UNIVERSITY OH AT non'^^-^ ^'2RAR- ^' U'^SANA-CHAMPAIG.^i NATURAL HISTORY SURVEY NOTICE: Return or renew all Library Materialsl The Minimum Fee tor each Lost Book is $50.00. The person charging this material is responsible for its return to the hbrary from which it was withdrawn on or before the Latest Date stainped below. Theft, mutilation, and underlining ol books are reasons lor discipli- nary action and may result in dismissal Irom the University. To renew call Telephone Center, 333-8400 UNIVERSITY OF ILLINOIS LIBRARY AT URBANA-CHAMPAIGN C ' '' ^ '- A PvEYISlOX LKPIDOITEIUHIS FAMILY SPHINGIDAK. 1-1 e^ \i)\ irAi'Ks ZdDi.ocrr i:. \oi . IX. Siti'I.kmi A REVISION L EPIl )( )PTEROr S FAMILY SPHINGIDAE. Hon, WALTER ROTIISC'IIir.r). Pir-D. AND KAEL JOKDAN, M.A.L., Ph.D. {WITH r,; PLATES.) Issued at the Zoologkai- ]\Iusfum. TniNc;. ^FAiiriT 190:1 riilXTFB T'.Y IIAZETJ,. WATPOK & VINKY, \a\. I.ONPOX AXD AVLKSnUTiY. 1903. v.l Introduction CONTENTS L riKNKHAL fSlBJECT XV II. Systematic Section I 111. ( ATALOlilE AND 1nI>K.\ 8l0 INTRODUCTION. /\ll{ ret;eiu\'lies in Spli'uKjiilnf, which liiivc liceii ciu-riril mi i-oujoiiitly tor ^"^^ scvfral years, uiul uf whicli we now imblish the resiiils, were uiiderlakeii with liie imriiose nl' giving ii suund basis to the classitictitioii of these insects hy uu extensive simly of their iuor])hology. Chissitictitioii is an interpretation of facts. The facts are to a great extent details uf I lie anatomy and morphology of the beings classified. As if, is, therefore, largely eirennistantial evidence wliicii guides the classifier, the first step towards a correct classification is to find out as many facts as [lossibie. In interpreting these facts or characters presented by the individuals— the indi- vidual is the basis of all research — one starts with the assuni[ition tliat what has lieen found to lie true in the necessarily limited number of specimens investigated, liokls gooil also in the vast multitude of individuals not compared. The possibility of an error in this respect can be lessened by the comparison of a large material of individuals. How large it should be, nobody can iiredict. To ascertain the extent of variation of the chief classificatory unit, the sjiecies, the material is never too extensive. On the knowledge of the extent of variation of the species of a family depends the stability of the superstructure of genera, tribes, and subfamilies. All ideal classification could be drawn up, if all the species were known which are in existence and which have been. As this premiss cannot be fulfilh^d, we have to be content with the species that are known. And here, again, the foundation of the superstructure will be the safer the more species have been examined. ^\ e have endea\(iured to comply with these three primary demands on a classifier a> far as it was ])ossible for us : many facts, many individuals, many species. We have not restricted ourselves to a comparison of the pattern, general appearance of the body and wings, and of the neuration, but liave taken into the scope of our research every part of the skeleton of the imagines, and hence have given a broader basis to our conclusions than has ever before been done in this famiiy. .Since the structure of the Hawk Moths has never been studied to any extent — the Sp/dnyu/uc lieing in this respect one of the most ( viii ) 1ii-;::1ccIim1 I'aiuilies of large insects- we cnulil sciirwly asnid discdvi'i'iiig organs liithi-rto nvcrloolu'd, and tlimwing a new light on others which, though known to exist, hatl no! hcen studied e(in)|iai'alively. Tiie organ of friction on the ciasper and eiglitii lergite, fouml in many HI" I IS : \\w structure of tlie inner surface of the [lalpi ; the devfdo|iniCMl of the liiiifrr and the aniennul I'ud-segiuenl ; tiie special stiuctui'e of the iiienini of tlie nieso- anil luetacoxae : the diverse deveiopiuent of the alidoiuinal spination ; tlie reduction of (he pul villus and paronyciiium of the ciaw-st'gments ; certain dillerences in tlie end-segment of the antennae ; the pecidiar mid- and liindtarsal coiul), etc., have ne\er been taken into account in the elassitication of the Sj)/iiii(/i(l(tr. However, we were very mueli hampered in one I'Cspeet. We should have liked to extend our researciies in the same degree to I lie early stages. The Sj)liiH(/i(/<ic being for (lie greater part exotic sjiecies, the larvae and pupae are known only of a comparatively very small proportion. We were failed nearly everywheiv liy a lack of material. Though we have tried, during the years devoted to study of these insects, to obtain larvae and jnipae from all countries where we have corres])ondents, we had very scant success, and scarcely any success ill procuring the first larval stages. The larvae and juipae of many of (he commonest Hawk Moths are still unknown i[iuintities.'' Nevertheless, our studies of the early stages Lave not been i^uite fruitless, inasmucli as tliey showed us that it is us dangerous to generalise from a few specimens or species in the case of larvae and pujiae as it is in the case of imagines. The studies proved to ns on the one hand that the deductions commonly drawn irom the characters of European Splunyidae are faulty in many respects, and on the other hand that one meets in the larvae and pnjiae witli similarities in not nearly related species and conspicuous dissimilarities in close relatives, .just as one finds such puzzles in the adults — puzzles which are stumbling- blocks lor the classilier, and a source of pleasure for an impiisitive mind. The figures of the larvae and pupae are, with few exceptions, of little use, it drawn by amateurs who do not recognise and emphasise the essential points. The illustrations of pupae especially leave much to be desired. We draw attention to the following particulars, which we venture to hope will not be so often neglected in future by the artists who are trying to give an As the iirilinary. iioii-residoiU, coUectur in tlif Tropics slums iiillatiag larvae uii aeetmiit ut tlie lime it occupies and the trouble it gives him, we think it ailvisable to call attention hcic to the following very simple mctboil of preserving xmall larvae. Put the larva, after it is suffocated Ijy benzine or chloroform, into a gla-ss tube heated over a flame. The specimen will confraiM, then e.xijand and burst, an<l dry in this expanded state. ],et the tube cool and take the larva out. In absence of a glass tube, one may use a piece of tin or anything that will stand heating. ( i'^ ) ;iili'(iMutc j)ic.liux' i>r u clinsiilis : position of tiiu gla/A'd |iiirlioiis ot' tiu' eye ; lcii,i;tii of the lonyiie ; [Ulterior femur (extorually visiiih' or iml): slriu^ture uC sligiu;ilie,;il ureas of the iilnl()iueii ; shape ami ariuature of the ereiiiasler. The luuuliei' of iiKli\iiiiials and speeii's of Siil,in(j'nhic c-oiitaiiied in the IVin.i;- Alusenni is eonsideralde, there lieing in the collection ni'arl\- Hi, nun speci- niens, lielonging to (.)(i<i-odd s]iecic^. Though tin's inalevial is sastly larg<'r I han thai contained in any other collection, it was nevertheless not sniiicieut to form the basis of a thorough revisiou of the family. Since many of the species of Spliuiijidiii' are very dilHcult to distinguish, and tlierefore the descriptions and often also the tignres not exact enough, it was necessary for us to examine the sj)ecimens on which the names proposed by previous authors were founded, in (U'der to find out how many species are known and which are the [iroper names for them. It has been our good foi'tnne to have our ap[ieal for help addi'i^ssed to private collectors ami museums responded to with the greatest lib<u-alily.-' ^Vilhout this kind assistance it would have been altogether impossible to disentangle the synonymy and to firing the species which were insulticiently described iido their proper place in our classitication. We gratefully acknow- ledge liei'e the help received from the authorities tif the British Museum, of the Museums at Oxford (Hope Department), Dublin, I'aris, Uruxelles, Berlin, Dresden. Jliinehen, Stettin. < 'openhagen, Stockholm, Vienna, Madrid, New York, and Brisbane : and we are no less grateful for the kind assistance rentlered by II. Druce. h. W. Distant, W. tScliaus, Colonel Swinhoe, M. ('. Piepers, r. ( '. T. Snellen, G. Weymer, G. Semper, Dr. A. Pagenstecber, A. Bang-Haas (coll. Staudinger), P. Doguin, P. Mabille, Dr. W. J. Holland, and Charles Obertliiir ; who all either sent us photographs, types, and other specimens, or allowed us to visit their collections and to study the material contained therein. — Mdiinia.f colleyis gnitius .' An important point for the satisfactory progress of our work was a com])arison of the long series of types of Walker's and Boisduval's descriptions, contained respectively in the magnificent collection of Mons. Charles Uberthiii' and in the British Museum. A closer study of these specimens than had hitherto been attempted was absolutely necessary. For Boisdnval, when visiting the P>ritish Museum in the forties of the last centiu'y, had named in manuscript and made notes upon the Si^jldiujidae of that collection, which names were for the greater ])art adopted by Walker in is.")'), but oftcLi ;i|)plied Id other s))ecies tha,n those for which Boistluval had intended them to stand. This muddle became intensified by I'oisduval, who, in his monograjjli ])id)lished * Only two letters of iii(|iiiry have been lult ui)aii>\vtre(l. The names oi the aildre-sseos may be passtd ijvLT in silence. (■ X ) ill l^To, j^iivi' (lc's^.•l•iJlli(lll^ lalvi'ii tVniu lii> own s|ii'(iiuciiN iuid ii|i[ilieil the bei'ure- luc'Utioiieil iiiiiiuisi'ri|il iiiuues to spucics wliirli In' liclievcd to Iju iIiu insects JR! liiul SO luiiiicil ill the lii'itish IMusi'uiii, lull wiiicli were (jt'tcii iiol llu; siuuc. Moreover. I'»iisilii\;il raiicil lo ieeo,^nise luany of I lie W'aikei'iaii species, and descrilied llieiu auaiii iiiuler new iiaiiie>. 'I'lie eonriisioii liiii> occasioned luis, we lio[ie, lieen siicce.s.sfull \ cleared n|i in I In' |iresenl, Kevisioli. 'I'liere are '.'>* species conlained in this Kevision. Of these \vv. have iiol seen llie iollowiiiL;- : — ll'lliiirii.s, /riiiii/;i, \k \'-\-i : known lo lis iVoni llic descrijil ion. — Kansas. LiijKirn iiiiuinii, p. I.ij : known lo lis IVoni the de>criplion and ligures. — N. York. Lnimni. /iiilii-nnuKc, p. l.')o ; luiown lo lis iVoni liic description and figure. — Florida,. I'ol'ijifi/i-h'i.'i (jiiiiili, p. '^Ah ■. known to ii> I'roiii I he description and tigure. - \V. Africa. SiHci i nth alas {'^) i/ciiiniins, |i. :iii:J : known lo n.s IVoiii the descriplion and a sketch.— iSikhiiu. S"ta.sj>ct< riljljci, p. -JT4: known lo lis I'roiu llie description and lii;iin'. — ('elelie^. I'lujiroscrpiniis cutn-pc, p. lil.i; known lo us lixuu the descripl ion. — ( 'alitbriiia. Arctonotas tcrlooi, p. (i'Hi ; known to us I'roiu the descrijition and ligiire. — W. Mexico. Jlijijiotiou Ijutkri, \). 7GU ; known to us from the description und ligiire. — Madagascar. The names of wliich we have seen the tyjies are marked with an asterisk (*) in the bililiography. Since Linue and Fabricius the Sphiiujida.c of the glolie have been five times ckissitied, revised, or catalogued. Hiibner, in his Ver. bcIi. SchinettUmii: (about l>i2'l), was tlie hrst to propose a detailed classitication of Lepidojitera. Tiie work was, in sjiite of all its glaring mistakes, far in advance of the time, and was therefore almost entirely neglected by the contem]iorary entomologists, and sub- sequently forgotten or treated as not being worth consideration. \\'e do not see any reason for rejecting the generic names j)ublished by Hiibner in the Verzeichnisa. The deliuitions are insufficient and often incorrect, and the species considered geuerically the same belong often to widely diilerent groujis, while close allies stand widely separated. That is ipiite true : but the Ijadness of the classitication and of the definitions is — perha|is unfortnnately--no valid argument against the adoption of the names. if it were, we should likewise have to reject a multitude of names proposed b\ more recent authors, whose definitions ( xi I do nut iiiiply 1(1 the specii^s guuericiilly ilctiued, coulaiiiiiig iTroiieous iuid niiiLe misleiidiug shitcmeuts, di- whose genera couliiiu very lieterugeuedus eleiueiils, as dti, tor iiishuiee, many of Walker's genera. Tliu\igli, in tlie original detiniLion of rntlop<irrr (lype: rH.sllai), Burmeister slated that the iKi[ia had no jirojeeting tongiie-case (whieh it has) ; though Staudinger erroneously said of his new genus DolbiiM thill it had only one spur to the hiniltihia, and lluwe made a similar mistake in (he definition of Smcriitthalus : and though the defitnliou of Moore's llothid is so vague as to ujiiiiy to a liost of other Siiliiinjidu,: as well (as many of Jloore's definitions doj, — these names eau and will not be rejeeled on that aeeonnt. There is no line to draw between good and bad definitions, sullieient anil insulKcient deseiijitions : and every descripliou is incomplete. The rirsi after Hiibner to treat again upon all the SiiJunnidac was Walker, who, in the LlM of Lc/iit/optrra llflcroreni of tin: Urilislt Miisruin vol. viii. (18o<i), gave deseri[itions of till the known genera and sjieeies and uuiuerons new ones. His bibliography is generally good, but his deseriptitjns are often so bail that it is im2)0.ssible to recognise the species without seeing his specimens. He has lieen much attacked on the Continent, and his names liave been ignored to a certain extent by a few authors. Walker did not attempt u classificatiou of Ihe S[ihiii<fi(lai\ lie simply descrilied the genera in the order he thought proper, without Ijringing them into groujis. His genera are very ofteu as unnatural as many of liiibuer's. Boisduval followed in February of In?.") with a monograph of the I'auiily, containing more exhaustive descriptions of the species ami genera. The Sjjhinyulac are divided in this work into a number of subfamilies, of which definitions are given, rather a rare occurrence in lepidopterological works of that time. In the nomenclature of the subfamilies he followed the old French custom of employing scientific terms in a gallinised form — a bad custom, which was formerly in vogue also in the nomenclature of species and genera. The monograjih, though far above that of Walker, had two great drawbacks : it did not contain all the species described up to l5;74, aud names already emjiloyed by \\'alker were used again for other species (see above). We mention inci- dentally that Boisduval's names have priority over those of Butler which were published in the Proc. Zool. Sor. Loud, of 1875. The Uevision of Spkintjltlor by Butler — which came out early in isTT, not in l8Tii, as is ((Uoted by some authors — is scarcely more than a synonymic list with occasional remarks. The genera are grouped into four subfamilies, but not defined, except the new ones. Though the definitions of the subfamilies are Inised almost entirely on the ijuite imperfectly known early stages, the grouping is, nevertheless, an advance on Boisduval's classification. The work would have ( xii ) iicpii luiicli iuii>n)ve(l if Hoisdiiviil's s|)i;cies iiml .ueiiL'ra hail \>wn iiicor]ioi;ttr(l into Mr- body of (lie Kevisioii iiistciul ol' being given as an apiJendix. The yonngesl work on I lie .V////?//y/c/rtc of the gioliu is contained in Kirliy's <'a/fi/or/'ic of LepiiloptiTd I Ictiivrfni, (IMI:.'). A« a list of names this cataiugne has been of great help to ns. The dassitication adopted in it has been mneh Munied liy some anihors us b('ing arljit rary. Bnt we think tliat one sinjnid not exjieel loo ninch fiuni a catalogne. Even the liest is full of errors, as a eataloguer of in.seels cannot possibly have iiitriiisically worked out all the groups catalogued. Besides these live general works, there are numerous treatises dealing with (lie Sj/Ziini/if/ac of certain restricted districts. Apart from a host of poimlar handbooks, there are two works on Ihe I'alaearetic Hawk Moths wordiy of special notice. Tiicse are liy Ikrtel, in Kiilil, ( i roti-^iscli m . vol. ii., and liy Tutt, I hit. Li' II. vol. iii. Hartel gives lengthy and generally accurate descriptions, but relies (uo much on others, whose errors he rejieats without having examined the insects himself and formed his own ojiinion. Tutt's work is of quite a dilferent kind. It is the most intrinsic ever written on Palaearctic Lepidoptera. The third volume comprises only a portion of the Spliiixiiilnc ; the remainder of the family will appear in the fourtii. The work will be uf the greatest help to the scientist who knows the matter well enough to lie alile to distinguish between what is scientific and what appears merely in a scientific garb. The usefulness of the work could have been improved, we think, liy a condensing of the contributions of the collaborators, and mistakes could iiave lieen avoided by the omission of references to foreign sjiecies with which the respective collaborators were not sufficiently acquainted. However, as it is, there is nothing written anywhere on European Lepidojifera coming up to it in thoroughness. The Indian •Spliingidat' are dealt with by Hamiison in Blanford, Fmina Brit. India, Moth» \(il. i. (1892). The volume should be consnlfed with some caution, .since many distinct species are treated in it as being identical. ■flie species occurring in the Philippines are contained in Semper, >ivlniidl. I'l'ilijjp. vol. xii. (I'^i'li), where many figures of larvae and pujiae are given. Miskin gave a catalogue of the Australian Hawk Moths in the I'ror. No//. Soc. Queiui.'iland vol. viii. (ISHl). A nionograpli of the North American SjihiiKjiduf by J. Smith is contained in the Trans. Amer. Knt. Soc. vol. xv. (188S). It is the liest work on Xearctic Hawk Mot lis, though the classification is faulty in many respects, owing to Sinitli's limited accjuaintance with the forms not found in North America. The Cuban sjiecies are described and catalogued by Gmte and Robinson in the .Joar. Ent. Soc. Plnludelphiu vol. v. and vi. (18G5. IftlOT^, and again by ( xiii ) (iiiiullarli in his i'oiilr. Kiil . ('»iliimn (ISsl). Tlic lunucv iiapcr was tlir host wvitton on SiJiiinjIiliK- np to tliat timo. The C't'iifral Anu'rican Hawk Moths arc onnmei'atcd and partly described and figured li)' H. Drnce in tlic liiol. Cfiitr. Aiiu'rirana, l.c/i. I let. riS.>*li — ISOO). Till' sjiories occnrring in the Aru'iMitine l{o]>nl]lic arc dcah with i>y I'nrnieister in his Prsrr. Rr'/i. Anji'litiiir Vol. V. (ISTS) and Atliis (l^T'.lj. Tin' di'scriptinns and lii^-nres of the earlier stau'es arc ol' inipoi'tniiee, Imt several mistakes in identitication oi-cnr. Then; is no list ol' the Alriean s)ieeies. Resides these more imi»)rtn,nt. win'l;s, which are pnrely systematic, tlii're are treatises oi' another mitni'e, dealini;' with the markings and eolonr of lar\a,e, like' Weismann's Stvdiis hi Ihr 'I'/n'orii ol' Drscciil, and Piepers's pa.per mi the laTvae of i<,plihiijiihtc pidilishcd in the Tijiliichr. Ivnt. for lsi)7. All the systematic works referred to siilferei] from a, la.clc of knowh^dgc of the raorjiliology of the S/ihiii(ii<l<ii'. Nenration, the organ jrencrally r<died upon in the classilieation of IjCjiidoptcra, ln'ing of little assistance in tlie llawlc Moths, tiie authors seized upon any superficial characters, and thus were led astray. We hope to fill up that gap in our science by the present Revision- at least to a certain extent. Nobody can lie more aware of the incompleteness of our researches than ourselves. The subject is far too large to allow of being treated after a, comparatively small number of years of study in any degree ap]iroachiiig completeness. Nevertheless, we may fairly claim to have given a new fonndaiion to the study of the Sjihiiifildnc. The groundwork for future researches is tlici'c : future authors will more clearly see tluM'r way, and be able to concentrate their ciforts on the elucidation of the numerous points only cursorily touched npun in this He\ision. We have generally abstained from giving detailed description of any stage of the known species ; but the keys to the genera and species, anil the indication ol' some trenchant feature of shape, colour, pattern, or structure under each species, will, we trust, be an (efficient guide also to those who wish to use the work as a means of determining the names of the material in their collections. Though we hope not to have missed any naines, we have not attempted to give a coiiijilrff bililiography of all the sjiecies. The work is divided into three ]iarts : — I. General Subject. II. Systematic Section. Ill, ("ataloo'ue and Index. I ( x\' ) I. GEXETJAL STn'.JECT. Tlll'i resoavclies cniboilioil in ;i work like; tlio ]irospiit, avo of two vory dilTorent kinds. "\Vo liad, h'lstly, to study the insects dealt with ; and, seenmliy, to study the names l)estowed upon tlietn Ijy previons authors. It may sound almost ridienlons, Imt it is nevertlieless true, that in many cases the time one h.-is to spend over the nouienclature of a form, in order to clear np nomenelatorial muddle and to find out what form autlifirs have meant to desi,!i-nn.t(> with a oerta,in name, eipials or surpasses the tiiue oik^ can, ioi- certain reasons, devnte to the study of the uaturnl history of tlie aiiimnl. Surely this is wronu'. The Natural History of tlie a,nim;il lieinn' the sul)ject of our science, the acci>ssory suhjcci of nomeu- idature should nevei' have a,ssnmed such mn-u'uirude. it is waste of energy. However, we have patiently to hear the i'rnits of the sins of our forefathers in science, and those who eouK^ ai'ter ns will a^-aiu mutter bad langun.i;'e. Oni' may kick, but one has to sutfer. We have no sympathy with those of our contem- porari(^s who eontril)ute unnecessarily to the burden, which is in itself superfluous, and detracts from the eiforts devoted to our science. The system of naniing the groups of individual specimens of animated nature lias been invented as a helj) to tlu^ student of science, but it has been carried ont from the beginning in such a way that it necessarily developed into a bother as well. Why? Because the PRINCIPLES OF NOMENCLATURE were not strict enough. Science is a republic wliere everybody may do as he likes. Tiiere ;ir(^ no laws which can be enforced ; and nobody can ln' ])revent(>d from ]iublishing what lie jdcases. Tliis fi-eedom is a great boon to st'icnce. f'uibrtun.ately,' the results of scientific research and those of nomenclature are of ipiite a difFer<'nt standing. If tlie i)nrported results of scientiiic investigations prove to be (M'roneons, they ari' rejnidiatod and forgotten. If somebody ]iro])ouuds "laws'" of develo]iment wliich arc^ ibun<l to he err(nieous, science jiasses on witliont tronbling any longer about them. If somebody considei-s the battledore scales of /.