No. 395.] THE WINGS OF INSECTS. 851 
thicker layer of hypodermis on its dorsal side, and the thinness 
of the hind wings steadily increases with their expansion in 
area. The hind wings are greatly expanded at the final trans- 
formation, while the elytra are almost as large in the pupa as 
in the imago. 
Comparing now the two types of wing development (external 
and internal), we see that, despite great superficial differences, 
there are important common features. In both cases the 
wings arise in early life and form a double plate-like fold of 
hypodermis, between whose layers trachez shortly penetrate. 
In the former the extension of the wings is gradual and mod- 
erate, excepting at the time of transformation; in the latter 
they early settle down into deep hypodermal pockets, in which 
their extension is of necessity retarded, although cell multipli- 
cation seems not to be. 
The principal structural elements which enter into the mak- 
ing of the insect wing are hypodermis, trachez, nerves (which, 
though always mentioned and once or twice figured by other 
students, we have rarely seen in wings), leucocytes, embryonic 
cells, and, possibly, sometimes fat cells. Of these, the first two 
only are essential structures; and these are so important as to 
merit special treatment. 
II. THE ORIGIN OF THE TRACHEATION OF THE WING. 
In wings developing externally like those of a dragon-fly one 
sees the principal trachez passing very early out into the wing- 
bud, branching freely and forming by multitudinous terminal 
anastomoses a network of capillary tracheoles. In a horizontal 
section of a nymphal wing one may see how the branches of 
the tracheze are formed. Fig. 84, D, is from such a section. It 
will be observed that the terminal tracheoles are intracellular, 
the trachez intercellular; but that there is easy transition from 
one condition to the other.' 
In a wing so mounted that the tracheal system is filled with 
1 There are no such distinct transition cells between trachez and tracheoles as 
Holmgren found (Arat. Anz., vol. xi, pp. 340-346) in the spinning NRE of cater- 
pillars, although the cells at a, 4, c might seem to stand in the same relatio 
