428 



SCIENCE 



[N. S. Vol. LIII. No. 1375 



and found that no disturbance of righting 

 reactions resulted. Ewald (1892) plugged and 

 cut the canals in the pigeon without throwing 

 them out of function. 



I have ligatured, cut and plugged the 

 canals in the dogfish (1910) without in the 

 least disturbing their functions. But I have 

 put the canal current theory to a more 

 decisive test (1919). In the dogfish the com- 

 pensatory movements to rotation in the hori- 

 zontal plane are mediated only by the ampullae 

 of the horizontal canals, and each acts only 

 for rotation toward its own side. If one 

 horizontal ampulla, say the left, is removed, 

 rotation to the right around the dorsoventral 

 axis causes both eyes to deviate to the left, 

 but rotation to the left has no effect. I ex- 

 posed a right horizontal canal for nearly its 

 whole length, cut it through as near as 

 possible to its posterior connection with the 

 vestibule, then without injuring its connec- 

 tion with its ampulla I raised it up and fixed 

 it in the vertical plane at right angles to the 

 longitudinal body axis. It is evident that 

 with the canal in this new position rotation 

 in a horizontal plane could not cause a cur- 

 rent in its endolymph. But rotation to the 

 right around the dorsoventral axis actually 

 caused both eyes to deviate to the left, while 

 rotation to the right (or left) around the 

 longitudinal axis, that is in the plane of the 

 new position of the canal, did not cause such 

 a movement. It is evident that under the 

 conditions of the experiment rotation in a 

 horizontal plane could not possibly produce a 

 current in the canal, and hence the stimula- 

 tion must have been produced in some other 

 way. 



Since no further consideration need be 

 given to the idea that the excitation on rota- 

 tion is due to currents in the semicircular 

 canals, we may briefly consider other possible 

 causes. These might be effects due to (1) 

 the inertia of the mass of fluid in the vesti- 

 bule, or (2) due to the inertia of the ampullar 

 contents, or (3) due to the inertia of the 

 sensory cells themselves. The second and 

 third of these possibilities are eliminated by 

 the fact that when the membranous connec- 



tion of the utriculus has been cut off the com- 

 pensatory movements to rotation in the hori- 

 zontal plane are entirely absent although the 

 motions can be as easily elicited as ever by 

 mechanical stimulation. 



Since the transection of the utriculus 

 abolishes the reflex it is clear that the 

 utricular (and possibly the saccular) struc- 

 tures are an essential part of the mechanism. 

 It is to be noted that the direction of rota- 

 tion which acts as a stimulus to any canal 

 is that which carries foremost the side of the 

 ampulla bearing the crista. In looking over 

 the large number of figures given by Eetzius 

 and by A. A. Gray I find no exception to this 

 rule. The mouths of the canals at their 

 ampullar ends are so connected with the 

 vestibular parts of the membraneous laby- 

 rinth that the inertia effect of the mass of 

 liquid (endolymph and perilymph) in the 

 vestibule must cause an increase of tension on 

 the part of the ampulla bearing the crista 

 when a rotation is made in the direction in 

 which the crista leads. A careful examination 

 of the anatomical relations will show that 

 even if it were possible for the inertia effect 

 of rotation to cause a movement of liquid in 

 the canal and thus exert a pressure on the 

 cupula (the hair cells of course could not be 

 acted upon directly), a much greater effect 

 must be produced on the membranous sti-ue- 

 tures in the vestibule. The relatively large 

 mass of liquid in the vestibule with its pro- 

 portionally small surface area exposed to the 

 friction of the walls must show more inertia 

 effect during rotation than the small amount 

 of liquid in the canal with its proportionally 

 large area of contact with the canal walls. 

 The membranes which form the sacculus and 

 utriculus are virtually sti-etched through the 

 mass of liquid in the vestibule and must 

 necessarily be put under tension when any 

 rotational movement is given to the liquid. 



If the above conception is the correct one 

 it should be true that change of tension and 

 not change of pressure should act as the 

 stimulus. I had previously shown by experi- 

 ments on Phrynosoma (1912) that the pres- 

 sure due to centrifugal force has nothing to 



