214 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY. 
In Figure 6 we have a surface view of a portion of this network (mbr. 
pr.), formed by the nucleated stellate cells, and in Figure 5 the membrane 
is seen edgewise. In Figures 4 and 5 it is to be observed that this mem- 
brane is widely separated from the hypodermis as a whole, though from 
nearly every hypodermis cell a greatly elongated process (prc.) reaches 
down to and unites with the Grundmembran. This separation of the 
two cell layers, and the consequent formation of a space between them, 
were also observed by Semper, but he appears not to have observed that 
this space contains lymph corpuscles. I find that these corpuscles are 
present in large numbers, and this fact complicates the question as to 
the origin of the formative cells of the scales, 
a question to the dis- 
cussion of which I shall return later. The presence of lymph corpuscles 
between the Grundmembran and the hypodermis renders it probable 
that the former is not, as Semper supposed, absolutely uninterrupted. 
However, I am not able to affirm from personal observation that there 
are direct communications between the sub-hypodermal spaces and the 
chief lumen of the wing. It is rather a matter of inference. 
In another particular, too, my observations add to the account given 
by Semper, for at rather regular intervals the Grundmembran of one 
wall becomes continuous with that of the other by means of hollow 
tubes (¢b.), which seem to be formed by the folding of the membrane 
itself (Figs. 4 and 5). The cavities of these tubes are direct continu- 
ations of the sub-hypodermal spaces of the upper and lower walls of the 
sac, which are thus put into communication with each other at frequent 
intervals, and it is worthy of note that leucocytes are frequently found 
within these tubes. 
Slender thread-like prolongations of the hypodermis cells (pre., Figs. 
4, 5, 7, and 8) are seen to extend inward from the hypodermis to the 
Grundmembran, as already stated. Each of the hypodermis cells gives 
rise to one, and only one, of these processes. The process is rather 
sharply marked off from the cell from which it arises ; but after tapering 
rapidly for a short distance it is prolonged into a thread-like structure 
many times as long as it is thick ; this diminishes slowly in calibre until 
it reaches to and fuses with the Grundmembran. It is probable that at 
first the hypodermis cells are simply converted into columnar epithe- 
lium, and that the sharp distinction between cell body and cell process is 
brought about only at the time when the process becomes greatly elon- 
gated. Occasionally a hypodermis cell is seen without any such process. 
The wings are still hardly more than simple outpocketings of the 
general hypodermis of the chrysalis. In fact, in the larva itself the 
