THE DEVELOPMENT OF THE NERVOUS SYSTEM, 481 
resemble those exhibited in my sections, but I am unable to 
regard the large cells as a layer of neuroblastic cells, although 
the nervous system is undoubtedly differentiated from this 
layer, and I think it probable that each ganglion is formed 
from a small number of these cells. Many of the large cells, 
perhaps the greater number, are probably the origin of the 
muscles, and originate from the ccelomic sacs (Pl. XIV., Fig. 2) ; 
and certainly, in the earlier stages of the development of the 
Blow-fly embryo, it is not possible to distinguish the cells in 
which the ganglia of the ventral chain are first seen imbedded, 
from those which subsequently form chains from which the 
muscle fibres of the larva are developed. 
Hatschek, Will, Heiden, and others, have described a median 
inflection of the cutaneous epiblast, between the two lateral 
chains of ganglia in various Arthropods. Kowalevski did not 
find any such inflection in Hydrophilus, nor have I seen any- 
thing of the kind in the Fly-embryo. It appears to me, from 
the very advanced condition of the embryos figured by Hatschek, 
that the involution is secondary, and is perhaps a rudimentary 
entosternum. To those who accept Dohrn’s views, the existence 
of such an involution is of primary importance, if it is developed 
previously to the ganglia and really enters into their formation : 
but the evidence derived from the study of the Blow-fly embryo 
appears to me to be quite incompatible with such a supposition. 
Balfour, commenting on Hatschek’s statements, says: ‘ It is 
probable that Hatschek was entirely mistaken as to the entrance 
of a median element into the constitution of the ventral cord ;’ 
but in this I think Balfour is wrong. There is considerable 
evidence that a median element is present in certain stages of 
development; even at an early stage, a median inflection of 
the caelom is seen in certain of my sections, with a distinct 
mesoblastic wall. It is probably from this that the trachea 
and connective elements, which exist in the median septum 
of the neuroblast of the larva, originate. 
At a later pericd I have obtained excellent sections, showing 
the ventral cord (as represented in PI. XIV., Fig. 2) entirely 
separated from the surface epiblast. In some of these sections, 
