Compensatory Regulation. 87 



*'the reversal in some cases at least does not take place or is incom- 

 plete." Brues (Wilson, '03, p. 210) adds the interesting fact 

 that in A. heterochelis the nerves supplying the two chelae and the 

 ganglionic centers from which they proceed do not differ percep- 

 tibly in size. 



2. The Data. 



The experiments about to be described in the present section of 

 the paper were performed at the Naples Zoological Station in the 

 winter of 1902-03. They confirm the general facts of reversal 

 of the chelae as given above. Their main object, however, was 

 the determination first, of the effect of the removal of one or both 

 chelae upon the rate of moulting of the animal, and, second, of the 

 influence of the presence or absence of the opposite chela upon the 

 rate of regeneration of a chela. 



a. The Influence of the Removal of One or Both Chelce upon the 

 Rate of Moulting of the Animal. Three sets of specimens were 

 operated on. In one set {Sn) the snapping chela alone was 

 removed. In a second {Cu) the cutting chela alone was removed. 

 In a third {Sn + Cu') both snapping and cutting chelae were removed. 

 The animals were kept in isolated dishes for 59 days after the 

 operation and were fed either every day or every other day on 

 small pieces of fresh fish meat. Without taking into account the 

 cases where the legs were accidentally autotomized a second time 

 during the experiment, or in which other disturbances occurred, we 

 have the following relation between the time of moulting and the 

 post-spinous thoracic length of the animals without reference to the 

 character of the operation: On the coordinate paper (Fig. 28, 

 p. 86) the abscissae represent the thoracic lengths in millimeters 

 and the vertical columns (ordinates) the days after the operation 

 when moulting occurred. The animals were killed 59 days after 

 the operation. The data from the first set {Cu) are represented 

 by the symbol o , those of the second set {Sn) by the symbol and 

 of the third set {Sn + Cu) by X. It will be seen that in general 

 the moulting interval increases with the size of the animal as repre- 

 sented by the thoracic length. 



On pp. 88 and 89 the data are put in Tables X, XI and XII, 

 each of the sets being placed by itself. The interval of time in 

 days between the first moult and the second moult is put down 

 in a separate column. Upon averaging this interval in the three 