//c/ij'ttii/oi' to lie fiiiKji, or the maxillary jialpi of Vitlicidnv to be antennae, such statements are disproved, and are no further encumbrance to science. Statements of fact, and conclusions, once proved to be erroneous, no longer occnpy the time of the scientific student : science is rid of them. Science can never get rid of a name of an animal or plant once ])ublished — unless quite a different system of designation be adopted than that employed since the time of Linne. We cannot simply ignore a name which is a record ol' an animal or plant. For we mnst keep a record at least ol' all the forms which have become known to science, since we cannot have a record of all the forms that exist nnd have existed. Even names which are synonyms cannot be dropped ; they must ( ^vi ) 111' ciirricil nil fill- two reasons. l>'irstly. if tlioy \vov(> drojijiod and forpjotfen. tlicy would in many casos ho. oniployrd ajjain for sonn'tliiiiL;' olsr, and thus land lis inevitably in a ninddle. Secondly, closer researeh often proves that what was considereil flic sanu' at one time is really different. A form may lor a lont; time 1)0 lost si<;lit of. Imt scientists will sooner or later become aware of tiie oversight, if the name is kejit on record. For instance, in onr case, tli(> liawlc Moths described by Liinu' and {''abrieins respectively as s^)/,/„.r ///i/r/irt and horrhiiri'ir have been treated us the same for aliont ]",'m years. AVIkmi we looked np the original records, we tonnd I hem to relVi' to two wi<lely dillerent insects belonginj^ to dillei'ent <;'enera. However, if it is granted tliat if is necessary, for the sake of completeness ol' onr knowledge, to keeji on record all the names uiven to forms of aninnils und phints, it will also be conceded tliiit it is an nnjnstitiable act because it adds unnecessarily to the burden— to suppress a mum' ;ind I'ephu'i' it by another. Some of the older writers did not seem to think much of recording an already named species under a new name ami treating the older nam(> as a synonym. I'^abricins — a great and intluential man in his time— set a very bnd example to his ibllnwers not only by his insufHcieut deseriiitions, lint especially by his arbitrary changing of names. For no reasons whatever he superseded names given by |)rnry. Cramer, and others, by names of his own invention, and eniploycMl - Worst of all — the rejected names for other s]iecies, thus entangling the nomenclature to such nn extent that it is difficult to find one's way throngh the impasse. With such an exnmjile before them, one caimof wonder that others followed suit. Pxiisduval especially seems to have fonnd great jdeasnre in his names being iirinfed. One cannot helj) smiling when one linds him coolly re])lace Pa/iilio curlK'nor by a new name, '■'■ (i.rloii I'oisd.," and sees the mannserijit-names which he had tiesfowed at one time or tlie other njion S/)li>iiiii(l(ir appear in his monograph of the family under species which had meanwhile been ba])tised by others. It may be comforting to an author who conies too late to be nevertheless able to lannch his names on the scientific world, but it should not lie done. AVhen Science was in its infiincy, a little l)laying like this may have licen pard(niable, Imt nowadays there is no excuse whatever for playing at nomenclature. " The species described by dones as cotiforml.'t stands in my collection under the name of dsrll'is mihi," or something to that effect, is not only a foolish thing to jniblish, but is an intolerable crime, which should always be met by an energetic rebuff. Vanity has something to do with this kind of proceeding, though there is really nothing to be ]irond of in giving a. name to a sjiecimen and .avoiding criticism by shunning jiublicity. Hut we d<i not (piito understand what is the object of those who are busy jiublishing mannscript-names which are given by others and which they tind on specimens in collections. As it is of no .advantage whatever to science, whether it becomes known or not that a bird or butterfly which has a valid name stands in this or that museum under this or that manuscript-name, there must be some other reason for wilfully increasing the list of synonyms. Is it to jn-ove ( xvii ) that tlie respective antliors of the matmscript-uaraes were too hizy to write out a description and make it pnldic, or that they were not sure if tiie forms named were really new? Is it to show that the respective authors who gave names to individual specimens of one and the same species really did not know enough of the things they baptised ? Is it to demonstrate the carelessness of the respective authors who bestowed, in the collection, a name on an animal for which a name had already been published ? Surely if the authors of the names had intended to publish them, we may leniently assume that they would have found time to reconsider the matter before rushing into print. We should not pry into the private foibles of others, and thus detract from their fame. Only jiublislied matter is common property, which scientists are bound to critically examine. We have seen many collections with numerous manuscript-names, but we are glad to state that the bad habit of naming specimens in collections without troubling about publishing a descri{)tion is very much on the decrease — at least among scientific systematists. The habit has come down to us from a time when few people worked at the same grouj). There is another class of no less objectionable names which gives little credit to those who are responsible for their iutroductiou. It is a matter of self-evidence that, if somebody claims credit for a discovery, he has to state what his discovery is. Let us assume that A publishes a note maintaining that he has found a new component of air, which lie calls so-and-so, but abstains from explaining what it is he has discovered. Another, B, working in the same line, also finds a component of air, which he describes and designates with a name. Then A (or one of his followers) gets up and claims priority for his name. — ^Another case. The morphologist C announces that lie has found in a certain group of animals a new secondary sexual organ, to which he gives a name. There the matter drops ; nobody can possibly tell what the new organ is. Some time after, several secondary sexual organs are discavered in that group, and described and named. Now the knowledge of the structures has become common property to scientists, somebody examines the preparations of (', and, finding that the naked name published by 0 applies to one of these organs, maintains that the name given by C should be employed for it instead of the later name, which was accomi)anied by a proper description. There can be no doubt what the verdict of scientists would be in either case. Science is knowledge of nature. Anything new which does not increase our knowledge of nature is outside the pale of scientific work, and what we do not know is not yet part of science. Facts professed to be new, and new interpretations of facts, do not advance our knowledge if they are kept secret. We know a priori that there are many facts to be discovered and new interpre- tations of facts to be oftered. A naked name or technical term, however, does not tell us what is the nature of the conception for which the name is meant ; and as long as we are left without this knowledge, the name or technical term has no standing in science. Name and technical term are nothing but arbitrary means which science employs as a convenient abbreviation for expositions of b ( xviii ) fiicts niul for tlie result of lengthy indiictious. The name as such is not part of science ; we miglit employ a number, or a letter, or some other sign instead without interfering in the least witli that part of knowledge which is thus designated. It is obvious that these deductions * which ajiply to science in general ajiply also to the nomenclature of classificatory work, if the work is meant to be scientific in all its branches. Facts and ideas in classification require explanation like any other facts and ideas in science. Families, subfauiilies, and all the other classificatory units down to the individual varieties require exposition by definition. The definitions bring into order the chaotic mass of individuals which forms the subject of classificatory research. However, instead of operating with the defi- nitions, the systematist employs, for the sake of brevity, names for them, thus simplifying reference. Every name is a tenu for a definition. It follows from tiiis that a name which is not a term for a definition — i.e. for which no definition has been given — has no standing. Naked names, with which classification has been favoured in abundance, are no valid terms ; they become so only from the time when the fact or idea is published for which they are meant to be employed as a convenient means of reference, and therefore cannot take jtrecedence over a name which has been defined before that time. An author who publishes a name for a genus, variety, family, etc., either has some kind of definition in his head — and then he should not keep this definition a secret, — or he has not — and then he should not propose a name for something he docfs not know, and of which therefore he cannot be certain that it exists at- all. The action of an author who publishes naked names is as indefensible as would be that of a describer who published names for the respective subspecies, for instance, of those Oriental Papilios which are as yet not known from certain islands, but which doubtless occur there, and which are certainly difterent from the subspecies of all other places. We appeal to secretaries of scientific societies and to the editors of scientific journals to suppress all new names which are not accompanied by some kind of definition. Systematic work should no longer adhere to the bad habits of the middle of the last century, when the Linnean method of classification, though so young in years, had already become weak as if from old age and had lost its vigour, and classifying was to a great extent more a pastime than a science. A catalogue of names like Dejean's, containing thousands of nomina nuda, jmblished there for the first time, is, we hope, au impossibility in our time ; but single nomina niuht still at the jireseut day appear even in works professing to be scientific. It follows further, that, if we do not wish to jeopardise altogether the efficiency of nomenclature as a convenient means of reference and communication, and thus eftace the motive which induces scientists to burden themselves with a nomenclature, it is absolutely necessary that a definition should be replaced We iindoi-.-t:iU(i under deduetion tlie process o£ reasoning by which we conclude from a general law the correctness of single cases; under inductUni the process of reasoning by which from single cases a general law is formulated. ( six ) only by one and the same name, and tliat a certain name shonld apjdy only to one and the same animal everywhere. Whoever adheres to this princif)le of stability of nomenclature must concede that this end can only be attained by adhering to the first defined name for every animal or plant. No compromise is possible. Personal preferences for euphony, so-called purity of langnage, etc., must be sacrificed by all those who sincerely advocate stal)ility of names ; there is no help for it. A publication is meant to distribute knowledge acquired by the author. The publication of a new scientific fact or idea is meant to enable the reader to understand what is discovered. As even nowadays names without any attemj)t at exjiosition are considered valid by a good many systematists, it cannot be wondered at that the definitions published are not always so precise as to advance the knowledge of the reader beyond the fact that something hitherto unknown to the anthor is defined. We have the description, and do not know what to make of it ; we have a name, and know perhaps what it signifies jihilologically, but not what its meaning is in natural science. There is no excuse for nnnecessarily vague definitions of varieties, species, genera, etc. ; but there is much in the method of so-called systematic work which explains the frequency of inadequate descriptions. Incompleteness is an inherent character of classificatory research ; the best definition is not complete, and may, therefore, any day become insufficient for recognising the species and variety defined, (U- must be modified in the case of genera and the higher classificatory cate- gories. A character apparently not worth mentioning to-day may become very important when more allied forms are known. But then, what is the use of having descriptions at all ? They are nothing but a record sufficient for the time (or meaut to be so). They do not profess to be final, though the author may aim at finality. Here, as everywhere, the advance towards completeness is gradual. As our knowledge increases, the definitions of species, genera, etc., become widened or restricted. The definitions change in scope, while the name which was valid for the original definition remains the same. This contrast between a stable nomenclature and a labile knowledge is a nomeuclatorial evil and a source of much disagreement among systematists. If we tried to make tlie names as labile as is our knowledge of nature, the remedy would be worse than the evil. Albeit we cannot do away with the evil altogether, we are at least able to mitigate its severity by the application of a dose of common-sense. We all know that the number of specimens on which the definitions of new species and varieties are based — it is of no consequence for our argument which classificatory category the reader designates with the word "species" — is extremely small compared with the number of individuals existing. Innumer- able species have been described from single specimens. Though this specimen or these few individuals may have been abnormal, though the definition is after- wards found not to cover the entire species at all, as an original definition seldom does — unless it is so vague as to cover other species as well — tlie name, given to the first-named individual or individuals is accepted for the vast (XX) nnniher of specimens which sire later found to belong to the same species, be they i)ractically identical or be they very different in appearance. Preference is given to the first name, though the species may later have been much better described under another name. Nobody with a sense of responsibility will now- adays re-name a species, variety, or genus of which he knows that it has a name, on the ground that the original definition does not apply to all the specimens of the species, or all the species of the genus, for which the original name is now employed. Everybody who agrees that for the sake of a stabile nomenclature the first name should strictly be preserved, gives to the first individual or individuals which became known to science an importance in respect to nomenclature which none of the later-discovered specimens can acquire. Now, if a definition is not sufficiently precise to recognise by it the species or variety, there is one way of solving the riddle, accepted by all systematists, we think. If there should be somebody who objects to this means of finding out the meaning of published but insufficiently defined names, and advocates that such names should be dropped, he will doubtless retract the objection, if he comes to think of the consequences. To drop such names, though theoretically justi- fied, is impossible, as such a procedure would give the careless worker and the ambitious amateur of the worst sort an excuse for inventing new names wholesale. The means referred to of ascertaining the meaning of an original definition is the comparison of the original specimens. If they are not pre- served (or if the author has based the name on an inaccurate figure or on an insufficiently precise description of an earlier writer), the name cannot take priority over another name ; it may be put down as a query synonym under some species with which the definition agrees best, or may be enumerated as species incleterminata at the end of the catalogue of the group. There are very few defined names of Sphingidae which we cannot refer with certainty to any species known to us : Sphinx ixion and Sphinx delis of Linne ; Sphinx leuco- phaeata and Chaerocampa thalassina of Clemens ; Smerinthus decolor, Sphinx trojanus, Chaerocampa brasiliensis, Macroglossa tristis, and Oenosanda chinensis of Schaufuss, are examples. If the originals are there and are sufficiently well preserved, we may be spared all difficulties, or we may get more deeply entangled in the meshes of nomenclatorial controversy according as we find one or more originals. Let ns consider the two cases separately. (1) If the species (or variety) was based on one individual, or, at all events, if only one individual (authentic, of course) is preserved, and there is nothing in the description which distinctly points to the definition being based on several different specimens, we are quite certain of what the name applies to. And that is all we require. (2) If the species was based on several specimens, we may find that they belong to one species (or variety), or to more than one : — ifi) If they are actually of one sjiecies (or variety), there is again no uncertainty about the application of the name. But we must remember that to pronounce two or more individuals to be specifically the same is nothing else ( xxi ) but a conclusion, and that a conclusion may be erroneous. Those who have some experience in systematic work will know that every now and again it happens that the specimens which one author considered to be the same species (or variety) are proved by another to represent several. The reader will find a number of instances illustrating this experience, if he looks over the synonymy in the present work. Therefore, what appeared to be certain may become uncertain again, if there are more than one original specimen. Some authors will, indeed, accept the identification even if it is based on some such mistake, because they consider a name far too nnimportant to justify a strict adherence to principles, if a change of names is involved. However, the majority of classifiers will oppose a name which is incorrectly applied. This spirit of opposition against all mistakes is very healthy. We should deplore its absence ; for we are sure, because we know instances, that he who intentionally overlooks errors in apparently irrelevant matters, will treat in the same spirit also details of fact which appear to him trivial, which may, however, be of the greatest bearing upon general questions, and, therefore, mislead altogether the generaliser who has to depend on the accuracy of the Specialist. (b) If it is proved that the original specimens belong to more tlian one species (as do, for instance, the originals of Walker's Macroglossum sitiene, corythus, Nepkele viridescens, etc.), systematists have adopted several methods of narrowing down the conglomerate to one species. These methods are as follows : — (<!>') First method of restriction : The name of a composite species is to be restricted to that component which is the first to which the name is afterwards applied bv the same or some other author. Illustration : Macroglossum corythus fA, ot 1856 consists of three species -( B, Of these B is the first mentioned as lo. corythus after 1856; ergo, the name of corythus is restricted to B. — To be certain of the result, it is necessary to know which is the first, and that requires a knowledge of all the books where the name occurs, and, moreover, a knowledge of what is meant therein by the name. These premisses may sometimes easily be got over, but they present more often difficulties which are as intricate as those which the method professes to solve. (4-) Second method of restriction : The name is to be restricted to that component of the composite species which remains after the other components have been subsequently separated under new names. Illustration : Macroglossum (A, corythus -^ B, As A and B are described as new in 1875, the name corythus ic. remains for 0. — To arrive at this result, one has to inquire into the descriptions of the new species, in order to find out whether the new names really apply to A and B ; the new species will in many cases again be found to be composite. The method, therefore, creates new difficulties in trying to remove the old ones. ( xxii ) (c") Tliird metliod oi' restriction : As the first aud second methods are opposed to one anotlicr, ditlering nearly always in the results attained, we reject them both. The energy spent on the book-research which either method requires is misai)plied, reminding one too much of the famous fight against windmills. Nomenclature is not part of nature ; it is an auxiliary means invented by the classifier for his own convenience. What in the name of common-sense compels us, then, to turn a convenience into an inconvenience? There is a wide scope for research in nature recpiiring all the energies of scientists. Why, then, impose upon scientists those unnecessary labours which have only a nomenclatorial, but no scientific result ? The method adopted by ourselves is at once logical and very simple, and removes all the difficulties as far as that is possible. Our method of dealing with composite species (and genera) is to narrow all cases down to the case dealt with under (1) by simply applying also liere the law of priority recognised by nearly every classifier as the only means of arriving at a stable nomenclature. P^rom tiie sequence of the localities under a composite species, or from the characters mentioned in the definition, or from the bibliography referred to by the author of a new species, one is able to draw up a sequence of the components of the sjiecies. If Macroglossum corythus, as conceived by Walker in 185(i, consists of three species. A, B, C, we have : — TA = M. cor/jf litis ; Macroglossum cor)jthus \^ = M. cori/thus ; \(_] = M. corythus. Each of the three components is M. corythus, according to Walker. According to the law of priority, the same sjiecific uame cannot stand twice in the same genus, and tlie name occurring more than once can be valid only for the si)ecie8 which was first published under that name, or which stands first in the book where the uame is defined for the first time. This rule being applied to the above case, it follows that the name corythus can stand for A only. All we have to do, therefore, is to find out the sequence of the components of a composite species. This is mostly easy, especially in the case of geographically separate forms. lu the case of composite genera the sequence is given by the names of species mentioned, there being very few genera defined without reference to one or more species. Strict adherence to the above rule makes the first species mentioned the type of the genus. One might object that this mechanical application of a rule leaves it entirely to accident which species becomes the type of the genus, or to which j)artieular portion of a composite species the specific name is restricted ; and, lurtlier, that tlie autiior did not intend to give the first species or the first specimens respectively any such pre-eminence, and that the "type" thus fixed may be just the one to which the description ajiiilies least. We reply, firstly, that we do not know the intentions of the author, as he did not state them ; and, secondly, that, if the description applies accidentally less well to the species ( xxiii ) 01- specimens first in the sequence than to the