THOMPSON: METAMORPHOSES OF HERMIT CRAP. 
181 
usually within the first forty-eight hours of its life as a glaucotlioe. 
What then is the significance of the shell in the ontogeny? It is 
clear that it cannot be regarded as an essential factor, but may it 
not perhaps influence the development to a lesser extent ? 
If such an influence were present it obviously might act in two 
directions: to affect the rate of the metamorphosis, or to modify 
the order of the metamorphosis and possibly, but less probably, the 
resulting structures themselves. In addition to this, despite the fact 
that the dextral asymmetry is fixed in the ontogeny, appearing before 
a shell is taken and completing itself even when the latter is absent, 
it is conceivable that the entrance into a dextral shell which nor¬ 
mally occurs at a definite point in the development may have some 
controlling influence, so that exposure to other than dextral shells 
might not be without effect on the course or integrity of the meta¬ 
morphosis. 
Effect on the rate of development .— In connection with our first 
possible effect of the shell, that on the rate of the development, the 
influence, if any, will involve the glaucotlioe more than a later 
stage as this is the period of most extensive change and the time 
when the shell is normally assumed. Therefore a number of experi¬ 
ments were performed on glaucotlioe by which they were exposed 
to different environments relative to the time when the shell was 
taken and also to the shape of the shell, whether dextral, sinistral, 
or uncoiled. 
For these experiments the age of the larvae used, had to be known 
within as narrow limits as possible. So fourth-stage zoeae were col¬ 
lected and placed in a large jar, where during the night many would 
moult to the glaucotlioe. New zoeae were used each day, and in 
the morning the glaucotlioe were removed and reared in one or 
more experiments as the case might be. For I made it a rule never 
to mix glaucotlioe of different ages, but kept each batch separate 
from all preceding ones. The experiments were examined daily at 
the same morning hour. At first, an additional mid-period record 
was taken, but it was found that the larvae moulted chiefly in the 
“small hours” of the morning and this habit rendered a mid-period 
record less necessary and as the pressure of other work became 
severe it was abandoned. The fairly definite time of moulting also 
served to reduce the unavoidable time error which might otherwise 
have had a range of twelve hours. The large number of specimens 
