are probably the least specialized; as they possess a frenulum, they cannot justly be derived from the Lasiocampidæ themselves, but have probably a common ancestor not far removed. From this original family are derived four branches, viz. (1) the Bombycidæ (this name has often been wrongly applied to the Lasiocampidæ, but here denotes Bombyx mori, the "silkworm" moth, and its allies) and Saturniadæ, which have lost the frenulum entirely; (2) the Notodontidæ, Polyplocidæ (Cymatophoridæ of some), and Sphingidæ, stout-bodied forms, whose larvæ are commonly furnished with various prominences; (3) the Uraniadæ and Epiplemidæ, in which veins 6 and 7 of the fore wings are normally stalked; (4) the several families formerly called Geometrina, whose larvæ have usually lost two or three pairs of prolegs.
Lastly, the Caradrinina contain seven families, of which the Ocneriadæ (Liparidæ of some) is doubtless the most ancestral, making in fact a close approximation in many points to the Psychidcs, and showing a tendency to exhibit similar apterous females. In this, and the allied family Hypsidæ, vein 8 of the hind wings is connected by a bar with the middle of the upper margin of the cell. In the Agaristidæ, Caradrinidæ, and Plusiadæ (these two latter forming the old group Noctuæ, whose name is untenable, belonging by right of priority to an owl), this is modified so that 8 anastomoses with the cell-margin very shortly near base, the Agaristidæ being characterized by the apically swollen or sometimes clubbed antennae, the Caradrinidæ by the obsolescence of vein 8 in the hind wings, which in the Plusiadæ is well-developed. In the Arctiadæ a further modification takes place, 8 anastomosing with the cell-margin for a considerable distance from base. In the Syntomididæ is reached the extreme of change in this direction, 8 becoming wholly absent by coincidence with the cell-margin and 7.
In this scheme the Caradrinina, Notodontina, Papilionina, and Tortricina are all terminal developments, i.e. growths which lead to nothing beyond themselves, and in translating this scheme into a linear form it would be possible to take any one of these as top, and the other branches in any convenient succession. But, considered as a whole, the Caradrinina, from the difficulty of sharply defining the families (which implies comparatively little
