A feature of generalization is offered by ii and iii of secondaries which appear completely separate.

Nymphalida. This term is used in a restricted sense, equivalent to the Nymphalinæ of Comstock, or typical Nymphalids, apparently taken from Scudder.

Nymphalina.-Characterized by the position of i, ii and iii, of hind wings, which spring from one point owing to the fact that ii and iii are absorbed or fused up to the origin of i, which remains nearly constant in all the butterflies examined. This character is secondary in its nature and I have not yet studied the phylogeny of the genera fully. In this subfamily the suppression of the media reaches its widest extent and is only paralleled again in the Attacinæ. In the most specialized forms the cell entirely opens, all trace of the cross-vein vanishes on both wings. Vein iv, becomes radial. Vein iv, leaves upper angle of cell and does not fuse with radius.

Argynnina.-Characterized by the fusion of ii and iii on hind wings not attaining the point of origin of i. No taxonomical features of neuration clearly define the minor groups, which are generally bound together by steps in the grade of specialization shown in the gradual suppression of the media. The "Goat Weed Butterflies" belong probably to the Charaxinæ, a specialized form having lost the "long fork" through absorption, but are not so specialized as the Nymphalinæ or "Purples," as might be inferred by their position in Comstock's Manual. In this work, as well as Mr. Scudder's, the sequence, as based on a specialization of the wings (and no other characters or class of characters allow of such fine distinction) is irregular. In the Check List of Dr. Skinner (1891) the disarrangement is nearly complete.

Agapetida (Satyrida).—Wings (except in the Pararginæ) as in Pieridæ, but vein viii of fore wings entirely absent; vein iii, of fore wings to apex. The veins in many forms show a secondary sexual character in the enlargement of vein ii, the cubitus, or vii at base in male. This character is indicated in the Nymphalidæ, in Potamis and some Fritillaries and in the Ager.

Parargina. The cross-vein of hind wings, or its traces, joins the cubitus; in other words the union of vein iv, with cubitus is complete, since this branch of the media has left the cross-vein. Here there is, in this apparently restricted group, a complete parallelism with the Nymphalidæ, from which the butterflies differ by

the position of vein iii, of fore wings. Cross-vein degenerate between iv, and iv, or cubitus, as might be expected, on hind wings, while on fore wings the specialization has not proceeded so far. Genera: Pararge and Lasiommata.

Agapetina (Type: Agapetes galathea).-Vein iv, of hind wings springs from cross-vein as in Pieridæ and next two succeeding families. All the North American genera I have yet examined (but many remain), and most European Satyrids belong here. The cross-vein is partially degenerate, but as long as vein iv, keeps its position and does not fuse with cubitus this may not here disappear. Vein i of hind wings varies in expression and, almost vanishing in Coenonympha, is quite absorbed in Pyronia. It is diminished in Cercyonis. Probably its study may give us a better arrangement of the European forms. In Eumenis it terminates squarely as in the Pararginæ, and again in Nymphalis. In the other genera it is pointed. Owing to the inequality and slight nature of the specializations in the Agapetinæ, it will require a minute and patient comparison to straighten them out. Any rough classification or sequence attempted on "general principles" must be always nearly valueless. Eneis is evidently a generalized form.

Heliconida.-Study of the type: Heliconius antiochus. As in all the "brush-footed" butterflies, the radius on fore wings is in a five-branched generalized state, while iv, springs from upper corner of median cell. Cells completely closed, the cross-vein merely thinning a little below iv. No trace of vein viii, hence more specialized than Limnadide and agreeing with Agapetida. Vein iv, nearly central, a little radially inclined on fore wings and considerably more so on hind wings, where the cell is small, retreating, the veins long. Vein i determinate, pointed. The radius of fore wings is more specialized than in Limnas, where iii, leaves the stem opposite cross-vein. Here vein iii, arises beyond the cell. A more generalized wing than that of the Agapetida, more distinctly a Limnad type. All traces of the base of media disappeared; no trace of backward spurs from cross-vein.

Limnadida.-Study of the type: Limnas chrysippus. On the five-branched radius of primaries vein iii, springs from a point opposite cross-vein. Vein viii on fore wings present strongly developed. Veins strong; cells closed; a backward spur from cross-vein on fore wings opposite iv,, the position of which is central. On hind wings this vein is slightly radial. Vein i of hind wings imperfectly

fused with radius at base; cross-vein angulate. The curious stigma below v, is attended by a rounded retreat of the vein, which is here slightly swollen. On comparing this type with that of Heliconius it is seen to be the more generalized. To separate Danaus from Limnas we must encroach apparently upon specific characters.

Libytheida.-Vein iii, to costa before apex; cross-vein partially degenerate; vein iv, on primaries central, on secondaries radial; vein viii of fore wings strongly developed as in Limnadida. Outline similar to Polygonia. On secondaries the cross-vein reaches vein iv, just immediately before cubitus. Specialization here almost like the Pararginæ. This isolated group, with its strongly developed labial palpi, cannot be referred to the stem of the Nymphalidæ proper (in sensu mihi) on account of the position of iii, and the presence of viii of primaries. It must be referred back on an independent line to the matrix from which the "brush-footed" butterflies originally sprang. It is now a specialized form as is seen by the extent of absoption of ii and iii, on hind wings, to the point of issuance of i, thus equaling the Pararginæ.

