. S. S. MAXWELL 351 



other ampulla whose canal is still horizontal. On the other hand, 

 rotation of the animal around its longitudinal axis (in the new plane 

 of the canal) never produces a deviation of the eyes to the left as it 

 might be supposed to do if the rotation causes a current in the canal 

 and the current excites the ampulla. This experiment, then, shows 

 conclusively that the excitation of the sensory structures in the am- 

 pulla is due to some other cause than the production of a current in 

 the canal. 



Since no further consideration need be given to the possibility of 

 currents in the semicircular canals as the cause of the excitation which 

 on rotation gives rise to the reflex compensatory movements we may 

 consider other possible causes. These might be (1) effects dependent 

 on the inertia of the mass of liquid or other material in the vestibule, 

 or (2) due to the inertia of the contents of the individual ampullse, 

 or (3) to inertia effects within the sensory cells themselves. It would 

 be impossible to decide between these a priori. 



In my earlier experiments I found that after destruction of the 

 structures in the vestibule I could never obtain compensatory move- 

 ments on rotating the dogfish around its dorsoventral axis. For a 

 long time I was inclined to think that the absence of the reflex was 

 due to some sort of injury to the ampullae, although these appeared 

 to be as sensitive as before to direct mechanical stimulation; the 

 slightest pressure caused decided eye movements. When, however, 

 I was finally able to remove the otolith from the recessus utriculi by 

 slitting open the utriculus lengthwise without tearing it across I found 

 that the compensatory movements to rotation in the horizontal plane 

 were not abolished.^ Since the destruction or the transection of the 

 utriculus abolished the reflex with no apparent reduction in the direct 

 sensitivity of the ampulla it became clear that the utricular (and pos- 

 sibly the saccular) structures are essential parts of the mechanism. 



In attempting to analyze more closely the arrangements of the 

 parts concerned it is to be noticed that the movement of rotation which 

 acts as a stimulus to any given ampulla carries foremost the side of 

 the ampulla which bears the crista. Thus the cristae of the anterior 

 canals are on the lower side of their ampullae and a rotation of the 

 head downwards excites them; the cristse of the posterior canals are 

 also on their lower sides and a rotation of the head upward (back part 