others, this argnment holds equally good in the case of the types fixed by any other method of restriction. We invite the reader to find out the types of such genera as Papilio, Sesia, and Zygaena by all three methods. A glance at the original definitions of these genera sufiices to fix priamus, tantalus, and Jilipendulae as the respective types according to tiie third method. A study of several families is necessary before the types can conscientiously* be ascertained by the first and second methods, since the species originally included in each genus belong to different families. In mammals and birds and several other groups of animals the second method has almost generally been adopted, for genera at least. If the systematists have there really arrived at a stable nomenclature, no change is necessary, a stable nomenclature being the main aim of the principles of nomenclature. In Lepidoptera, however, and other insects, the first and second methods, less often the third, have been followed, and that has landed us iu such a muddle that there is no question of stability having been attained. This being so, we should have adopted, as a matter of course, the surest method of restriction for the sake of avoiding waste of energy, even if our method was not the logically correct one for all who agree that strict priority has to be adhered to. If the authors of the names for varieties, species, genera, etc., had done from the beginning what we now have to do with their names ; if the authors had restricted every name iu the way that we now are compelled to restrict it, much time would have been saved. We all agree that a specific (or varietal) name based on one specimen, and a generic name founded on one species, are as valid as names based on more material. Further, if all names were based on one individual or on one species respectively, there would be no composite species and genera ; and if the original individuals of each species and variety were preserved, scarcely any difference of opinion would arise among careful workers abont the application of the names. We cannot alter what has been published ; but our contemporaries and the scientists who come after us have it in tlieir own hands to simplify nomenclature in the way here indicated b}' making all names monotypical. We do not know who was the first to fix a ti/jie (= ti/pus) for the name of a sjiocies, variety, or genus. The nomenclatorial term appears already in 1816 (Dalman). The word ti/pe was jierhaps not the best that could have been chosen, as it had already a definite meaning also in science, signifying that wJiich is typical for a group of units. But as we frequently use in science tlie same word for different conceptions (claw, wing, tarsus, lip, mandibles, tail, tongue, etc.), tlie philological objection against the term " type " is not of much weight. However, tlie difference between the meaning of the word as used in ordinary language and tlie meaning of the nomeuclatorial term has occasioned confusion, and hence led to another kind of objection. * Some writers have simplifieil matters fur themselves by iguoiing the exotic species altogetlier ! I ( xxiv ) Some antliors, acceptinjj (lie wurd "tyiic" in tlie ordinary sense implying that till' sjieeiniens called tyjx's arc tyjiii-il iii(li\ idiials, very pnijierly rejily that these t\]<L-s are often alierrunt sjircinicns, and very seldom the most typical for the group of individuals to which they belong. This confusion of the verbal and the technical meaning of the word "type" misleads those authors to insist further that, there being no " types " in nature, one individual being no more a pre- eminent rej)resentative of the species (or variety) than another, the word " tyjie " as a nomenclatorial term has no standing. It is ohvions that those authors fall into a deplorable error of confounding the names, which are the jjroduet of scientists, with the objects named, which are the product of nature. Certainly there are no types in the nomenclatorial sense in nature, but there are also no names. Tiie type is as such not at all the type of the species, Init is the type of the arbitrary name given to the first specimen or specimens, and applied by common consent to all the specimens which belong to the species, of which tlie type-specimen is only a member, like any other individual. Those who have the stability of nomenclature at heart, and are unwilling, when pro})osing a new name, to lay an avoidable burden on scientists — and who do not consider themselves infallible — should mark one individual as type (= ti/pus) of the name, and make a clear statement to that effect when publishing the name — • and one individual only. Every care should be taken to have such individuals preserved. There is neither justification for opposing this usage by which the systematists benefit enormously, nor for employing the purely nomenclatorial term " type " in any other nomenclatorial sense than the one here advocated. As a name is not valid if the animal or plant has already an earlier valid name, we reject also all tliose names of composite species and varieties of which one of the components has an earlier valid name, aud the names of composite genera aud higher categories which comprise the type of an earlier validly named genus or higher category respectively. Temnora brisacus of Walker (18.56) has no standing, because it is a mixture of several species, of which one is Cramer's pylas (1779). Dalman's Ilemaris (1816) is a synonym of Macro- glossum (1777), because it includes the type of Macroglossiim. Aud for the same reason the subfamily name 2Jacro(/Ioss/nac (1875) becomes a synonym of Sesiinae (1819). Ambul(/x of Walker (1856) cannot stand, as it includes Amplypterus of Hiibner (1822). In general terms : — If A is based on a, or on a and b, which are not cosjiecific, respectively not congeneric, etc., ovm a,b,c, „ „ „ „ „ and B is based on a, or on i and a, or on a mu\b, or on ^ and a, or on a and ^, or on b, a, c, or a, b, c, or a, c, b, etc., „ „ „ „ then JJ is a synonym of A. ( XXV ) Tliis does away with some names {jroposed in scamped work, of wliicli science would he well riti, es[)eciully with those names which are foniided on actually the same material. Before a new name is introduced, the author should ascertain to the best of his ability that the material for which the name is meant to stand has no earlier name. This is a demaud on a describer, on the fulfilment of which classitiers should rigorously insist. In an ideal nomenclature a name should convey to the systematist the characters of the animal or its place in the system. As long as the classification is not final this is not possible, and final it will most likely never be. One step towards this goal was made by Linne himself when he established it as a rule that an anim.al or a plant was to be designated by a generic and a specific name. In Lepidoptera he tried to go even further by indicating, in the case of some groups, by means of a certain ending to the specific name {-clla, -aiia, etc.), to which family tlie species belonged. But this proved to be a complete failure. The Linnean binominal system of nomenclature had the one great advantage that, when the number of forms became larger and larger, there was no serious objection to employing the same specific name in several genera. In Linnd's time, when so comparatively few animals and plants were known, the generic name was indeed sufficient to tell the scientist the position of the form in the system. Tliis is no longer the case. By far the larger number of generic names do not convey any idea to scientists, except to a few specialists who happen to know them ; the family or even the order to which the genus belongs has generally to be added to make clear what is meant. So far students of natural science have adhered to tlie Linnean binominal system with that tenacity with wliich human beings generally cling to what they have become accustomed by long usage. However, a great change has taken jdace in one direction. When the theory of descent threw an entirely new liglit on the forms of animated nature, the study of variation became an all-important subject. That the species were more or less variable was long known. But now the variability assumed C|uite a different aspect. It was found that there were different kinds of variation. AVhereas formerly the chief object of dassificatory research was to separate the individuals into species, and group these in genera, and so on, now there were in addition the several kinds of varieties to be carefully studied. For this purpose a nomenclature of varieties is as necessary and as convenient a help as the Linnean binominal nomenclature is to the student of species and higher categories. Systematists agree that the name of a species, genus, or family must be of the same form, so that one recognises by the name (or rather the formula) which dassificatory category of units is meant. A family name must be different in form from a subfamily name, and a genus name from a non-generic name. The name itself must show us whether it designates a species, a genus, a variety, a family, etc. The efficiency of nomenclature would be nil if one could not see from the name that Charaxes castor designates a species, Papilio a genus, Ayaristidae a family, according to common agreement, Sphinx ocellata x Amorpha populi a hybrid, Arascknia levami f. t. prorsa a seasonal form, etc. This is so ( xxvi ) self-evident that we ask onrsclvos in wonder liow it is possible that there are systematists wlio do not— should we say will not? — recognise the necessity of it. If Frlmjilla coelehs is accepted as a formula for a species, Sphinx atlunticus is also a designation of a spei-ies, and not of a genus or a variety or a subspecies. To speak of " species " Sphinx ocellata and of " subspecies " Sphinx atlanticiis is a contradiction unworthy of science. In i'ornicr times varieties were looked upon as freaks of nature. They were to many a classifier an interesting nuisance, which often threatened to upset the balance of his well-fixed species, and were on that account more often entirely put aside than welcomed as an object for research. Esper, who went perJiajis deeper into the phenomena of variation than most of his contemporary entomologists, already distinguished between ordinary varieties (Abweichungen) and abnortnal individuals (Ausartungen). However, as long as the principle of evcilution underlying these varieties was not recognised, there was no need to study them systematically, and to work out a system of nomenclature which would bring into order the chaos of varieties, as did Linne's binominal system tlie chaotic mass of sj)ecies. From Linn(5 onwards varieties, if provided with a distinctive name, are recorded in various ways. The following names may serve as illustrations : Papilio iris luteus ; Columba oenas j3. domestica ; Phasianus gallus /3. gallus cristatus ; Phasianus colchicus (/3.) Phasianus varius. The word varietas, introduced by Linn6 as subordinate to species, meant anything deviating obviously from the normal individuals of a species. The practice of putting the term in an abbreviated form, as rarid. or m;-., before the varietal name does not seem to have sprung up before the beginning of the nineteenth century, and the use of the term aberratio {ab. = aberr.) is still younger. Not rarely the "variety" was in reality the normal form, while the "species" happened to be described from aberrant specimens. There was no strict rule for the emjtloyment of var. or ab. ; some treated well-marked varieties as var. and less obviously different individuals as ab., the distinction between var. and ab. being merely quantitative ; while others employed ab. for abnormal specimens occurring singly among the normal ones, and var. for the regularly observed varieties. There are many collectors and classifiers, representing the stagnant element in this department of our science, who look at varieties still from either of these standpoints. Since the middle of the last century, when natural science stepped from childhood into manhood, the study of variation has gradually become more methodical, with a change in the PRINCIPLES OP CLASSIFICATION, and has now attained a height of which our forefathers in science did not dream. New lines of research bring to light new series of facts ; and new kinds of facts recpiire a new terminology. It will not do to have the same nomenclatorial formula for a sj)ecies as for a genus ; and so it will also not ( xxvii ) do to name all kinds of varieties in the same way. It was Staudiiiger wlio first separated the varieties of Lepidoptera iuto two categories: geographical and non-geographical varieties. There is indeed a great difference between these two categories, as we shall see later on. Unfortunately Staudinger adopted for the geographical variety the old term varietus (car.), and for the second kind the term aberratio, giving both terms a definite meaning which they did not originally have. He should have invented a new term at least for the geographical variety. As it was, the application of rar. and ah. remained in as great a muddle as before. While there was thus some sign that the relatiou of the varieties towards each other was assuming a clearer aspect, at least to some authors, a curious misunderstanding crept in, which prevented many classifiers from perceiving the true relation of the varieties towards the species. It is self-evident that two or more diflferent-looking animals whicli are found not to be specifically distinct from one another belong to the same species, i.e. are components of this species. All the components together are the species. This is a truism. How- ever, systematists became accustomed to look upon that particular component which was first described and named, and of which the name was accepted as the name for the entire species, as being the species, while it was, as a matter of course, only one of the components of the species. It is utterly wrong to say that the first-described form is the species and the later-described forms varieties of it. We know, for instance, that Araschnia lemma and provsa are the same species ; neither levana nor prorsa is the species, but the pale spring-form letana and the darker summer-form prorsa together are the sjjecies, thus : — levana \ \-= species. prorsa J It is j)urely conventional, on grounds of nomenclatorial efficiency, and has nothing to do with the relation of the summer- and spring-broods towards each other, or with the causes and the origin of such horodimorphism, that the first-given name, levana, is applied to the s]ieeies. In doing this people forgot that the name levana, originally standing for part of the species, was now employed for the whole as well, and that therefore the true relation between the species and its components is this : — spring-brood levana \ , , = species levana. summer-brood prorsa J Or, if we use the term/, t. {= forma tempestatis) for horodimorphic forms, we have as formula for the insect : — f f. t. levamt ; Araschnia levana -\ „ , \ i. t. prorsa. Instead of employing this formula, classifiers spoke and speak of a species levana, meaning the spring-form, and a variety prorsa, meaning the summer- form, co-ordinating the one with the other : — Araschnia levana ; Araschnia levana var. prorsa. ( xxviii ) This is wrong, as tlic spring-form is no more the species than is tlie summer-form. What we have said licre in regard to seasonal varieties, applies also to geograjiiiical and individnal varieties. Wliich of the components of a species is the first-described and -named form depends in nearly every case entirely upon accident. The Hrs(-named form may be the most aberrant and the very youngest devehipnient of tlie species, having originated from one of the later- described compounds of the species. To call this accidentally first-named portion of a sjjecies the species and the later-named forms varieties of tlie first, is a ludicrous confusion of facts. And yet, systematic work, from mammals downwards, teems with this glaring misconception. As nomenclature is a convenient auxiliary to classification, as it is sub- servient to science, and must therefore be accommodated to the latter, it should not form a liard-and-fast structure, into the compartments of which the results of ciassificatory research have to be squeezed somehow. The distinction between the scientific part and the accessory nomeuclatorial side of classification should never be lost sight of. Tiie aim of scientific research is to discover and elucidate the phenomena of nature. Classification, as part of science, aims at an understanding of the connection between the individuals. To attain this object it relies on facts discovered by two lines of research : firstly, on the facts relating to the body ; and, secondly, on the facts relating to life. And here, as in all scientific research, we find tlie primary question underlying all investigations to be difference or no difference, because science is always comparative, consciously or unconsciously. Morjihology and anatomy provide the classifier with the knowledge of the body. In a vast number of instances there is no other knowledge available than this, to build a classification upon. The corporeal facts of tlie morphologist and anatomist are, however, no absolutely trustworthy basis for a superstructure. For the primary units of the classifier, the individuals, are always different from one another to a certain extent, and therefore cannot be proved to be classi- ficatorially identical by corporeal comparison alone. As in inanimate nature identity can be established by action and reaction, so also in animated nature. The observed differences and apparent identities in tlie bodies of the individuals have to pass the higher criticism of the knowledge of the phenomena of life. Two individuals may appear very different to the morphologist ; but the classifier, who knows from observation of the living animals that one is the offspring of the other, cannot establish any other connection between them than that of parent and offspring, however conspicuous the bodily differences may be. The difi'erences between young and adult, male and female, parent and oflspring, brothers and sisters, however prominent they are, lose all the ciassificatory importance which the morphologist and anatomist (and the classifier misled by him) thought they had, when biology establishes the true relationship of such individuals. On the other hand, apparently insignificant corporeal differences, which the morphologist may scarcely deem worth noticing, often turn out to be differences between ( xxix ) animals wliich are entirely independent i)f one another. The conclusions based upon the facts of anatomical and morphological research must always give way to the direct proofs of biology. The significance of corporeal characters is established by biology. Anatomy and morjjhology give the quantity, biology determines the quality. It is therefore obvious that the classifier should not let himself be guided in his judgment solely by a consideration of the quantity of bodily characters, bnt always keep in mind the higher criticism of quality. If he does this, he will not easily fall into the error of treating two groups of individuals as being of the same classificatory category (variety, species, genus, etc.), if biological considerations are against it. If the classifier had no other guide than corporeal similarities and differences, the classification would merely be an artificial arrangement, without regard to the true connection between the animals classified. Such a classification, perhaps very useful for the mere collector, might be likened to an arrangement of minerals according to their external features, without any regard to their chemical composition. Although there are no biological data available of most animals, of which we know nothing but what the dead bodies or portions of them (embryo to adult) tell us, the scientific classifier has nevertheless another guide. Tiiis guide is the principle of judging from analogy (the word used in the logical sense). We give two illustrations. If in a certaiu country the spring- and summer-broods of a certaiu grouj) of species have been proved to be different in all the species examined, we may safely conclude that they differ also iu those species of the group which have not yet been examined. Therefore, forms of that group of which it is only known that they differ iu the same way as the horodimorphic forms of the better-known species must also be treated as seasonal varieties, and not as distinct species. As we have found that the genital armature, though individually variable, does not exhibit any seasonal differences in those species of Lej)idoptera of which forms proved to be seasonal have been examined by us, we are bound to conclude that Lepidoptera which do constantly show differences iu those structures are not seasonal forms of one species. The j)remiss is the better established, and hence the conclusiou is the more likely to be correct, the more species have been examined. This deductive reasoning, though logically exact, will never give us certainty. For the animals as we know them are the product of evolution ; the result of evolution is dependent on the nature of the animal and of the evolving extraneous factors ; these extraneous factors being independent of the animal, their combination with the internal factors — which combination dete-mines the line of develop- meut— depends on accident ; results determined by accident canudt be deduced from an a priori law. That means, we do not a priori know that what holds good in all the cases examined is true also in every case not yet examined. Nearly all the so-called " laws " in biology are nothing but more or less well- formulated rules of probability admitting of exceptions. A rule may apply to every individual case within a certain group of animals ; but the further away we go from that group, the greater becomes the chance of exceptious turning ( XXX ) up. It is hasty to generalise from a limited group of animals, though biologists arc nowadays in tiie habit of doing so. Every group first requires indejicndent investigation before a rule found to be true in another group can be pronounced to be true also in the one under consideration. The " laws " of development hiive all a jironiiss ; tlicy are true only under certain conditions. The scientist must, therefore, always remain aware that the unexpected may happen, and individual cases turn out to be exceptions. For instance, what the classifier considered to be a generalised character may turn out to be a specialisation, and what is a sjjecific difference in one group may only be seasonal in another. Since the classifier is in the same position as the commentator of a book which is written in a language only im])erfectly known to him, and from which many words and whole pages are missing, there is necessarily a certain amount of assumption in classificatory work which makes the results proportionately unstable. However, the degree of instability can be lessened to a great extent. Just as we know a priori that a certain character which appears generation after generation in the larger number