247 
1914-15.] The Eeflexes of Autotomy in Decapods. 
worked out. The corabined result of these reflex contractions must be 
to weaken the leg at the region of the breaking-plane, and so to facilitate 
removal by plucking on the part of the chelipeds. If any one of the 
muscles be divided, autotomy cannot take place. 
It is thus seen that, when the hermit crab is removed from its shell, 
the normal unisegmental reflex is not sufficient to produce autotomy. 
It must be reinforced by spread to other segments, and participation of 
effector organs belonging to these. 
Something may be said of the nature of this spread. Firstly, if one 
cheliped be damaged it is promptly plucked off by its fellow of the opposite 
side, only a few seconds intervening between the times of injury and of 
autotomy. The movement is always well aimed. In the second place, 
if the next pereipod be damaged, the aim is not so good, and it may take 
the chelipeds almost half a minute to seize and remove the offending stump. 
Sometimes it seizes the wrong leg and pulls violently at it, but removal 
never results, probably because the unisegmental reflex contraction of the 
basal muscles has not taken place, and therefore instability at the breaking- 
plane has not been produced in this sound limb. If, thirdly, the seat of 
injury be in a limb innervated by a segment of the nervous system still 
further removed, quite a time may elapse before the leg can be plucked 
off, and many attempts on sound legs may have been made during this time. 
The speed and degree of certainty of autotomy, therefore, vary inversely as 
the distance of the innervation centre of the injured limb from the region 
of the chelipeds. Another point indicating the faulty means of connection 
between different levels of the nervous system is seen in the behaviour of 
the damaged leg. When the chelipeds seem to be doing their utmost to 
get hold of the stump, this part is calmly going on with its rhythmic walk- 
ing movements, and instead of bending up to meet the chelipeds it always 
seems to be in the process of retreat when the claws are about to seize it. 
These plucking movements of the hermit crab have been mentioned by 
Morgan (1) in one of his papers ; but as his observations do not agree with 
those described above, some discussion is necessary. He, in the first place, 
takes no account of the fact that the crab is in a most abnormal condition 
when it is autotomising outside its natural shelter. I have shown that 
the reflex removal of damaged limbs is unisegmental in normal circum- 
stances, but that reinforcements by arcs of other levels is necessary when 
the crab is removed from its shell. There is no need, therefore, to invoke 
complex instinct when dealing with the matter. 
Furthermore, Morgan states that plucking only takes place in those 
cases where there is a definite breaking-plane at the base of the injured 