Nemeobiida.-Not a typical " brush-foot," but with the fore feet reduced in the male on the Riodinid type. Special examinations of this structure are needed to bring out the points clearly. Wings of the Pieri-Nymphalid pattern, not of the Lycæni-Hesperid. Radius five-branched, generalized. It is thus impossible to bring the butterfly into the Lycæni-Riodinid series in which the radius is specialized, three to four-branched, while the other neurational features contradict the supposition that it could represent a generalized type of the series. The neuration runs parallel with Libythea and the resemblances lie between this butterfly and Pieris. Vein iii, seems to join costa just before apex. Cross-vein entire, cells closed; on fore wings vein iv, is central, on hind wings radial. Vein viii of primaries seems to be degenerate and I represent it by dots in my original figure. Subsequent studies lead me to believe it wholly or partially tubular. Veins ii and iii of secondaries at base fused nearly to point of issuance of i, hence nearly as specialized as Libythea, much more so than in any Riodinid or Lycænid yet examined. When writing my original paper (in 1896) I failed to note that the family Nemeobiidæ had been recognized, though I have found no description and the study of the neuration seems to have been neglected. To unite this butterfly with the LycæniHesperid branch appears to me a physiological impossibility. It

must rather be relegated to a distinct line, running parallel with the Libytheidæ and leading to the main stem of the Hesperiades. Its affinity with the Pieridae is marked by the position of iv1, which, on secondaries, has left the upper angle of cell and is fused with the radius to a point much beyond the median cell, as in the Pierinæ. Since there is a parallelism in the specialization between the Lycænid group and the Pieridæ in the reduction of the radial branches, a further parallelism might be made to account for this, especially as on primaries vein iv, is fused with radius as in the Thecline. But this will not explain the position of vein iii, on external margin, the radial position of iv, and the more unequal spacing. We might appeal to the imperfection of the geological record and conjure up extinct and intermediate series; but, independent of the fact that such flights of the imagination would lead us nowhere and would excuse even the arrangements proposed by Mr. Meyrick, we cannot do away with the main difficulty, that the wing of Nemeobius is developed upon the Pieri-Nymphalid pattern and that we should not logically graft it upon the Lycæni-Hesperid. The radius is also generalized, five-branched and cannot be derived from a three to four-branched group, which it should have preceded. But the five-branched Hesperiadæ are formed upon another pattern and could hardly have given rise to Nemeobius. The five-branched Hesperiada have most plainly produced the three to four-branched Riodinidæ and Lycænidæ. The wing of the latter is just what we might expect from a reduction of the radial branches of Hesperia. The conclusion we may come to is, that we should seek for the origin of Nemeobius in an independent line, and that the structure of the fore feet has been probably independently acquired. There is no difficulty in this, since aborted fore feet are also characteristic of certain moths belonging to the Hypeninæ, notably of Pallachira bivittata Grt. There seems to be a latent tendency in this direction which has broken out strongly in the day butterflies.

GENERAL COmparisons.

Before entering upon any comparison as to the amount of specialızation in the Pieridæ and the "brush-footed" butterflies (=Nymphalidæ of Scudder and Comstock) it will be well to get a mental picture of the neuration of the Pieri-Nymphalidæ as a whole. This can best be obtained by contrasting it with that of an allied wing group in the same structural series, the Lycæni-Hesperidæ. Inde

pendent of relative breadth or shape of wing we have in the latter a simpler pattern, the veins more equidistant, an indisposition to fuse and furcate shown by the retention of a central position by vein iv,; so that as the suppression of the media takes its course this branch tends to degeneration in situ, from resisting the attraction of either radius or cubitus. As opposed to this we have a willingness in the Pieri-Nymphalidae to preserve vein iv2, which latter tends everywhere to becoine radial, except in the isolated case of Leptidia, where it becomes cubital. We have a spreading of the veins and abundant traces of unequal specialization. Except in the lycænid reduction of the radial branches, the Lycæni-Hesperiada offer few neurational changes to aid our formation of classificatory categories; the Pieri-Nymphalidæ plenty. United by the presence of the looping vein viii, or its traces unequally expressed and sometimes quite vanished, the Hesperiades offer in this way two groups characterized by the peculiar neurational wing pattern; giving us also an instance of parallelism in specialization, in that the Pieridæ sustain an analogous position with regard to the "brush-footed" butterflies (Nymphalidæ, etc.), to that the Riodinid-Lycaenids show with respect to the Hesperids or "Skippers." In both these groups the reduction of the radius takes place; the Pierids still showing phases embracing and intermediate between the five and three-branched radius, while no five-branched Lycænid is yet known to me. Thus the gap in the Lycani-Hesperiade between the subgroups is greater than that between the subgroups of the Pieri-Nymphalidæ. But the fact that the reduction of the radial branches has been independently taken up by the two main wing groups of the Hesperiades comes clearly out. I have been unable to find any characters which will always distinguish the neuration of the Hesperiades from the moths. Not so with the Parnassi-Papilionidæ, a distinct major division entirely left out of sight in the present studies.

Having thus endeavored to trace the outlines of the neuration of the Pieri-Nymphalidæ as a whole and to enable the reader to grasp more or less fully the wing structure of this waste of butterflies, we may more in detail compare the wings of the "Whites" with those of the other butterflies in their group. That the radius is specialized in the Pieridæ and generalized in all the other families is the first and obvious difference, one which strikingly throws the balance of specialization to the side of the "Whites." So that in this direction of secondary specialization, which the Pierida share with