of individuals of a species will finally be present in all (if the conditions of life remain the same), so the probability of the correctness of a conclusion in classification will become certainty, if all the details of the animals and all the facts of their biology render it probable that the conclusion is correct. In order to ajiproach this certainty the classifier has to give his conclusions the widest possible basis. The results of anatomy and morphology must be cliecked off by biology, and the conclusions derived from one organ by the coni'lusions based on every other part of the body. If one organ contradicts the other, there is a mistake either in observation or in interpretation. The more snch luistakes are discovered and corrected, the safer the superstructure of conclusions will be. Grouping the animals according to the apparent develop- ment of one organ leads to an artificial arrangement ; grouping them in accordance with all organs checked off by biology results in a final classification. However, finality, even if the classification is restricted to a small group of beings, entails such an enormous expenditure of energy that it can be ajjproached only gradually in the course of time by continued co-operation between the various lines of research. All classification begins with the discrimination between individuals. Every- body distinguishes between the specimens he observes around him and brings tiiem into groups, thus setting up a rough kind of classification. Experience shows which individuals stand in the possibly closest blood-relationship — that of ])arent and offsi)riDg, and of brothers and sisters. The classifier starting with this experience, which can be corroborated over and over again, has therefore as foundation for classification a solid fact which will remain firm when a flimsy superstructure that a wild fancy may think fit to erect is blown away. Obser- vation of the individuals so closely connected as parents and offspring teaches ns that there is always a certain amount of corporeal difference between them, every individual having an individuality. We notice this variability in all organs ( xxxi ) when studying them closely. There is no constancy. How far the varial)ility extends is a primary subject of investigation. Lepidoptera are the most convenient group for the study of variation, as they can be comparatively easily reared and experimented upon. Variability may be fairly equal in all organs, or may be excessive iu one aud slight in others ; variability of one certain organ may always be accompanied (within one group of animals) by variability in certain dtlier organs, or the variability may be quite independent. Variability (= state of being different) is to be accounted for by two kinds of variation (= process of becoming different) : firstly, individual variation, pertaining to brothers and sisters ; and, secondly, generator)/ variation, pertaining to parent and offspring. Individual variation is normalli/ such that there is a gradation from one extreme to the other, the number of individuals becoming, how^ever, smaller towards the ends : monomorphisrn, respectively — chromatism. Or the individuals which are all the offspring of the same parents fall into two or more groups, which may or may not be connected by iutergradations : di- or pohjmorpliism, respectively — chromatism. In every case there appear occasionally specimens which stand outside the ordinary range of variation. Such aberrations are generally aberrant in one organ only, and otherwise quite normal. Aberrations should not be confouuded with monstrosities, in which the deviation from the normal is due to injury of the early stages. The dimorphism of commonest occurrence is sexual, the jiurely sexual characters being nearly everywhere in sexually separate animals accompanied by differences in size, colour, aud some kind of structure. Marked dimorphism in both sexes is comparatively rare. But where such di- or polymorphism exists, and where the similar individuals of both sexes are referred to under one name, it must not be implied that the similar specimens are biological entities. They constitute like aberrant indi- viduals a morphological or anatomical unit named in contradistinction to the other individuals. A black c? of Papilio machaon is not more nearly related to a black ¥ than to its own normally coloured brothers and sisters, and the brown cJcJ of Mimas tiliae are as much the Si of the green ? ? as of the brown ones, in spite of corporeal similarities and dissimilarities, and in spite of similarly coloured specimens bearing the same name in classification. In most groups of Lepidoptera the variability of the males is inferior to that of the otiier sex, there being numerous species in which the male is monomorpliic, while the female is polymorphic. In such cases the females are generally protectively coloured or otherwise adapted in several directions, and may occasionally become mono- morphic again or more constant than the males, in consequence of the best- atlapted form becoming prevalent and finally appearing alone. The habit of naming varieties has been in vogue among entomologists and conchologists more than among any other systematists on account of the great variability of insects and shells ; and as the student of variation requires names for the sake of brevity of reference, there is nothing to be said against the habit. But since every individual deviates in some organ from the mean, there is some danger of the naming turning out indiscriminate. Fortunately, Lepidopterists have so ( xxxii ) far (■(iiiliiu'd tli(.'nisclvfs to anli'iit liaplism dI' iiidiviilual vai'iuties showing some distinct ion only in colonr, jiiittern, or sliapc. There is, of course, no reason for such restriction. There is just as much justification in Le])ido{)tera as there is in beetles for naming structural varieties. If it is not objectionable to baptise sculjiture-varieties of Carabus, there can also be no objection against providing with a name, for instance, the numerous individuals of Chalcosiidae which are different in neuration. However, it is our oj)inion that the naming of individual varieties should not go further than is necessary, and that only the student of variation can decide how far it is necessary. Conspicuous individual variability is observed among Sphingidae both in the larvae and the adults, the caterpillars of Macroylossum and Cephonodcs and tlie imagines of some Ambidicinae being very variable in colour. The differences between jiareut and offspring are no less marked in many instances than those of the offspring inter sc. The generations as tiiey follow one another are either practically the same, or there is a cycle of more or less different generations. In Lepidoptera the cycle is generally restricted to two or three broods. As these coincide in time of appearance with the seasons as a rule, this particular kind of " gemratonj " variation is called seasonal. Though as a matter of course the individuals of each separate brood are a biological entity varying individually in the way explained above, there is nowadays a great inclination amongst systematists as well as biologists to confound seasonal with individual variation. If two different specimens are offspring of the same female, they are snrely not seasonal varieties. At the bottom of the confusion lies the assumption that, since seasonal variation depends on meteorological differences of the seasons, all differences are caused to appear by such factors. This assumption is certainly incorrect, and therefore the indiscriminate treatment of different- looking specimens as seasonal quite misleading. Everybody may draw conclusions as he pleases, but the facts must be represented as they are ; they must not be tampered with. Seasonal variation is observed in several instances among Sphingidae. The most remarkable cases are found in Iliemorrhagia and Hyloiciis. Though we know that generatory variation is often structural in insects (^Aphidae, Physopoda, etc.), we were nevertheless surprised to find a structural difference in the claw-segments of tiie two forms of Hi/loicus perelegans, which forms we presume will turn out to be seasonal. Having studied the differences between the individuals proved by rearing to be parents and offspring, the student will be able to select from the individuals at large those which a comijarison of their bodies convinces him to be so nearly identical with the specimens reared that they might very well be brothers and sisters of tiiem. But knowing that corijoreal similarity is no absolute proof of biological identity, he will prudently test his conclusion — perhaps gather some female moths which come to his lamp and which seem to him identical to all intents and i)urposes, get eggs from them, and rear the larvae. To his astonishment he finds that the apparently identical females have ( xxxiii ) produced two very different kinds of caterjjillars, from wbicb lie obtains in tbe course of time a series of imagines again very niucli alike, bnt presenting now to liis searcbing eye and suspicious mind some sligbt differences. Continued experiment witli tlie two insects proves to liim tliat lie bas to do, not witb a dimorpbic larva, but witb two entirely independent beings, wbicb fly togetber and feed as larvae on tbe same plant, and wbicb are scarcely distinguisbable as adults, and are nevertbeless perfectly kept apart in nature. He realises tbat tbere is a gap between tbese two kinds of insects wbicb is utterly different from tbe gap between tbe varieties wliicb stand in the relation of parents and offspring ; tliat tbere is an effective barrier wbicb lies in tbe nature of tbe specimens tliemsclvcs, separating tbe two sets of individuals completely, tbough tbere is no extraneous barrier between tbem. And by studying furtber tbe life around bimself, tbe student will perceive tbat tbe animated world is composed of a multitude of sucb sets of specimens, of such entities, all separated by tbat same kiud of barrier. The knowledge of the existence of this barrier is essential for the classifier. What the liarrier is the student cannot know witb certainty. The corporeal differences observed in the individuals are not the barrier, but are only accessory to it. Passing now into a neighbouring country, tbe scientist will find practically the same composition of tbe fauna, tbough some old friends may be missing and some strangers may meet bis eye. A good many of the entities will indeed be identical with what he knew before, but others appear in an altered garb. In one the range of variation has remained the same ; but the indi- viduals wbicb were in the minority in the first place are here in a majority, tbe mean of the variability having changed. Another entity, wbicb was known to liim as being monomorphic, is dimorpbic in tlie new locality. A third, which was seasonally variable there, does not exhibit seasonal variability here. Among the specimens of a fourth entity there appear individuals different from what tbe student had hitherto seen, tbe range of variation having become shifted or widened. In others, again, he finds the proportion of sucb different individuals to be larger and larger, until the student comes to entities of which all specimens exhibit some distinction from the individuals of tiie former country. They are the same entities, but with a difference. Now, in which relation to each other stand sucb geographically separated entities ? Tbe two extreme cases are tbese : — (1) The geographically separated entities are, each taken as a whole, identical, witb tbe same range of variation and the same mean of variability. (2) The geographically separated entities are different from one another in all individuals. Between these two extreme cases there are all intergradations. The difference may be found only in a very small proportion of the specimens, or only in one sex, or in the greater number of individuals, or in nearly all, or in all. Tbe difference may be structural, or chromatic, or both ; it may be ( xxxiv ) consi)icHOiis or scarcely perceptible ; the gap between two geograjihically separate entities may be small, or large, and it may be completely bridged over by individuals from other countries. A careful observer of any group of sexually separate animals will know ■ many instances referable to the various cases alluded to. We will call such geographically sejiarate animals which are different, qeoiivnithical representatires. In order to see clearly what the difference means, it is necessary to know the meaning of identity in geographically sejiarate individuals. If two geographically separated specimens prove on comjjarison to be identical, or are considered identical, the student has to conclude that they are not separated by tiiat barrier which we have above characterised as effectively keeping the entities of the same country (= synoecic entities) apart. As each of the two individuals is a portion of an entity in its native country, these two respective entities are also not separated by that barrier. That means, the two entities together are only one single entity separated from all the others by the above barrier. We cannot help adding, that it appears very strange to us to see some indi- viduals pronounced identical, and to iind them, nevertheless, treated by the same author as belonging to two " species." * It is clear, from what we have just said, that geographically separate animals which are identical in some of the individuals, or, in other words, of which the range of variation overlaps (A' varying from 1 to 5, A- from 4 to 12, A' from 7 to 19, A^ from 18 to 21, etc.), are one biological entity; that is to say, the geographically separate different specimens are the product of geographical variation of the same animal. It is a difficult task to find out the limits of variation. The material which comes into the hands of the student being insignificant compared with the number of individuals at large, it depends upon chance that the extreme individuals become known. Ex[)eriments, es])ecially with Lepidoptera, have proved that the range of variation can be increased artificially. That is to say, the capability to vary is greater than we generally observe it to be in the material collected, and therefore we must expect that the range of variation is in many cases actually more extended than we see it, and that, consequently, very often also those geograjihically separated different animals overlap in characters which differ constantly in the specimens contained in collections. If the number of individuals is small (we know frequently only one or a few specimens, often only one sex), it is the merest assumption to say that all the specimens of the respective locality are difterent from all the individuals of a certain other locality. And an author who makes such a statement is no less h'asty or superficial in his judgment than the one who waves the distinctions aside as being of no importance. However, if there is any reason for expecting the characters to overlap, it follows from what we said before, that the individuals from the two places together belong to one entity. The accom- panying diagram (Fig. 1) demonstrates ])erhaps more clearly the correctness of * See \o)\ Zool IX. p. 459 4(J3 (1902). ( XXXV ) our arj;-umeiit than any words. The diagram ilhistratcs an extreme hypothetical case. The curves A' and A- represent the variation * of two geographical representatives. The extent of variation, a e, is the same in both repre- sentatives, but the means are different. The small number of specimens, a d, belonging to A- are identical with the majority of the individuals of A', while the small proportion b c oi the specimens of A^ are tlie same as the majority of A-. It is obvious that there is little chance of the student getting any of the specimens a b of A- and b e of A', if he has not a really large material at disposal. To him A' and A- would appear to be separated by a gap ; they would appear to be constantly different, thongii they are, each taken as a whole, morphologically identical. Two sexually or otherwise di- and polymorphic animals do not differ in all specimens in the same way ; one sex or one set of individuals may be distinguishable by colour, the other by shajie ; one form may be paler, another darker, than the respective forms of the animal with which they are compared. To find out the differences between two geographical representatives which are di- or polymorjihic, it is necessary to comj)are the corresponding forms with one another, just as one has to compare sex with sex, larva with larva of the same stage, pupa with pupa. There are also monomorphic geographical representatives which differ from one another in a similar way, some individuals differing in shajie, others in colour, others again in pattern or in structure. The difference between the two representatives is also often constituted by the ensemble of the characters in eacli individual. Such cases lead over to those in which the geographical representatives are completely separated by a gap in their morphology, either in one or in several organs. If A', A", A', A*, etc., are the representatives of one certain t3'pe of animal, inhabiting, for instance, each one particular island, and W, B-, B^, B^, etc., those of another type of animal found on the same islands respectively, there is a corporeal gap between A' and A', etc., and between B^ and B^, etc., which do not live together, as well as between A' and B', A' and B", etc., which live together. And now tlie question arises. Are the gaps between tiie various A's * The ordinate gives the number of individuals, the abseissa the degree of difference. ( xxxvi ) and tliose between the various B's ])ioloy,ic.ally equal to tlie gaps between the synoecic A's and B's ? Does tlie barrier which we know to exist between the synoecic A' and 15', which occnr together, exist also between A^ and A-, wliich are geographically separate ? Tiierc arc many scientists who say Yes, and many who say No. Let ns consider first tlu; arguments advanced for the correctness of an affirmative answer. (1) The geogra])liical representatives A' and A- are constantly different ; ergo, there is a constant barrier between them, as in the case of A^ and B'. — Firstly, this is begging the question. Secondly, we have seen above (p. xxxiii) that corporeal differences as such do not constitute the barrier existing between A' and B'. Thirdly, in a vast number of cases it is mere guess-work to maintain that A' and A^ are constantly different ; all we know of them is that they are different under the special conditions under which they are living, similarly as seasonal varieties may be constantly different if the conditions are constant. If the constancy of the special conditions falls, the constancy of the corjioreal difference between A^ and A^ will certainly or perhajis — we cannot tell a jjriori which is correct — also break down. (2) A' and A^ are geographically isolated. ' Tiiey form therefore separate biological entities which do not interbreed and fuse. — We reply, firstly, that this is again a restatement of the question ; secondly, that, as there are numerous cases of identical individuals (as far as there is identity in individuals) being geographically separated from one another, geographical isolation as such is no criterion whatever ; and, thirdly, that the facts of A' and A^ being geographically separate, and A' and B' living together, constitute certainly not an agreement, but, on the contrary, a difference in the relation between A' and A'^, and A' and B', respectively. Onr arguments for geographical representatives not having a. priori the same biological standing as the synoecic animals which are separated by the barrier before characterised (p. xxxiii), are as follows : — (1) The geographically separated and morphologically distinguishable representatives A', A^, A^, etc., are morphologically and anatomically more similar to one another than to B\ B^, B', etc. The A's are modifications of one and the same type, the B's of another, the C's of a third, etc., and each type as a whole stands in contradistinction to the other. This is a statement of fact, not of opinion. Now, since the existing modifications are the result of evolution from the ancestral types, it follows tliat phylogeuetically A'. A^, A', etc., stand in closer relation towards each other than do the synoecic animals A', B', C, D\ etc., the A's forming one branch, the B's another, the C's a third, etc., of the phylogenetic tree. This difference in the degree of blood-relationship between the geographical rejjresentatives (= geographically separate components) of one ty[)e and between synoecic types is very conspicuous in all classes of animals where adequate material has been conscientiously studied. From the point of view of a morphologist alone, all the animals which are clearly geographical representatives of one ( XXX vii ) auothcr have a different stamling iu classification from the non-geographical entities. (2) If a small proportion of the specimens of A' and A^ are the same, and A' and A^ (or one of them) are brought under conditions which favour the appearance of the characters of this small proportion in each case, ' Lionomics teach us that A^ and A^ will in the course of time become more and more similar, overlap wider and wider, and become finally identical. Tliat is to say. A' and A- cannot live together without fusion. However, if A^ and A-, which diff'er, say, in !i9-9 per cent, of the individuals, cannot exist together as separate entities, A- and A', which are connected by intergradation in the intermediate countries, or A^ and A'', of which the ranges of variation are contiguous, or A^ and A*, which are separated by a small gap, have no chance of remaining separate entities, if by accident brought under the same con- ditions of life. As there is a gradation in the morphological difference between geographical representatives (see Fig. 2, in which the curves D', D", D', etc., are the geographical representatives), some being sliglitly, others more distinctly, and others again widely different, it is obvious that the chance of the representatives not overlapping in characters, under those altered conditions of life above referred to, becomes larger and larger, the wider the morphological gap is lietween them. From ttiis point of view it depends, therefore, entirely upon the characters of the geographical representatives whether these can exist together or not, i.e. whether there is a similar gap between A' and A- to that between A' and B'. In other words, considering the gradation in the morphological difierences, there are morphologically different geographical re])resentatives which can certainly not exist together without fusing to one entity, and there are others whicli may be so far settled in tlieir characters that a fusion will not take j)lace. From what we iiavf said it follows tliat it is wrong to treat all geographical corporeally different forms indiscriminately as being separated by that same barrier wliich prevents synoecic animals from fusing. As science does not knowingly perpetuate errors, the contention which we have demonstrated to be erroneous shonld be abandoned bv those who claim their ( xxxviii ) vvuil< t(i lie wliolly scientific ; ami iioinciiclutiirc, Mie language of classification, should therefore have a diflerent formula lor the two different conceptions, if classification is meant to be i)recise. And the language recording in formulas the results of classification must lie precise, because science has specially invented the formulas to prevent confusion arising. However, Ijefore going further, let us consider anotiier side of the question. It might be argued, with some degree of justification, that nomenclature is a practical invention for the convenience of the classifier (and those who make use of classification), and, as we said aljove, must not be turned into an inconvenience. Now, if tlie classifier is compelled to distinguish between geographical 1}' isolated fonus wliich are e(|ual to distinct synoecic animals, and such which are not, and again between geograjthical and non-geographical corporeally distinct animals, an amount of work is thrown on him which he could easily avoid by treating all definable forms in the same way, as being all co-ordinate. Those of our readers who follow classificatory literature* are aware that we are not stating an imaginary case, with the object — as so often happens in science and elsewhere — of demolishing an opinion which nobody entertains. The procedure above advocated is certainly a great simplification of work. But we may be allowed to state in resjMnse that a far greater simplification can be attained by putting all geographical modifications of the same type down as identical, nomenclatorially and classificatorially. If the splitter who indiscriminately calls everything definable a " species " claims the right to live, ourselves as impartial onlookers must concede the same right to the lumper who treats everything not conspicuously different as being identical. These two opponents are almost hopelessly at loggerheads. We say almost ; for we perceive some glimmer of hope of a good ending in the fact that, as in the struggle between two nations the victor becomes generally influenced by those details of civilisation in which the vanquished excels, both the lumper and splitter will profit from one another in the course of time, both remaining victorious in the struggle till the end, the one knocking over ■what the other considers distinct, and the other separating again what the former has put together as the same. Do not let it be truly ad infinitum. We sympathise with both ; for we believe that we understand the cause of the struggle and the reason for its unreasonable continuance. When Linne invented his nomenclature (binominal, except most Lepidoptera) for his classification, he started with the conception that the animated world was composed of a great number of forms which, though variable to a certain — mostly small — extent, were fixed entities. These entities he defined as species, and designated each with a name. The introduction of a precise discrimination (or what was meant to be precise) between the species, and their fixation in classification by a name, gave a great impulse to collecting and studying the forms of animated creation. During the last century the material increased at such a rate that the number of species known to Linne became soon insignificant as compared with what * See, for instance, Vtrli. V. Intern. Hool. Conijrcss p. 910 (1902). ( xxxix ) more extensive collecting brought to light.