140 University of California Publications in Zoology. [Vol. 4 



tachecl tube feet of these rays causing the starfish to turn a direct 

 somersault. 



Frequently the first pair of rays that become attached take 

 the awkward first or second position. If now a second pair of 

 rays becomes attached in the more favorable third position, then 

 usually these determine the direction of turning-. The ill- 

 adjusted pair are released, and the animal turns on the pair that 

 can pull in a coordinated way. Similarly, if two adjacent rays 

 are first attached in one of the uncoordinated ways, another 

 single ray may become attached in such a manner as to make a 

 coordinated pair M'ith one of the first two. Thereupon the star- 

 fish is likely to turn on this coordinated pair, releasing the third 

 ray, or moving it merely in unison with the other two. 



This factor, like all the others, is not always decisive. Often 

 the direction of the impulse is determined by some other factor 

 before the rays have become attached. Then lack of coordination 

 in the rays may not change the direction ; the starfish persists 

 in its original direction, it turns awkwardly, twisting and un- 

 twisting the rays in various ways during the process. 



Another conceivable factor should perhaps be mentioned here. 

 Often when the starfish is resting on its ventral surface in the 

 normal position, some of the rays lie close together, while others 

 are far apart, with a wide angle between them. When such a 

 starfish is turned on its back, it might apparently take hold and 

 turn itself with the two rays that are wide apart, much more 

 readily than in any other way. One might therefore anticipate 

 that the direction of righting would be predetermined by the 

 position of the starfish before it had been turned on its back. 

 But experiments show that this is not the case. As soon as the 

 starfish is turned on its back, its rays begin to wave back and 

 forth, sideways as well as up and down, so that the original ar- 

 rangement of the rays is quite lost before any of them become 

 attached. The starfish is not more likely to turn on the two rays 

 that were widely separated than on any others. 



(11) Experimentally releasing attached rays. — The direction 

 of turning may sometimes be determined in the following way : 

 Three or four rays are attached, some tending to pull in one 

 direction, some in the other, — the unified impulse not having 



