No. 1.} VARIATIONS IN LIMULUS POLYPHEMUS. 33 
brates renders any relation between them at first sight very 
improbable. But the unexpected discovery that in this par- 
ticular arthropod the appendages are frequently invaginated, 
leaving in their place a series of slit-like gill openings, such 
as those shown in Fig. 1, is a potent factor in favor of my 
view. 
If in Fig. 8, the thoracic appendages had been provided 
with gill leaves like those on the abdominal appendages, we 
would then have, in place of typical arthropod appendages, a 
series of respiratory sacs, the cavities of which, by the persist- 
ence of embryonic conditions might, after the manner of 
vertebrate gills, communicate with the alimentary canal, either 
through its nephridium and somite, or by a secondary opening 
at the apex, where ectoderm and mesoderm appear to be con- 
tinuous. The exact relations in Limulus of the invaginated 
appendages to the mesoblastic somite, and of the latter to 
endoderm have not been determined, as in the cases studied 
the cavity of the somite had disappeared. But there seems to 
be no reason to doubt, from what occurs in the normal embryos 
of Limulus and other arthropods, that something like the 
condition mentioned above might arise, for each of the assumed 
conditions is known to occur. 
The great difficulties involved in determining what is ecto- 
dermic, mesodermic, or endodermic in the vertebrate head, we are 
just beginning to realize. The use of the terms is founded on 
the supposition that the embryological processes in vertebrates 
present a modification of those assumed to occur in some real 
or imaginary invertebrate, which is further assumed to be a 
more or less remote ancestor of the vertebrates. While the 
problem is still in this uncertain condition, the fact that these 
‘a priori’? methods of interpretation do not harmonize with 
the suggestions here made cannot be used as an argument 
against them. 
II. ABSENCE OF APPENDAGES. 
We shall consider all embryos showing the absence of one 
or more appendages under the above heading. In all cases 
the reduction of an appendage seems to be accompanied by a 