* Among the new materials the systeniatist (bund a mnltitnde of forms connecting by all kinds of intermediate grades a great nnmber of hitherto well-separated " species." In dealing with these intermediates systematists adopted three methods. The one class of describers, disregarding variability, thought to carry out in the correct way Liuue's method of classification and nomenclature by applying Llune's term " species " to every group of individuals which they found to be definable. As every individual differs to some extent from the other, every specimen was naturally a trap for tliese authors, wiio continually considered individual differences to be specific, and lieuce described an army of " species " which bad no standing at all. Another class of systematists, noticing the links connecting what otherwise appeared to be distinct " species," were inclined to regard everything similar as being the same. They forgot that the first object of the describer is to distin- guish between what is distinguishable, and they were still further carried away by the reaction against the indiscriminate creation of species which necessarily set in. There was, therefore, a kind of excuse for the one as well as for the other class of systematists ; the one student erring in being too zealous in applying throughout what he thought to be the generally adopted Linnean method of dealing with the animated world, and the other falling into mistakes by over-zealously correcting the errors (or what appeared to him as such) of the former. The right })ath lies, as it mostly does, in the middle between those followed liy tlie extremists. A third class of students, keeping their mind unbiassed, were led along this middle path by their own power of discrimination. They learnt from the investigation of the mass of material in museums and private collections, and from observations on living specimens, that neither everything similar is identical, nor everything dissimilar is specifically distinct. It is the lack of discrimination which prevents either extremist from finding the right path. However, the work of the splitter has a great advantage over that of the Inmper. The differences which he points out between tne * TLe numbers o£ species described by Linne in Syst. A'at. ed. x. 1758 are as follows:— Mammalia • 184 Aves 551 Amphibia 218 Pisces 378 .Colcoptera 665 Hemiptera l'J7 [jepidoptera ......... 542 Insecta • Neuroptcra ... 60 Hymenoptera 22!) Diptera 1*^8 >-Aptera 22'.l Vermes 027 Total . . . .4371 ( n1 ) animals are tlieie. In tlie statements of tact he is correct ; but he errs in the interpretation of tliese dift'erences. His sight is keen, liis reasoning— less so. Tiie Inmper, on the other hand, does not perceive the differences, or lie perceives them only in a hazy way ; he jnits them down as insignificant and passes on, halting only if there is a conspicuous quantity of difference wiiicli impresses itself on his mind. His eye and reasoning are on the same level, and his work is generally scamped. As he depends in his judgment on the conspicuousness of the characters, he naturally falls constantly into the error of treating as the same what differs in non-contrasting colours or in vaguely perceived structures, and as distinct what differs in contrasting colours, in shape, size, and other easily noticed characters. It may ajipear ludicrous to the non-initiated, hut it is nevertheless true, that in one and the same group of animals— for instance, Lepidoptera — the same kind of difference is considered by the same authors as being of no significance in the case of small forms, wliere it requires careful research to find the distinctions {Ejnplemidac, Geometridae, Tliyrididae, etc.) ; while it is treated as being specific in the case of large ones, where even a dull eye cannot fail to perceive the difference {Papilio and other butterflies). Sombre- coloured animals (some Sphingidae, Noctuidae), small forms {Epiplemidae), and such with a uniform pattern {Micronia, for instance), are generally great stumbling-blocks for the lumper. The author who treats everything definable as distinct has at least method in liis errors ; the author who depends entirely on what appears to him to be a sufficient quantity of difference is quite arbitrary in his judgment. The errors of an author arising from physical shortcomings, lack of training, and a certain flightiness in reasoning may leniently be passed over, if the mistakes ai"e not persisted in when they have clearly been pointed out. However, the direct misstatements as to variation and distribution, which the Inmper is more liable to make than the splitter, are scarcely pardonable. If it is the object of the systematist to elucidate facts, and not to conceal them or to misstate them so as to mislead, it is wrong for an author to suppress distinctions whicli he has noticed; to refer to differences which are geographical, as if they were individual ; to consider characters as seasonal which he knows not to be seasonal ; to pronounce a form constant, if he knows only one or a few specimens ; to treat another as being individually variable, though he knows only single individuals from different countries. In short, it is wrong — and may become objectionable, because bordering ou charlatanism, if the misrepresentation is made deliberately — to represent anything to be what it is known not to be, or what on a little consideration it would have been known not to be. Neither tlie author who considers every form definable as being specifically distinct, nor his opponent who treats as identical everything of which the distinguishing characters do not appear to him to be conspicuous enough, fulfils one of the primary demands on the conscientious classifier— namely, to discriminate carefully between the differences presented by tiie various forms of animals he lias to deal with, instead of lumping and sejiaratiug indiscriminately. ( xli ) Wl' are aware tliat a gnod many systematists, botli among amateurs and jirof'essiouals, have no higher object than naming and arranging the material in their collections, and maybe istsning liooks to enable others also to name and arrange the specimens, the knowledge aimed at being a knowledge of some distinguishing characters, and especially the name of the " species." However, the knowledge of tlie alphabet does not carry with it the knowledge of the language, and lie wlio knows the words and speaks a language is not yet a philologist. So there is also in classification a higher object to be attained than merely describing, baptising, and arranging in some arbitrary order the forms of animated nature. Tliis higher object is to understand tlie ])hylogenetic relation between the forms, and on this understanding the scientific classifier bases his system. In order to comi)rehend the connection between tiie forms, it is necessary to know what it is that separates them. It was the one kind of difference to which we have referred before, separating the animals which e.\ist side by side, the one effective barrier consisting of differences in the organisation of the animals themselves, which was the keynote to the Linnean Reforniation of Natural Science and to the Darwinian Revolution. The individuals within the barrier form an entity which has an existence indejjendent of all the other entities. Each entity was in the Linnean classificatioii understood to be a special creation, and the effective barrier to be intended to ])revent fusion of the entities. This was the conception to which Linn6 applied the term species. And this was again the conception which formed the subject of Darwin's Origin of Species. The great mystery which the theory of descent sought to explain was the fact of the co-existence of such innumerable independent species, all separated by that gap which we know to keep the species apart. How did this specific barrier come into existence, if not erected by special creation ? That is the fundamental question which is before scientists. The question is not solved by looking it straight in the face and then shelving it by applying the term " species " to something else than what it originally meant. No friend of true research should let pass unchallenged what so many classifiers nowadays try to do — namely, to sul)stitute for true species the geograpliical form. We know that a specific barrier exists between synoecic animals ; we know that there is a morpliological distinction between geographical representatives. Can the one barrier which we know to be specific honestly be replaced by the geographical difference which at the highest may be assumed to be specific, and of which we know tiiat it is not always specific? We shall oppose any such attempt at uiulerh;ind sliiftiug of tlie meaning of the term " species," which would misguide the public and prevent the student himself from seeing clearly the question at issue. If the specific barrier is the result of the evolution of the organic world by natural causes ; if, further, the multitude of species is the outcome of the divergent development of species into a greater number of species, — then we have to search for the rudiment (= beginning) of the specific barrier among ( xlii ) clifTeroncps ionnd within a siiooios Ix'twcen tlic various com])niicat varii'ties and iiulividiiiils. Tlicrdore it is nec.essar)', in order to nnderatand tlie origin of the specilic liarrier, to study the varii'tal difTerences, and find out among which varieties tiierc is a ruch'uieutary sjie('ific harrier, and lience wliich varieties are rudimentary (= incipient) species. It has been shown by one of us* (and therefore we do not again full}' enter into the same question) that the development of gamogenetic species into two or several species is not possible without an effective extraneous barrier between the varieties, whicii barrier prevents the fusion (if tli(! varieties, as does tiie specific barrier the fusion of the species, and, furtlier, tliat this extraneous barrier is provided by geographical separation. Isolation of one or more mutating fiictors is the cause of the portion of a s])ecies subjected to them becoming different from the other components which stand under other influences. All our researches confirm tliis conclusion based on the facts of variation, and all attempts to demonstrate the possibility of the separation of a species into several without some kind of local isolation are fallacious in reasoning. Geocjraphicdl variation leads to a muHi plication of the species; non-geographical variation at the highest to polgmorpliism. Geographical variation is, therefore, of another kind than non-geographical variation, and therefore geographical varieties have a different standing in the evolution of the organic world from the individual and generatory varieties. Geographical varieties as incijnent species are the next classificatory category below species, just as subfamily is a degree lower than family, and no better term could have been invented for them than subspecies. With subspecies we designate, therefore, nothing else but the geographically separated different components of one and the same type, which components represent together a species. The criterion of a subspecies is not a certain amount of difference, but bodily difference and geogra]ihical separation. Synoecic varieties — i.e. varieties from the same locality— are never subspecies. We have to empliasise this distinction, as many authors constantly confound subspecies with non-geographical varieties. Tiiere are comparatively very few species which do not vary geograi)liicaily. It was an ardent opponent of Darwin — Wiegand — who put forward as an argument against the theory of evolution that geographical variation was a conditio sine qua non for the correctness of the theory of descent, and that there was no sucli general basis for evolution. Systematists have proved by their minute research that geographical variation is the rule and not the exception, and they may be justly prond of this result of their untiring labours. Curiously enough, uon-systematists do not generally seem to be aware of this result, nor to fully comprehend its bearing on the theory of descent. A sjiecies which has not developed into subspecies (= geograpliical varieties = geographical races or forms) may be individually or seasonally di- or poly- morphic, and similarly the individuals of a subspecies may all fall into seasonal and these into individual varieties. As the species of a genus are co-ordinate * Boe " Meclianical Selection " in iVor. Zool. iii. p. 126 (1896); " ReprocUictive Divergence, etc., in Natural Science xii. p. 45 (189i<). ( xliii ) with one another and subordinate to the genns, so are the subspecies co-ordinate with one another and subordinate to the species of which they are the components. Since Linne applied the term tarietas to the forms which are not speciiically different, we do not see any reason against the use of this very convenient word in the same sense for all the components of a species which differ from one another. We understand, therefore, under rarieti/ not a particular category of the components of a species, but employ the term for all the different members of a species indiscriminately. The different categories of varieties must receive special terms in a precise classification, and special formulae must be employed for them in a jirecise nomenclature. We distinguish three categories of varieties, namely : — I. Individual variety. — The following terms are employed by us : (1) ab. = (iberratio for individuals which stand outside the normal range of variation. (2) /■ = forma in the case of di- and polymorphism. If a form occurs rarely, it may be termed f. ah., in contradistinction to /. norm. (3) ?-y. or S-J., if the respective form belongs to one sex only. (4)y. loc. = Jorma alicuius loci, if, in the case of polymorphism, a form is restricted to one portion of the range of the respective variety or species. II. Generator// variet;/. — This variety is seasonal in Lepidoptera, and is designated as (5) /. t. = forma tempestatis. III. Geographical tarietij or subspecies. — This is the highest category of varieties. As the term varietas includes also other varieties, it cannot be employed as such for the geographical variety except in a precise nomenclature ; either a specifying attribute must be added {car. geogr.'), or an abbreviation of another term chosen (subsp.). But we do not see that it is at all necessary to put any such abbreviation of a term before the subspecific name. We can do without the encumbrance of the abbreviation — what we can do without is unnecessary ; aud what is an unnecessary encumbrance in nomeuclaturc, common-sense comj^els us to drop — by (6) Simply mutually agreeing that a subspecies is designated by its name added to that of the species without any abbreviation before the subspecific name. This means simplification of nomenclature, nothing else. The following diagram illustrates the nomenclatorial relation towards each ( xliv ) otliiT of tlie various categories of c lassificato -y nmti- from tlie imliviibuil variety U]l\V!U-(ls to tll(! gciHis : — ' subsj). sulisp. - f '■ '■ ■ If.,. • f. t. ' f. f. . al). ?-f. I ?-f. ' S-f. (?-f. 1 sjiecies subsp. ft.- ?-f. ?-f. , ab. ab. ab. ab. f. norm. genus species species species subsp. f.t. f.t. f f. norm. ( f. ab. [I Tliei"te seems to us to be some confusion existing in the minds of some systematists with regard to the degree of phylogeuetic relationship of the animals classified and the nomenclatorial position of the units towards each other. Although it is the highest object of the researches of the systematist to elucidate the phylogeuetic connection between the classificatory units from the individuals ujjwards to the family, order, etc., the linear arrangement of the units in the system gives but a very scanty elucidation of their evolution. If a geuus is a development from another, it may be put behind the older one ; but this method at once breaks down if there are several genera derived from one, especially if one of the derivatives has again given rise to a series of genera. And uomenchitorially the classifier can do even less. In nomenclature all the units of one category are co-ordinate units ; all the species, all the subspecies, all the genera, etc., are respectively co-ordinate with one another. The classifier cannot make any nomenclatorial difference whatever between phylogeuetically younger and okler genera, between tiie jiareut- and daughter-species, between the generalised and specialised subsjjecies. We have the same nomenclatorial formula for every genus {Papilio, Fringilla), for every species {Papilio priamus ( xlv ) Frinqilla coelehs), for every subspecies {Papilio priamiis poseidon, Pnpilio priamns priamus, Papilio priamus coelestis). If the classifier wishes to represent the probable phylogenetic origin of the nnits which stand co-ordinated in his system and nomenclatnre, he has to take recourse to the figurative tree, or must give a kind of pedigree, as we have done in the present work. Fringilla coelebs does not mean that coelebs is a derivation from Fringilla, but that it forms part of the genus Fringilla : and Papilio priamus poseidon must not be interpreted as signifying that poseidon is derived from priamus, but that it is one of the several components wliich together form the species Papilio priamus. This confusion of ideas has been occasioned by the unfortunate habit, which many classifiers cannot shake off, of regarding the first-described comi)onent as the typical form of a species, as the " Stammart,"' * as the phylogenitically oldest portion of the species, or in the case of a genus as the phylogenetically oldest species of the genus, while it is merely the accidentally first-baptised form. Is the distinction between what is phylogenetically and what is nomenclatorially the oldest really difficult to perceive and to comprehend ? The classification of the lower categories from individual variety to species is in a different position to that of the higher categories (from genus upwards). While the species and varieties are realities which can be tested by observa- tion of the live specimens and by experiment, there is no such test possible in the case of genera, tribes, families, etc. These higher categories are definable groups of allied species. The criterion of their being realities, or, as one is used to say, of their being natural, is threefold : — (1) The grouj) must be definable — i.e. must not so intergrade with another that there is no line of division, or that the line of division is arbitrary. C^) The contents of each group must be homogeneous. Elements of different origin, though perhaps similar in consequence of convergent development, must not be brought together. (3) Each higher category must have a separate definition based on other characters than those referred to in the definitions of the respective lower categories. The characters which make an animal specifically distinct do not make it also generically distinct, nor can the same character \\\w\\ which a genus is based be employed to characterise a tribe or a family. To define genera and higher units is not always an easy matter. In order to render a definition precise, a close study is necessary of the forms which come under the unit defined, as well as of the forms of the allied units. The difficulties encountered have induced many authors, especially in Ornithology and Entomology, to propose names for genera, subfamilies, and families without attempting a definition. The naked names thus introduced are a fit testi- monium paupertatis for their authors. In our opinion, the thoroughness of the researches in systematic work can best be estimated from the degree of exactness * The word '■ Stamm.ait " is much eiiiplo3'ed by Gcrmau writers in this erroneous and