, and for simplicity suppose that there is only one. Starting 
from rest, let us suppose rotation of the head to take place about 
an axis at right angles to the plane of the canal. The bony canal, 
being part of the skull, of course shares in this rotation, but 
the perilymph lags behind, and thus the membranous canal, which 
floats in the perilymph, does not immediately follow the motion 
of the bony canal, but, as the membranous canal is continuous 
at both ends with the utricle, the relative motion of the bony 
and the membranous canal must produce a pulling or stretching 
of the forward end of the membranous canal. If this is the end 
at which the ampulla is situated, such stretching will necessarily 
move the terminal nervous organs in the ampulla, and may reason- 
ably be expected to stimulate the nerves. This stimulus will no 
doubt be greater the stronger the pull, i.e., the more rapid the 
rotation. We should thus with one semicircular canal have the 
means of perceiving, and of estimating the rate of, rotation in 
