UEACTION IN ASTERIXA GIBBOSA. 427 



the righting is carried out with great precision and rapidity, 

 often in 30 sees, or less. The two arms on which the starfish 

 rotates are at the end of the turn disposed at a wide angle with 

 one another, a condition recalling the usual method of righting in 

 Ophiuroids, where the animal pushes itself over on two arms 

 stretched out nearly in a stiuight line. The wide angle is 

 immediately compensated for when the starfish liegins to crawl 



If, however, the two arms turn their ventral surfaces in one 

 direction, so that the tips get a twist in either a clockwise or a 

 counter-clockwise direction, the turning is delayed and other arms 

 must be brought into play. In general, a disposition of the arms 

 with the tips pointing in the same circular direction is a very 

 unfavourable condition for turning, and much adjustment is 

 required before turning can be carried out. 



The turning on two arms is very often delayed b}^ the holding 

 on of one or two other arms, which later detach themselves. 

 The "unified impulse " of which Jennings (1907) speaks in his 

 elaborate analysis of the righting I'eaction in Aster las forreri is 

 certainly shown in Asterina, but nppears to arise I'ather late in 

 the I'eaction. I have often seen the third arm detached rather 

 forcibly by the pull of the turning pair. In other cases, how- 

 ever, one can observe a voluntaiy retraction of the tube -feet of 

 the impeding arm or arms. That the unified impulse does play a 

 great part is shown by the fact that, in general, a bad start is 

 always or almost always compensated and a, good solution arrived 

 at, for whatever number of arms is employed at the beginning 

 the turn is almost invariably made on two adjacent arms. If the 

 unified impulse did not at some stage arise the righting reaction 

 would be a long drawn-out and confused affair, coming only by 

 chance to a successful issue. 



A second neat solution of the lighting problem is afforded 

 when the righting is carried out mainly by one ai'm, helped 

 towards the end of the reaction by the arms on either side. This 

 solution is rare, and the method is not a particulaiiy speedy one, 

 the times taken ranging from 37 sees, to 120 sees. 



A third method sometimes adopted is to turn on three adjacent 

 arms. On conrlition that the middle arm of the three takes the 

 lead this metliod is quite successful : it is, in fact, hardly to be 

 distinguished from the method of turning mainly upon one arm. 

 If, however, the middle arm lags behind and the outer arms get 

 ahead of it, trouble ensues and a deadlock occurs 



Turning on four arms is really not met with. Four arms may 

 firmly attach and an attempt be made to turn on all four, but the 

 actual turning must take place on tlie middle pair, the outer arms 

 being forced to let go. An attempt to turn on four resolves 

 itself into a turn on a single pair. 



If all five arms attach, no progress can be made until at least 

 two relax their hold. 



Comparison with the account given by Jennings of the methods 