misleading sense. .See Reichcnow, in Verh. V. Intfi-n. Zool. Cumjres.i p. yil (I'JO'i). ( xlvi ) of the definitions of jiencra and liiglier nnits. For tlie degree of exactness depends here (1) on tiie more or less intrinsic, study of a larger number of forms than in tlie case of species and varieties, and (2) on the correctness of tlie author's inductive reasoning. The classificatory category coming in our system of classification next to the species is the genus. Some authors iuterjiohxte between genus and species a category they call subgenus. From a comparison of a number of subgenera which have been defined, we are able to state that they had either no standing, being groups of species quite arbitrarily put together, or they represented well- defined homogeneous groups — i.e. were equal to a genus. We do not see any possibility of distinguishing between a defined genus and a defined subgenus. To call one defined group a genus and another defined group a subgenus is quite arbitrary. As there is a gap between every two species, and mostly between the varieties as well, and, further, as the species fall into groups different in extent or dififerent in composition, according to this or that organ being taken as the basis of the groujjing, the limitation of the genera would be entirely left to the personal opinion and ability of each individual classifier, if there was no general definition of what kind of classificatory unit a genus is meant to be. Systematists differ, indeed, very much in the extent given to genera, some authors adhering to large unwieldy groups of the Linnean type, and others erecting a genus for almost every species, and sometimes even for subspecies. In order to check arbitrariness, to escape uncertainty as far as possible, and to 'make generic classification more stable, we think it advisable to define a genus as a classijieatonj unit one category higher than species comprising one definable group of' species. In many cases the group contains only one species by the otiier members having become extinct, or by there being as yet only one sjiecies known, the other species being still undiscovered. In the present Revision of the Sphingidae it has been our special endeavour to give a solid foundation to the genera, supplementing and rectifying the vague or faulty definitions with which the workers in this gronj) of insects have contented themselves. Though many genera hitherto considered to be valid have been shown by us to have no standing, the number of genera of Sphingidae has been much enlarged, owing to the closer examination of the insects proving many groups of apjiarently similar species to be heterogeneous. We have laid special stress upon the genera as conceived by us representing stages in the evolution of the Sphingidae. There is nowadays a tendency among British Lejiidopterists to imitate some American leading spirits in Lepidopterology in shifting the term " family " (J'amilia ; designated by the ending -viae according to common consent) to a lower category than that to wliich it was originally applied. We do not see what good it serves to call, for instance, all the Hawk Moths together a superfamily, and the next divisions of it families. It is an entirely superfluous innovation, and only leads to confusion, like all shifting of ( xlvii ) terms. We divide the SphiiigiJae into the i'ollowing eight classitiiuitory categories : — I. Individual variet}'. II. Generatory variety = seasonal variety. III. Geographical variety = subs]>ecies. IV. Species. V. Genus. VI. Tribe. VII. Subfamily. VIII. Family. Before we proceed to state some of the general results of our study of tlie Sphinqidae, we give a summary of tlie raor[)hology of these insects, which will enable the reader to more fully comiirehend some of the conclusions bearing on phylogeny and distribution. MOKFHOI.OGY OF SFHINGISAE. The dorsal skeleton of the head (PI. LXII. f. 6) is divided by two transverse sutures into clypeus {cl), epicrauium {ecr), and occiput {occ). The last is always a narrow transverse plate more or less distinctly placed at an angle to the convex epicrauium ; it is as a matter of course longer transversely in those species in which the eyes have a more oblique position than in those which have less obliquely placed eyes. Comjiare PI. LXI. f. (i and PI. LXII. f. G. Its anterior edge is generally faintly incrassate in tlie middle. The ej)icranium forms laterally the sockets for the antennae, which stand nearer the eye in some Hawk Moths than in others ; the suture between epicrauium and clypeus is just in front of the antennae, ending in the antennal grooves. The clypeus is the largest plate of the three; it is more or less strongly convex, especially mesially. It bears at the anterior margin the labrum {Ir, PI. LXI. f. 6. 8. 9. 10. 11, also PI. LXII.). The two are almost merged into one, the suture being mostly not distinct. The labrum is in most instances raised to a large, transverse, cariniform tubercle, which is generally vertical in front. It projects sometimes frontad over the base of the tongue, concealing the mesial part of the epistome {ej>), and is occasionally very small (PI. LXII. f. 3). The epistome is a transverse plate of variable dimensions situated in front of the labrum, with which it is so completely fused that it is not easy to say where tlie one begins and the other ends. It is especially hirge in a number of Ambnlicinae with otherwise reduced mouth-parts {rp, PI. LXI. f. 10. 11 ; PI. LXIL f. 4. 5). This epistome covers the base of the tongue. When normal, it has a thin mesial lobe and a large process at each side. The mesial lobe varies in size and somewhat in shape, and is rarely absent (PI. LXI. f. 10 ; PI. LXII. f. 1), this occurring when also the other mouth-jiarts exhibit a high degree of reduction. ( xlviii ) The lat(M-al iirocesses {p, Tl. I>X1. f. 0— 11 ; PI. LXII. f. 1—5) arc designated "pilifer" by Kellogg*), and are often erroneously considered to be homologous to the niandildes. Tiie normal jnlifer is a curved obtuse process, concave and flattened on the inncrsidc, and is licset on tiie inner surface with a great number of long stiff bristles wliicii jiroject over the base of the tongue, which they touch. Tiie pilifer and its bristles undergo various modifications. The bristles become modified into scales, either partly or all (PI. LXI. f. 11), or they become fewer in number and disappear finally nearly completely (PI. LXI. f. 10). At the same time the pilifer may become shortened and lose the appearance of a process, Leing represented in the most reduced state known to us ])y a broad obtuse projection (PI. LXI. f. Id). Two other modifications are represented by PI. LXII. f. 4. 5. In fig. 5 the pilifer {p) is almost stalked, and projects far bej'ond the mesial lobe, which is large. In fig. 4 the whole epistome is enlarged together, jiroduced forward, the pilifers being close together and very Lroad, and i)roj('cting little beyond the mesial lobe, which is very small. Quite different from these modifications is that observed in all the species of Choerocampinae, and only in this subfamily. PI. LXII. f. 2 represents a species of Celerio : the pilifer {ji) is long, somewhat twisted, and the apical portion is clothed inwardly with short bristles, wliile the proximal portion bears the ordinary long bristles, the two kinds of bristles contrasting very strongly with one another. This character is of surprising constancy ; its significance in classification will be understood, when the characters of the palpus and antenna are taken into consideration at the same time. Between i)ilifer and eye, supporting the former laterally, there is a more or less triangular projection, which is an enlargement of the brim which separates the eye from the large labial cavity of the underside of the head. We term this projection " geual process " {gp of PI. LXI. and LXII.). The upper portion of this genal i)rocess is often distinctly separated by a suture, and corresponds to the mandible {;md, PI. LXII. f. 4), as pointed out by Kellogg, /. c. The genal process is very large in Macroglossum (PI. LXI. f. 0), Sesia, and allied genera, reaching often to the tip of the pilifer. In the Choero- campnnae (in all of them) it is smaller than anywhere else (PI. LXII. f. 2), not showing in a lateral view the distinctly triangular form observed in all other Sphingidae with well-developed tongue. The suture between genal process on the one side and labrum and clypeus on the other ends often (many Ambulicinae) in a deep groove (J', PI. LXI. f. lU. 11). If we examine the underside of the head after the removal of the labial paljii (PI. LXI. f. 7 ; J/p is the groove in which the palpus is inserted), we find again the pilifer and the genal process between eye and tongue, and observe below the pilifer close to the tongue on each side a short process, dilated apically in the s])ecies figured, which process is the remnant of the maxillary palpus (tnxp). It is in most cases densely clotlied with long wiiite scales, which project beyond the pilifer and genal process, being visible also in dorsal and lateral aspects of * Amer. NuUiralist xxix. Ip. 546 (1895). ( xHs ) the head (PI. LXI. f. 6. 8). Tlie size of the vestigial maxillary ]ialpns is not constant in the i'amily, nor has the ])alpns always the same sliape. Tlie transverse arched stripe of chitin between the labial palpi is the mentnm ; in front of it ^e find often a vestigial, very feebly chitinised submentam. The before-mentioned month-parts of Lepidoptera have attracted much attention on the part of scientific entomologists since ^Yalte^'s now famous paper on the mouth-parts of Miero/iti'ri/.r.* The distinctions exhibited by them within the families of Frenata have, however, not been made use of in classificatory work. Tlie parts are covered by the labial palpi as a rnle, and are not visible without jiushing the palpus away from the head. A drop of benzine, or, better, a drop of alcohol, ajiplied to the base, is generally sufficient to make the jialpus so flexible that it is movable, and allows the geual jirocess and pilifer to be studied without injury to the specimen. The two jiarts of the cajnital appendices which remain to be discussed, tongue and labial palpus, are better known to the i;lassifier, though the descriptions given of them go seldom beyond length of the former and imtline and general aspect of the latter. The length to wiiich the tongue has developed in the family Sjjliinyii/ae is an exceedingly striking character. Here we find the longest tongne of all insects. But what is far more interesting for the student of comparative morphology as well as the classifier is the fact that the length of the tongue varies in this family to such an enormous extent as it does, the extremes being represented by Coe^titis, in which the tongne is sometimes little short of 25 cm., and Poli//jt'/c/t>fs, where we find species with a tongue represented by two tubercles barely longer than 2 mm. A comparative study of the tongue (j/los.sa) of Lepidoptera is a desideratum. It is formed by the first pair of maxillae, and consists, as is well known, of two halves closely applied to each other (PI. LXII. f. 2). Each half is concave on the inner side, and bears at the upper inner edge a very dense fringe of ciliae. The trans-section is in Spldngidae short kidney-shaped, or nearly circular, apart from the inner concave jjortion. Laterally at the base the tongne has very often a patch of minute hairs ; in a few cases hairs are found all over the dorsal surface. Within the cavity of each half we find, in dry specimens, a large trachea and the residue of the dried-up muscles, nerves, etc. The sucking-tube itself (PL LXIL f. 2, tu) formed by the two halves of the glossa is closed above by the fringe, the ciliae of which are soldered together to form a membrane, which is often (pite smooth, showing no trace of transverse striatiou indicating the ciliae. When the tongue becomes reduced, the two halves are less firmly applied to one another, and the transverse striation of the closing membrane of the tube becomes distinct, till with the further reduction of the glossa the two halves separate and the closing membrane assumes the form of a fringe of separate ciliae (PI. LXI. f. 11 ; PI. LXII. f. 4. .3), this fringe finally disappearing (PL LXI. f. K) ; PI. LXII. * Jen. Zeitschr. Naturw. v. 5. p. 751 (1885). d ( 1 ) 1". 1). Thf I'liiictioiilcss tdiigne loses the transverse animlatiou ; it is very f'celiiy chitiiiised, and varies imliviiliiaily in iengtii. it not rarely bears scales, and is oci'asionally tnberciilated. The longest tongue diciirs in tiie tribe Sphingicae, which tribe contains, however, also species with a very short and functionless tongne, and one species with jnst a vestige of it {Elloihrrhiit). A short tongne is fre(piently found among the Amhiiliciiiae, in which subfamily it is always shorter than the body, but often strong, and in the latter case not showing any sign of reduction. In the SphiiHiulicae the tongne is also very weak and short in most species. In the other groups of Spldngidae the tongue is never excessively long and never very short, always preserving the ordinary structure and rigidity. It is scarcely necessary to point out that the functionless glossa is a derivation from a tongue efficient as a sucking-tube. As varied as the other month-parts are also the labial pali)i, shortly designated as " palpi " in Le[)idoj)tera. Though this designation is, strictly speaking, not correct, it is very convenient and quite precise enough, as there can be no doubt that the labial palpi, and not the vestigial maxillary ones, are meant, when speaking of the jialpi. The palpus, if not reduced, is large, broad in lateral aspect, closely contiguous to the head, and has a short third segment. A palpus like this does not occur outside the family. However, where the ])alpus is reduced, the general aspect is insufficient to recognise it as a Sphiugid palpus. The most slender and at the same time longest palpus is found in Tinostoma and in the S of Cressotiia, in which latter genus the S pal[)i are strongly divergent. A very large and rounded palpus is met with in Paehi/lia, Earypteri/x, Protoparce, and other genera. Reduced palpi occur abundantly among Amhiilicinae and Acherontiinae, while the palj)i of the other Sphingidae are mostly of medium size or large, very seldom small. The size of the palpus depends on the width of the segments and upon the scaling ; a broad segment covered by short scales appearing mucli slenderer than a narrower one with long erect scaling (PI. LIX. f. 20. 27). It is, as a rule, also the scaling which gives the terminal portion of the palpus its particular shape. The triangularly pointed palpus and the broadly rounded one, as described in systematic work, may have similar segments when denuded. If one sjieaks of the shape of the palpus, one means the palpus inclusive of the scaling as it appears in a perfect specimen. There is distinct sexual dimorphism in the palpus among Ambuliclnae, the palpus of the male being often larger than that of the female. There are always three segments ; the third is, however, nearly always very short and concealed in the scaling of the second, projecting as a little knob. In the few cases where the third segment is more distinct it is conical, sometimes naked and horn-like (Coryfi/is ; Xanthopan); it is never long, slender, and rod-like. In narrow jialpi it is often as broad at the base as the second segment is at the end. The first segment is the longest as a rule, but there are many exceptions ( H ) in wliicli the first is not longer or is shorter than the seeond. The segment is cnrvetl, lying along the eye. The inner surface (PI. LIX. f. 20. 27 ; PI. LX. f. 1) is more or less regularly annulatcd or wrinkled, flattened, or slightly convex, or somewhat concave. It is naked, except the edges, with some lono' hair-like scales ; or it is more or less loosely scaled for the greater part. The scaling at the apex of the lirst segment (PI. LIX. and LX.) is either short, or long and rough, or long and quite regular, affording in several cases conspicnous characters of taxonomic value — as, for instance, iji the subfamily Clioerocampinae, where the two Neotropical genera and the cosmopolitan genns Celen'o have tlie scaling always rough and irregular, while all the allied Old AVorld genera with the exception of Pergesa and Rkodafra, derivations from Celerio, have it regular (PI. LIX. f. 19 — 2L 26. 27). A character of the greatest importance in the classification of the Hawk Moths is found at the base of the first segment. Tliat is a patch of variable size of short (and doubtless sensory) hairs, which is always jiresent in one section of the family Qi, PI. LIX. f. 26. 27), excejjt a few reduced forms, and equally constantly absent from the other section (PI. LX. f. 1). The trustworthiness of the distinctive character was discovered after we had separated the Acherontiinae {= Aeherontncae + Sphingicae ■\- Sjihingidicae) and AmhuUcinae from the rest of the family on other grounds. This basal patch was found in Butterflies and treated upon at some length by Renter.* It is of wide occurrence in Moths. Its absence from Acherontiinae and Amhidicinae has nothing to do with the reduction of the palpus, as it is absent from the large palpi of Protoparce and other Sphingicae, while it is found in the other subfamilies on the relatively small palpus of Orjjha, Berutana, and other genera. The basal patch is present in Geometridae, Notodontidae, Agaristidae, Nochiidac, Pijralidae, etc., etc., and assumes sometimes an obviously distinctive form. ^Ve have not noticed it in Saturniidae and allied groups, nor among Lasiorampidne^ Bombi/cidae, luiptcrotidm', and some other groups. How far this organ can be made use of in the classification of these families more complete research must show ; but we are justified in maintaining that the basal patch will prove itself elsewhere an e(|ually good distinguishing character as we find it to be in Spkingidai\ The first segment of the palpus exhibits also on the outer side characters of taxonomic value. There is a kind of transverse crest near the eye in Basiothia, Alcuron, and Unzela, and the apex of the segment is strongly convex externally or angulate in these genera ; the crest is also found in (Jurelca and Sphingo- nnejiiojtsis. A peculiar modification is found in one genus of Acherontiinae {Mcgacorma) and in a great number of Choeroeampinae. It is illustrated on Pis. LIX. and LX. There is at the apex of the segment, ventro-laterally, a space devoid of the ordinary scaling, being either quite naked or clothed with a few long hair-like but flat scales. The scaling around this naked space, • Ada Soc. Si: Fcnn. xxii. 1 (18',t6). — Keuter says that the basal patch of Ileterocera is never elevated as in Rhopalocera. We finil that it is sometimes raised to a conspicuous ridge, for instance in Pijralidae. ( Hi ) wliicli is often soiucwliat concave, is more or less regular, especially ventrally, ami, surrounding the naked space, forms a kind of cavity {gr, PI. LIX. f 13. l(i. 17. 18. 25 ; PI. LX. f. 3). The naked membrane is doubtless seusory, but we could not find any external sensory organs in the dry specimens. The long hair-like scales situated in many species on this naked membrane are not of a sensory nature. There is every intergradation between a rudimentary cavity and a large regular one. This specialisation is found only among Oriental and Aethiopian Sphingidae, not in American ones. Tlie joint between the first and second segments is exposed in the species which exhibit the cavity. A naked and exposed joint is also met with among Ambulicinae, but no cavity. The second segment undergoes many modifications in shape ; it may be subcylindrical, quadrangular, triangular, ovate ; it may be longer or shorter than broad, or square. In Tinostoma it is three times as long as broad. It is angnlate at the upper outer corner in Aleuron and Unzcla. The scaling of the two palpi generally covers the base of the tongue ; but there are instances where the mouth-parts are exposed. This is the case when the scaling of the second segment is very short, as in PI. LIX. f 10. 11. 20. The inner surface of the second segment exhibits some remarkable s])ecialisa- tions. It is normally scaled all over, but we find tlie scales very small and rather dispersed in Euchloron (PI. LIX. f. 24), so that the membrane is partly naked. In all the other species of Choerocnmpinae these small rounded scales have all, or nearly all, disappeared, leaving the segment bare except at the edges (PI LIX. f. 19—21. 26. 27). The character is quite constant, and occurs only in those species which agree also in certain characters of the pilifer and of the antennal end-segment, and form the subfamily Choerocampinae. The scales at the upper apical angle of tlie segment with naked inner surface are either short or form a conspicuous tuft, wliich projects ventrad (PI. LIX. f I'J — 21. 26. 27). This difference is of importance, the absence of the tuft being characteristical for the fifty species of the purely American genus Xylophanes, which has, unlike the Eastern representatives of the subfamily Choerocampinae {Rhodu/ra and Fergesa excepted), the scaling at the apex of the first segment irregular, as stated above. AVo have not ascertained the true nature of the naked membrane, but assume that it serves as an organ of sense. A specialisation reminding one of that just described is found in the genus Pailogvamma of the subfamily Ackerontiinae. Here the segment bears a naked longitudinal stripe not far from the upper edge, the stripe appearing as a jirolongation of the naked part of the first segment. The modification does not occur in the otherwise very closely allied genus Leucomonia. A third modification peculiar to the Acherontiicae, but vestigial also in one genus of Spkingicae, is represented by figs. 1 and 2 of PI. LX. Here the second segment is concave on the inner surface ; the scales at the edge of the cavity {c) project over it, forming a kind of roof. The concave part is eitlicr practically naked or scaled. The transition from the vestigial groove of ( liii ) Xanthopan to the deep aud naked cavity of Arherontia and IL'rse is found iu Megacorma and Coelonia. The third segment bears at the end a small and deep cavity of a sensory nature. AVe have not studied this organ. The great variety in the structure of the antennae* of the Heteroc^era is repeated to a lower degree in most fixmilies of larger extent, and this makes it generally impossible to give of these organs a short family diagnosis wliich is true of all the species of the respective groups. The stereotyped description of the Sphingid antenna handed down from Linnean times applies only to a portion of this family. The prismatic, clubbed, and iiooked antenna is by no means common to all the Sphingidae, a large proportion of the Hawk Moths deviating widely from this type. However, comparing each type of antenna of the present family with the corresponding types of other families, we shall always Hnd some difference, and very often some striking difference, revealed by the closer examination of the special structures. The generalised type of antenna of Lepidoptera is, in our opinion, that in which the dorsal surface is wholly scaled, while the ventral surface is scaleless, being clothed instead with a dense covering of fine hairs. This type is derived from an antenna entirely ciliated. Besides the protective scaling and the sensory ciliation there are other organs present, such as sense-bristles and -cones, both of which are rather easily discernible and of considerable taxonomic value. The scaling of the butterfly, antenna is reduced apically and in a unmber of groups altogether absent, except the first segments {Danninae, Papilios allied to sarpedon, etc.). In Heterocera we find the scaling extending to the last or last but one segment, but meet also with antennae which have lost the scaling {SatKrniidae and close allies ; Feloc/ii/ta). The arrangement of scales iu two regular transverse rows on each segment which is prevalent among Butterflies, and is found also in a very great number of Moths, does not obtain among Hawk Moths except on the distal segments in a number of species, the scales being comparatively small and very numerous. Where among Butterflies and Moths the area covered by the fine sensory ciliae is reduced, the ventral surface becomes often partly scaled, the loss of the sensory function being followed by, or going hand in hand with, the appearance of scales, in Lepidoptera the normal clothing of the epidermis where no organs with special function are required. The proximal segments of the antennae of Sphingidae are occasionally nearly or totally scaled — namely, where the antenna is strongly clubbed. That means, where the sensory function is more concentrated distally, the proximal segments, which are reduced in width and function, have acquired an ordinary covering of scaling on the underside. Tliis is most evident in Rkopalopsj/che, a genus with also otherwise remarkable antennae. The appearance of scales as a sign of loss of function, or of weakness, is strikingly illustrated by other organs. We have seen above that the reduced tongue becomes scaled in some instances, and that the bristles of the pilifer are replaced in many reduced species by * Compare Bodine, 7Va«.*. Amcr. Ent. Soc. xsiii. p. 1 (1890) ; Jordan, Xov. Zool. v. p. 374 (1898), ( liv ) scales. A I'lirtlicr fdiitii'iiiiitioii <il' the view just c.xjji'csscd, that tlie veutral scaling of the aii(enna is a relatively young character, we find in the fact that there are sometimes scales on the underside of the antennae of hybrids where there should not be any, tliese scales being doubtless an expression of weakened vitality. We predict that by breeding in and in specimens will result which show a more extended antcnnal scaling than the normal individuals. The antennae of the Acj/eriidde and CastHiidae, whicli resemble in general appear- ance somewhat a clubbed Sjihingid antenna, differ essentially from the latter in having the greater part of the ventral surface scaled, agreeing in this respect with the antennae of a great many Tineidae. Among the latter family (or group of families, jjerhaps), and among the lAmacodidar, we find antennae which are almost entirely covered with scales. Before following tlie somewhat complicated development of the sensory surface, we shall shortly refer to the sense-cones and sense-bristles. The sense- cones discovered by Bodine do not occur in Rliopalocera, but are widely distributed in Heterocera. They are ventral, mesial, and apical in Sphingidae, one on each segment, except the end-segment and the basal one or ones (PI. LX. f. 4 — 29, ec), from which they are absent. The cones of the distal segments are generally more prominent than those of the proximal segments. They do not always stand exactly at the apical edge of the segment, being not rarely removed somewhat basad (compare, for instance, Ilerse conwlmdi). However, they are never absent and never abandon the mesial position . in this family. They are wanting in the Castniidae, Aegeriidae, and Zygaenidae ; their absence is a distinctive character not difficult to recognise. They are present on the clubbed antennae of Agaristidae and Cocytia. The sense-bristles are stiff hairs of varying length. The Lepidoptera most primitive in respect to these organs have a complete belt of such bristles on each segment. Among Rhopalocera we find sncli a belt preserved in Lycaenidae and Hesperiidae ; we meet with it again among the Jiigata and many Hetero- cerons Frenatu. Some of tlie dorsal bristles are, however, generally reduced and covered by the scaling. This is the case also in Sphingidae. The normal number of the bristles found in the Hawk Moths on the non-scaled surface, if we except the end-segment, is two on each side (PL LX. f. 14), one being dorso- lateral, the other veutro-lateral and basal. The number is sometimes doubled ; but there is never a complete transverse series, and tlie bristles are never apical, the Sphingid antenna dififering therefore obviously from that of Agaristidae, Zygaenidae, and many other clubbed and non-clubbed antennae. The dorso-lateral bristles situated close to the edge of the scaled area are very often so jirolonged and become so stout that the antenna has the appearance of being pectinated ; in fact, that bristle has often been mistaken for a pectina- tion.* We meet with this development very commonly in tiliform and setiform antennae, but never among Sphingidae. The ventral and lateral surface of the antenna of Heterocera, as far as * Noi: Hool. iii. t. i. f. 5—11. ( Iv) it is not scaled, is covered witli I'iliae, only the parts near the joints and close to the dorsal scaling being in many cases naked or sparsely ciliated. Pageustecher * calls tlie antennae of CuIlithilidK,' "naked,"' and Hampson t describes those of Ai/nri.sfidat' as being " not ciliated." Both authors are wrong, the antennae of Cdllldiilidne and of Agaris'ithte being scaled above and ciliated below. There is no antenna among Lepidoptera which is not ciliated, and the term " naked " can with some justitication only be emplo3-ed for antennae which are not scaled, and therefore have a naked dorsal surface (apart from some bristles and setiferous jiits). The evenly ciliated ventral surface as we find it in Rhopalocera among the Lijcrieniilae and Ili:-<pcrn(fac, and in great abundance among the Heterocera, more especially in the female se.\, far less often in the male sex, represents a generalised state of development from which a variety of specialisations have started, which, tliongh resulting in widely different structures, have nevertheless all the same tendency — namelv, to make the antenna more efficient as an organ of sense. This is accomplished by enlargement of the area bearing the sensory organs — i.e. by the increase in their number — or by concentration of the organs, or by enlargement of the organs themselves, these modifications obtaining either singly or together. The concentration of the ciliated area into such well-circumscribed grooves as are described and figured in Nov. ZooL. vi. p. 374. t. 14. 15, does not occur among the Heterocera. The enlargement of the distal part of the antenna into a club, which is normal for Rhopalocera, among which non-clubbed antennae are extremely rare {Pseuchpontia^, is met with in a number of Heterocerous families, such as Gastniidae, Aegeriidae, Sphingidae, Zijgaenidae, Agarisfidae, Callidu- lidae, and also among Geometridae and Noctuidae. As regards mere outline the clubbed antennae of rej)resentatives of different families are sometimes not distinguishable. The Australian Castniufuf, which appear to form a different subfamily from the Neotropical species of that family, resemble in the short and abrupt club certain I'ieridae, and some Hesperildae and Neotropical Ca.sfniidae have practically the same antennal outline as some Sp/ungidae. The most strongly clubbed antenna of Sphingidae we find in Ilacmorrliagia and Rkopalop«>/chc ; from these to the setiform antenna of Megaconiia there occur all intergradations in shape. It is by no means only the ciliated surfiice which becomes expanded in the clubbed antennae of Spliingidat\ The dorsal area is often projiortionally more enlarged than the ventral area. This can best be seen in a frontal view of a segment of the club of Ccpkonodes or Haemorrhagia, where the axis of the club will be found further ventral than in a segment from the middle of the antenna. By axis we mean an imaginary hollow cylinder of the width of the joints.^ There is no real axis, each segment representing, so to sjicak, a box with an opening each at the proximal and distal sides. The edge of this ojiening is more or less raised and joined to the edge of the opening of the next segment. The diameter of the cavity of the * TierreicU xvii. (If02). j Lepid. Phalatnae iii. p. r.l5 (1901). X Joint and segment should not be confounded, ( Ivi ) Regment containing the ncrvnlnr, niuscnlar, anil vasculiir ap])arat,us is larger than that of the openings, wiiicli serve as doors to the cavity. That is especially the case in the segments of the clnb, as can be seen in tig. 19 of PI. LX., which represents the left half of a segment, divided sagitally, of the clnb of the antenna of I laemorrhagia fucifonnis, viewed somewhat from the distal side to show the distal snrfaee. Tlie difference in the position of the axis is illustrated by figs. 17. IM. 25 of PI. LX. The tino ciliae are never essentially prolonged in Rhopalocera ; they are also short in some families of Heterocera — for instance in Castniidae and Agarintidnr. In such cases the two sexes do not differ, or differ only slightly in the aTitennae. Among the Splnnyidae there is only one genus (^Rhopaloji»i/che) in which there are no prolonged ciliae "in either sex. Sncli instances as these are ([uite an ex(e))tion among Heterocera, the antennae of which show generally marked sexual dimorphism. The specialisation refers to the length and arrangement of the ciliae and to the configuration of the ciliae-bearing surface. The ciliae are seldom nearly all prolonged and irregularly distributed. Of more common occurrence is that specialisation in which some of the ciliae have become long and are arranged in a subapical and a snbbasal vertical row, the two rows becoming fused into one dorso-laterally, forming a kind of flattened half-ring in a lateral aspect of the antenna. These seriated ciliae are apparently always fasciculated, two or more ciliae arising from a common tubercle (PI. LXI. f. 1 — 4), or from a common jiit. The ciliae of the two rows curve towards each other and project laterally, so that they form a kind of jiocket in a ventral aspect (PI. LX. f. 15). The apical row does not extend so far ventrad as the basal row, and the ciliae in and near the ventral mesial line are always shortened in Sphingidae. The ciliae of the apical row are shorter than those of the ventral row : compare PI. LX. f. 17, distal aspect. These fasciculated ciliae as illustrated on PI. LX. are not confined to the Sphingidae ; they are found in Notodontidae, Noctuidae, Arctiidae, etc., etc. However, they occur in all the males of Hawk Moths, excejit Ehopalopsijcke, and are also present in a good many females, though they are here always developed in a lesser degree than in the respective males. The fasciculated prolonged ciliae arc not i)resent on cylindrical antennae. Their apjiearance is always accompanied by a modification of the surface of the segments. The lateral surface encircled by the rows of ciliae becomes flatteneil or impressed, and the segment dilated ventrad and compressed, appearing some- what prismatical in a distal or basal aspect (PI. LX. f. 17). The depth of the lateral groove and the length of the ventral projection are very different, not only in various Sphingidae, but also in other Heterocera (PI. LX. f. 17. 25. 27). The ventral projections of the adjoining segments lie either closely together, or are more or less separate (PI. LX. f. 16. 22. 26. 27). The ventral outline of the segment is in most cases nearly straight, but in many Sphingidae and other Heterocera we find the segment sinuate in a lateral view (PL LX. f. 22. 26) ( Ivii ) to a more or less obvious degree. Such serrate antennae are not rarely fonnd among the Avihnlicinae. The sinnons ventral outline does not occur in clnbbed antennae, in which also the lateral grooves are less deep and less large than in nou-clnbbed antennae. A step further tlian the ordinary fasciculated antenna is that in whicli pectinations make their appearance. As there are no fasciculated antennae among lihopalocera, so there can be no pectinated ones, as the latter type is a derivation from the former. We are accustomed to speak of pectinated antennae, if the segments have lateral expansions or branches. The term comprises, however, morphologically very different elements. Among Sphngidae there occur only two kinds of pectinations, and these alone we shall deal with. We see from tig. K) of PI. LX. that the dorso-lateral part of the segment just above the groove is somewhat dilated laterad ; the expansion is more distinct in a frontal aspect (PI. LX. f 17). If this dorso-lateral expansion becomes more or less jirolonged, we have a strongly or a slightly pectinated antenna of the form represented by figs. 23. 27. 28. 29 of PI. LX. If the scaled dorso-lateral area takes ]iart in the expansion, tlie side-branches are scaled above (PI. LX. f. 2!)) ; if the exjiansion is restricted to tlie non-scaled area, the pectinations are naked above. Comparing figs. 16. 17. 23. 27, it will be obvious that, the branches being lateral expansions of the upper edge of the grooves, the ciliae are naturally confined to the under surface of the pectinations. This type of pectination is (if wide occurrence, being met with in many families of Heterocera, and is easily distinguished from another type, not found in Sphingklae, which has the ciliae all round the side-branches. At the tip of the branch we find generally a prominent sensory bristle (Nov. Zool. iii. t. 4. f. 3. 4), corresponding to the dorso-lateral bristle of setiferous antennae. The pectinations of Sphingid antennae have no such bristle at the ends. Rudimentary (= inciijient) pectinations are rather often observed among Amhulicinae (= Ambuh'cinae + Smerintkiivie of Butler), while distinctly pectinate antennae are rare ; PI. LX. i. 28. 29 represent those in which the branches are longest, one an Ambulicine species, the other an ally of Macroglossum. In the latter species the branches of the inner (anterior) side are somewhat longer than those of the other side. Such asymmetrical development is very common in the antennae. Tlie second type of pectinated Sphingid antenna is illustrated by figs. 1 — 5 of PI. LXI. In figs. 1 and 2 {Polypti/chus mutata) we ser the dorso-lateral expansion ((llj>) well develoj)ed, and observe that its underside bears a number of fascicles of ciliae. These ciliae stand mostly upon tubercles like the more ventral fascicles, and the ventro-lateral tubercles are seriated and are situated upon a slightly raised carina (yc). If these two carinae, one apical and the other basal, become higher and higher, or are produced laterad, what will be the result ? We shall have a segment with two processes on each side, the processes being compressed and bearing like the carina the fasciculated ciliae at the narrow edges, not upon the flat vertical sides (PI. LXI. f. 3. 4. 5). This very peculiar type, which comes close to the type fonnd in most Saturtiiidae ( Iviii ) uml near allies,* but is lint identical with it, dctuirs only in two siiecies of Am/julicinae, each representing a geuus of its own — namely, Monardu and Cresaonia. In figs. 1 and 2 both kinds of Sjihingid pectinations aie incipient. If the dor.so-lateral expansion (dip) Ix'caiue prolonged, the result would he a pectinated antenna of the first type ; if the snbventral carinae (/c) became j)rolonged, we should have a pectinated antenna of the second type. The dorso- lateral expansion is very rudimentary in Monarda and Cresso/iia, and the upper fascicles stand ajiart from it, having in a dorsal aspect (PI. LXI. f. o) the same appearance as in ordinary fiisciculated antennae, this being a si)ecial feature of the Sphingid bipectinated antenna. The asymmetry of the segment is distinct in fig. 4. As said above, the fasciculatiou, ])ectiiiatioii, and the compressed shape of the antennae obtain in a much higher degree in the male than in the female ; very often the female antenna is simjile where the male antenna is complex in structure. Now, the question arises. Does the simple antenna really represent the more ancestral state of development as maintained above, or is the simple cylindrical segment derived from a more complicated segment in consequence of the reduction or loss of the special structures ? Poulton, from researches on the pupae of some Saturniidae, came to the conclusion that the second alternative was correct — namely, that the short-branched Saturniid female antenna was a development by reduction from a longer-branched antenna. Though the con- clusion was perhaps rather hasty, inasmuch as the fact was not taken into account that the specialisations of the male are often transplanted on to the female, it was nevertheless suggestive, and served to draw the attention to a neglected point. If one considers the case of the similarity in the sexes of Rliojtalofjsi/che by itself, one must come to a conclusion similar to that arrived at by Poulton. For the absence of fasciculated ciliae from the male of Rhopalo- pgijche cannot be explained by assuming that this genus had preserved the original simple state of ciliation ; such an explanation seems to us to be almost absurd, coniridering that Rhopalopsijche is in all other respects very specialised, and is the only excejition from the rule among all the Sphingidae. And therefore there remains only the second alternative — that the male antenna of Rlwpalopsijc/ie has lost the fasciculated ciliae, and thus become simple. If that is true — and it cannot be seriously doubted, we think — one might conclude with some degree of justification that the same line of development from the com- plicated to the simplified observed in this male obtained also in the female sex of Sp/khigidae ; that is to say, that the simple female antenna of Sphingidae was a derivation from a fasciculated female antenna. And it might further be advanced, as a confirmation of the evidence upon which that conclusion is based, that there is a wide-spread tendency of retrogressive development in Sphingidae, to which we shall have to draw attention in many places of this Revision. However, when we take into consideration the two types of pectinated antennae found within the same subfamily of Sphingidae {Ceridia, PI. LX. f. 27. 28; • Smith, Ent. Amer. iii. p. 2 (18S7); id., Trans. Amer. Ent. Soo. sv. p. 230 (1888). ( lis ) aucl Cressoina, PI. LXI. 1". 3. 4. b), the (iiiestioii assiuues a ditfereiit asjject. FollowiDg the same line of argument, we slionld luive to attribute to the ancestor of Ceridia an antenna with long pectinations of the first type in both sexes, and to the ancestor of Cressoina one with long pectinations of the second type. Therefore, according to this, the common ancestor of both genera, or the ancestor of the subfamily Amhididnae, would have had an antenna combining both types of pectination — namely, a long subdiirs;il l)ranch and two long subventral branches on each side, a type which does not occur anywhere. If we go further and construe by a similar line of argument the ancestral type of antenna of Sataniiidae and Notodontidae, and then that of the common ancestor of these femilies and the Sfjlniigidac, we arrive at a form of antenna combining all the various types of pectination which are morphogenetically different. In short, if the above line of argument were correct, we should have to attribute to the ancestral antenna of the Lepidoptera all those special features found in the Order which are not derivations from one another. That would be absurd. And yet, if we look over the literature bearing on classification, we often encounter absurdities akin to the above. Going back to the alternative presented above, we hope to have now shown that the pectinated antenna is indeed a derivation from the fasciculated one ; that the pectinated antennae of the males are more advanced than the simj)le or the less strongly pectinated antennae of the females, these coming nearer the ancestral form. As said above, the specialisation of the male antenna is often observed in the female — the groove, the fasciculated ciliae, and the pectination being more or less distinct. We shall call such female antennae andromorphic in tiie body of this Revision. There are female antennae which are much more strongly compressed and deeper-grooved, and liave longer fasciculated ciliae than the male antennae of other species (compare the genus Foh/ptijchns). The sexual differences observed in the antennae of Spliimjidae and other Heterocera are either such of degree, where the male characters reappear in the female, or of kind, where the sjiecial male characters are quite absent from the female ; and the occurrence of such sexual dimorphism is an indication of differences iu function, and not of different capacity in the sexes of acquiring the sjjecialisations. In other organs the females are not rarely more advanced than tlie males ; there is therefore no general rule in this respect. However, it is true of the antennae that the female sex is never in advance of the male. This is cx])lained liy the different role the sexes play in courtship, especially in finding one another, the male antenna being very specialised in cases where this sex lias to search for the sedentary female. The sluggish Ambulicinae have, in accordance with this explanation, on the whole more strongly compressed, grooved, and ciliated antennae, and more often subpectinate ones than the members of the other subfamilies of Spliimjidae. Specialisations of one sex are latent in the other, and may occasionally put in an appearance where they are normally absent. For instance, the metallic Mx ) colours of ninles aro found sometimes in females in sexually very strongly dichromatic species, such as Papilio priamns. The occurrence of andromorphic female antennae is, we think, in many cases due to a similar transmission. In several instances the male characters are the stronger pronounced in the female antennae of Sphingidae the more marked the specialisation is in the male antennae ; while, on the other hand, there are species with almost simple female antennae in sjnte of the highly specialised male antennae. Poh/pti/chus carteri and Ciriisonia juglandin are good illustrations for the one and for the other ty])e. Where the male specialisation is not at all or only in a very slight degree transmitted to the female antenna, there must be some cause at work checking the transmission. The most distal segments, which we have as yet not taken into account, are more or less similar in the sexes. Tlie well-known hook in which ends the antenna of very many Sphingidae^ but not of all, occurs in all subfamilies. The segments are broader and longer ventrally and dorsally (PI. LX. f. 11. 12). The ciliae-bearing surface is thus more exjiosed, and doubtless made more efficient. The special male features are not, or slightly, marked on the most distal segments ; the cone is prominent, and the sensory bristles are often more numerous and longer than on the more proximal segments. The last but one and the previous segments are occasionally couieally jjroduced ventrad (PI. LX. f. 7). Of particular taxonomic value is the eud-segment. The length and shape, and the clothing with scales and bristles of this segment, vary very much and offer good distinguishing characters of genera and even tribes. Figs. 4 — 12 of PI. LX. are illustrations of various kinds of end-segments. The segment is very thin and very long, almost filiform, bristle-like ; or thin and short ; or broad, compressed, elongate- conical in side-view ; or short, broad, and conical. All the end-segments which are produced into a filamentous process like figs. 4. o. 9, or are very thin and cylindrical (figs. 11. 12), we call long ; and those which are not produced into such a process and are conical in a lateral aspect we call short (PI. LX. f 6. 7. y. 10). The scaling is often very sparse, loose, and rough (PI. LX. f. 4. 0) ; ic projects often as a kind of tuft beyond the tip of tiie segment, if this is short (PI. LX. f. 0. 7). The sensory bristles are in most cases irregularly distributed. Many species have two at the extremity. Li Haemorrhagia, Sesia, Macroglossum and genera with similar antennae the slender end-segment has several lung bristles at or near the tip (PL LX. f. 11), the segment resembling that of the subfamily Choerocampinae, in all species of which we find a slender but comparatively short eud-segment (PI. LXl f. 12), which bears six or more long apical and subapical bristles. The same end-segment is met with in Panacra, bnt nowhere else. This brush of hairs reminds one of the end-segment of Aegeriidae and Neotropical Castniidae. The brush of these insects (PI. LX. f 13) differs, however, very essentially in consisting of a great number of long, hair-like, flattened scales.* Incidentally we mention that Pa rant lire III', which is considered an Aegeriid, has no aegeriad antennae, * These scales look too much like bristles in the figwe. (Ixi ) but tiueiform ones, and snrely does not belong wbere it stands in the Catalogues. The only end-segment which would be confounded with that of a Castnia or Aegeria is found in Enijo, where the segment is provided with a brush of long narrow scales which stand all round the segment ajiically. Both the long and slender segment, and the short and broad one, are of equally wide distriliution among the Hawk Moths. The long type, as illustrated by figs. 4. 5. 9 and 11. 12 of PI. LX., does not seem to occur outside the family. It is found in all Sphingicae (except Oligographa), Acherontiicue, a few AiiibKliciiiaf, and many Sesiinae, Philampelinae, and all Choerocampinae. The short type of end-segment has no special Sphingid character by which it could be distinguished from the end-segments of all other Heterocera. The question whether the long or the short end-segment is the more generalised in Sphingidae lias puzzled us a good deal. Judging from the other Heterocera, one is inclined to pronounce the short segment to be the original one. And this is doubtless true as regards the Heterocera as a wliole. But if we consider the development of the antennae and other organs in Sphingidae, we find that the forms with short end-segment in the subfamilies Sesiinae and Pkilampi'linae are doubtless derivations from more generalised forms which have a long end-segment. Compare, for instance, Temnora, Antinephele, Gurelcd ; and Deilephila and Darapsa. Further, we observe in the Ambulicinae that the genera Amph/pterus, Protambuh/x, and Oxi/ambidg.c, wJiich have a long end-segment, are in many respects more generalised than the allied genera Trogolegnum, Orecta, Callambulgx, etc., which have a short end-segment. Therefore we conclude, that the general tendency of reductive development observed in the palpi, legs, tongue, Gic, applies also to the end-segment of the antenna. We attribute accordingly a long thin end-segment to the ancestral Sphingid as a distinctive feature. The eye does not call for many remarks. It is subglobular, its edge ])eing either nearly circular, or regularly rounded above and more straight below and behind. Its anterior edge is less widely apart from the mesial line of the head than the liinder edge, the eye being oblique in position, the head narrowing frontad. This is most evident in Macroglossum, Sesia, Haemorrhagia, and allies (PI. LXII. f. 6). The eye varies much in size ; the largest eye is found in Orgba. "Where tlie mouth-parts are much reduced or obliterated, the eye becomes also reduced. It is never hairy itself, but is often covered above by a kind of eye-brow, and below by a large tuft of hairs, which is especially large in Rhodoprasina, where the tuft covers the lower half of the eye (PL LIX. f. 12). We have not found a vestige of the ocelli. Before entering upon the descrijjtion of the thorax and its appendages, we think it necessary to emphasise what will have become evident to the reader : (1) That there is an obvious tendency of reduction in the head and mouth- parts ; (2) that this tendency is far more ajjparent in tlie Spl/ingidrif without basal iiatch of sensory hairs on the inner surface of the first jtalpal segment than in the others ; and (3) that the reduction of the cranium, eye, and ( Ixii ) iiioiitli-iiaits occur <;oiici'ally tojretlioi-, iiiul are acconipauieil liy the a])j)eai'aiice of niii(i;liiu'ss in tlie scaling of the liead and often by the ai)i)earance of scales on the pilifer and tongue in place of bristles and hairs. The roughness of the scaling is here a sign of weakness of the s[)ecies, suggesting an analogy with the shaggincss of weakly iiidiriduaU of mammals. Little lias liccn published bearing on the comjiarative morjiliology of the tnnik of the tliorax of Le])i(loi)tcra, authors having confined themselves generally to a comparison of the thorax of the Lephhptera Jiiynta with that of the Frenata One of ns has given an account of the mesosternite of the Rliopa- Incera* and tried to tinravel the homology of the various plates composing the sternal i)art of tlie mesothorax. Our researches on tlie thorax of the Jleterorera are not yet rijie for publication. The sterna of the various Heterocevons families are more ('(pially developed than in tlie families of Butterflies ; such sj)ecialisations as we have found there do not occur here. But there are also in the J]eteroc<'ra thoracic characters distinctive of genera, tribes, and families. The most variable part of the thorax within a Heterocerous family is generally the metanotnm. We shall restrict our description to the thorax of the Spkingidac. The denuded protliorax does not apparently offer characters of taxonomic value in tiiis family. The mesonotum (PI. LXII. f. (i, dorsal view ; PI. LXIII. f. ], lateral view), composed of the praescutum, scutum, and' postscutura (= scutellum), is very large, as in most Frenata, occupying by far the larger j)art of the notal region, the pro- and metauotum I eing quite small as compared with the mesonotum. Tlie praescutum [psc) is distinctly triangular in dorsal view, penetrating between the halves of the scutum, or it is transverse. It is vertical iu lateral aspect. The scutum (;«.sr) is widest behind and little longer than broad ; the mesial suture vanishes in front or is here vestigial. The ])Ostscatnm {mscl) varies obviously in size and shajie (Pi. LXII. f. 0. 8), the anterior and posterior angles being more acute in MacroglossKni, Sesia, and some allies tiian in most other SphiiujUhtr. The praescutum of the metanotnm is not visible in the figures. The scutum (mtsc) is divided like that of the mesonotum, but the two halves are widely separate, the postscutnm of the mesonotum projecting between them. The postscutum {mtscl) is always narrow. In Sesia and allies (PI. LXII. f. 6) it is overlapped by the mesothoracical postscutum, which reaches to the first abdominal tcrgite, while it is free in the otlier Spkingulae (PI. LXII. I 8 ; PI. LXIII. f. 1). The ventral jiarts of the meso- and metathorax are never so different in size in any Frenata as are the nota, though also here the mesothorax is the larger of the two. PI. LXII. f. 7 represents the mesosternite in a frontal aspect, the dorsal and ventral mesothoracical endoskeleton (endosc) being visible in the cavity of the thorax. The mesial j)lates of the sternite, the sternum (st), and peristernum {pest) arc not completely separated from one another ; the sternum is elongate-triangular ; thei-e is generally a small membranaceous area * Vcrh. V. Intern. Zooh Congreu p. 81G. t. 1. 2. 3 (1902). ( Lxiii ) at the u]ipi'r end of the mesial suture. Tlie i)eristermiiu (pest) is always snhvertical, as iu other Lepidoptera, witli the exception of the higher Rhopa- locera {Pajjilioniilae, Pierulae, Ni/mphulidae, and some Eri/cinidae), in which it is a transverse belt, and in which the sternum has assumed a pentagonal shape. It is large in Sp/nngidae, and remains broad at the obliquely truncate upper end, where it leans against the parasternum [past). This is a large plate, extending obliquely dorsad and mesiad from the meral sutnre {smu, PL LXIII. f. 1), separating the meral and sternal jiavts of the sternite, to the membrane connecting mcso- and pvotiiorax. The ])late is angnlate behind at its upper corner. It is large in llesperiidae and all Ileterocera, and always small in the Bnttertlies. Between it and the notum the mesothoracic tegula {nitg) is inserted. Below the parasternum we find the episternum {est), with which are fused the hyposternum {hijst) and the marginal stripes along the coxal cavity. The suture between episternum and sternum is distinct, wliile it becomes occasionally obsolete at the upper end near the i)arasternum, where it separates the episternum from the peristernum. The ei)isternnm is always obliquely truncate, with the upper inner angle more or less pointed, owing to the large development of the parasternum. In the Butterflies (except a portio.i of the llesperiidae) it is quite different in shape, the mesosternite of a Butterfly being always distinguishable from that of a Moth by the development of the para- and episternum. In Spliiiigidue and several other families of Ileterocera sternum and episternum are separated, wliile the latter and the hyposternum are merged together. There is another group of families in which sternum and episternum are fused, while the comparatively large hyposternum is free. This distinction is possibly of value in tiie examination of the true {)hylogenetic connection between tlie families of Ileterocera. Tlie division of tlie meral half of tlie sternite into plate.^, some of which are visible also in a frontal view of the breast, will be understood by comparing PI. LXIII. f 1. ;J, which are representations of the breast in a lateral asjiect. The meral snture is more heavily drawn in these sketches than the otlier sutures, in order to make the division into a meral and a sternal part more obvious. Along the meral suture we find two more or less strongly convex plates, which may be termed the iiaramerum {pani), corresi)onding to the parasternum, and the protomerura {prin), corresponding to the sternum. The area beliind these two plates is occupied by the large epimerum (epm), corresponding to the episternum ; from the epimerum is separated by a more or less distinct suture, wliich often vanishes behind, a marginal stripe (wp), situated along the meral cavity (= jiosterior jiart of coxal cavity bearing tiie merum), and divided by a vertical suture into a small fioutal piece and a large posterior stripe. The epimerum becomes membranaceous above. Id this membranaceous area, between fbrewing and episternal plate, we observe a curved stripe of chitin, serving most likely as a kind of spring. The metasternite is more simplified than the mesosternite. The episternum and epimerum are the only large plates ; the other plates are more or less ( hiv ) fused with them, the sternum heiii^r, however, always jjreserved as a narrow separate strijjc. The coxa is inserted in a groove lormed by the sternal part of the sternite ; it bears the trochanter {trocli), and is supported behind by the merum. It will be observed from the figures that the midcoxa is much slenderer than the merum, while tlie hindcoxa is larger than the respective merum. Tliere is a very remarkalile sjiecialisation found in the merum. The hinder edge of tlu' nienuii is normally rounded : but in I'dchylia, Nephele, and other genera the edge becomes carinate and angulate ; in the higher Sesiinae and Pliilam- pelinae. the angle assumes more and more the shajje of a sharp triangular process, reaching the highest point of sjiecialisation in Cephonodes (PI. LXIII. f. 2). The mesosternal merum shows this specialisation first, the metasternal one slowly following. The processes project so far sideways that there is room for tlu' femur between the process and the breast. The femur is generally found in this j)osition. The object of this specialisation is doubtless to prevent the legs and the incision between meso- and metathorax from being a hindrance in the darting flight of the insects. This becomes quite evident when we compare the abdomen, which is closely appressed to the breast in the forms with strongly angulated merum, and liears very strong and flat spines, making the abdomen very smooth. In the upper corner between coxa and merum we find the trochautinus (tc/nn), a small plate which is more or less subglobose. The femora do not offer any characteristic points. It is a curious fact, however, that they remain always simple, even in cases where the tibia has become strongly spinose. Tibia and tarsus undergo several modifications. Their relative length is often variable either specifically or generically. A tibia clothed with scales and some hairs, and truncate at the end, may be called a normal one. The apex of the foretibia is often produced into a strong process (a thorn), which character is sometimes of generic, sometimes only of specific value (PI. LXIV. f. 16). Such a thorn is found here and there in all subfamilies (except Choerocampinae). It is a specialisation often combined with other specialisations. It wonld, however, be erroneous to say that a species possessing the thorn was in every respect more specialised than the allied species which is devoid of that character. For instance, Cephonodes picus has an armed foretibia, while hi/las possesses no thorn, but hi/lns has a decidedly more specialised sexual armature. The spinosity of the tibiae is often an accessory character found in the species which show reduction in the appendices of the head and thorax. But the appearance of the spines is not invariably associated with a reduction of the legs in length, nor are the spines always met with on reduced tibiae. Spinosity of the tibiae occurs commonly among Acherontiinae and Ambulicinae ; it is also found in some genera of Sphim/idae semanophorne. The spines appear first near the apex of the tibia, and the foretibia is the first, the midtibia the second, and the hindtibia the last to acquire spinosity. The hindtibia is frequently much more densely spinose than the midtibia, and the foretibia ( Isv) has the spines mostly along the enter edge, where the_v become sometimes very stout and long {Proi<erpiiu(s ; Arctonotus). The spurs of the tibiae are organs of much importance in the classiticatiou of Moths, their number and length being constantly made use of in systematic works in the definition of genera. The epiphysis or spur of the foretibia,* which we consider homologous with a proximal spur of the hiudtibia, varies in length and jwsition in the Sphiiigidae. It is never absent, as in C/udcosiidae for instance. The most remarkable form is that in which the fringe is obliterated {Ceridia). The midtibia has one pair of slender si>urs, which end in a naked point as a rule and are proximally generally cylindrical. They are sometimes of the same length, but as a rule the outer or anterior one is shorter. A ipiite exceptional development found among the species of the American Choerocampiue genus Xijlophanes leads to the outer spur being the longer one of the two. The variation in length is considerable, the longer spur being sometimes as long as the tibia and frequently barely longer than tiie tibia is broad. There are on the spurs generally some long hairs, which develop often into sj)ines. The spinosity of the spurs is a feature often met with in species witii spinose tibiae, and is, like the latter, a sign of reduction or weakness. In some cases we iind a series of stiff scales along the side towards tlie tarsus, resembling a serrated crest (Mar roff loss tan) ; while there occurs a conspicuous comli of bristles on the shorter spur in Nephele and Centroctena (PI. LXIV. f 8). The comb of Nephcle and Centroctena has most likely the same function as the comb of the tarsus, as described below. The midtibial spurs are never absent from Sjdiinyidae. The hindtibia jiossesses normally two pairs of spurs, one terminal, and the other more jn'oximal, situated sometimes in or near the middle of tiie tibia, sometimes near tlie terminal spurs. The shape and structure of the iiindtibial spurs resemble tliose of the midtiliia ; they are longer on an average, the longer apical Iiindtibial one being never shorter than the longer midtibial one, but very often surpassing it in length. They are less often spinose ; the inner one is never shorter than the outer one of the same pair, and the comb or crest found in JS'epliele, etc., is less strongly developed. The proximal pair disappears very often. The gradation from a tibia with long proximal spurs to one without a trace of them is complete, tiiere being many species with very short spurs, and one {Paium porphyria) in which there are two very short proximal sjjurs or only one or no spur. Here we have a case where there can be no doubt about the direction of the line of develojtment. The absence or reduction of the j)roximal pair of sjiurs is a specialisation, the species thus characterised repre- senting a younger stage of development than those in which the proximal spurs are long. And therefore we can safely conclude that also in the case of the midtibial spurs shortness is a sign of reduction. .Short spurs are seldom found outside the Acherontiinae and AmbuUcinae, and the cases where the proximal pair of the hindtibia is absent are nearly all confined to those two subfamilies, • Kathreiner, llludr. Zcitxckr. Entom. iv. p. 113. Hil. Plate (1899). ( Ixvi ) tlic iiiilv spi'C'it's witli one ]i!ur of liiiiillil>iiil sjnirs not- belonijiiig to tlioso groups lii'iii^'' a iiu'inlier ot" the I'/iilamjjcliiine {MicrOKp/ii/Kc). We afiree with Kolbe in considering the spurs to be modified spines ; they are moriiliologically lioiuologons with hairs, but tliey are spines with a matrix of their own which develops seales, hairs, and s])ines giving the mid- and hindiibial sjuirs tlie same covering as have the tibiae. It sounds curious tiiat we are sjieaking liere of a scaled and hairy hair, but it is nevertheless correct. We refer the reader to anotlier instance where an ordinary spine has developed into a scaled organ similar to tiie spurs. One of the ventral rows of sjiines on the tarsus is in some Si/ntomidae more prominent ; tlie spines are longer and stouter than ordinarily, and are densely scaled ! There are normally four ventral rows of spines on the tarsus of Lepidoptera. The regularity of these rows is very often disturbed by inter- mediate sjiines, or by the disappearance of spines. A remarkable development commonly found in Noctiddae, Arctiidae, At/aristidae, Si/ntomidae, Coci/tia, etc., but not met with in Splihn/idae, is that in which the two inner rows merge together into one, so that there are ajjparently only three rows. In Zygaenidae, Lasiocampidae, some (ieometridae and Notodontidae the ventral spines are very numerous and arranged in more or less regular transverse series. The lateral and dorsal sides of the tarsus are also spinose ; the latter spines of the compressed mid- and hindtibiae, for instance, are very numerous. The anterior tarsus shows different modifications in the si)ines from the raid- and hindtarsi. The external row (or row 1) of the foretarsus contains often some long spines (PI. LXIV. f. 1 — o) ; not rarely these long spines alone are present and are enlarged to curved claws, there being generally present three such spines on the first segment, and one special one on the following (PI. LXIV. f. •")). In most cases the long spines are accompanied by some small ones, generally placed near the base of the segment. Such claws as these are commonly found in Acherondinne, accompanying other modifications (reduction of mouth-jiarts, spinosity of tibiae, etc.) ; they are wholly absent from all other Sphiiujidae e.xcept Clioerocampinae, where they occur occasionally in a lesser degree of development. Another modification obtains in Cocijtins. In this genus of Acherontiinae, especially in the males, the fourth row of the foretarsal segments resembles a short-toothed comb or rake, the spines standing rather close together and being directed ventrad and somewhat curved (PI. LXIV. f. 6). The spines of the other rows have, in some species, disappeared for the greater part (PI. LXIV. f. l(i). The foretarsal spines of Cephonod.es are mostly reduced to hairs. The midtarsal spines have also their specialisation. Here it is the ft)urth row. The basal spines of this are prolonged to long bristles (PJ. LXIV. f. 7), forming the midtarsal comb alluded to very frequently in the body of this Revision. The comb is very strongly developed in Cocytius, Frotoparce, and other Acherontiinae, and gradually disappears (as a comb) in the weaker members of the subfamily, the bristles assuming the normal length. In Andjulicinae it is not met with, not even in forms with such long legs as are ( Ixvii ) possessed by Campsiogene and Protawliidi/x. Here tlie corresponding spines ai'e slightly thinner than the more distal ones, bnt very little longer, the most basal ones being even shorter than tliese. The comb occurs again in the Sphingidae semanojihorue, especially in the lower forms ; bat tlie bristles never attain the lengtli they have in Ac/wro/it/hiar, resembling always more ordinary spines and becoming gradually shorter (PI. LXIV. p. 8). It will be noticed in the figure quoted that the third row of spines does not reach the base of the segment. The hindtarsus is generally longer than the midtarsus, but agrees with it in appearance. The comb, however, is less strongly developed, and is not part of tlie fourth but of the third row of spines, which is very strange, and, like all strange things in morphology, very interesting. The combs of mid- and hindtarsus serve doubtless both the same purpose — we think, of keejjing the abdominal fringe of tlie wing in order, and, in the cJcJ, perhaps also to brush the tuft of the abdominal scent-organ. As the hindtarsus is in a different position to the body and wings, the position of the comb has become altered, or, rather, another row of spines has developed into a comb. As the four ventral rows of spines of the three tarsi are homologous organs, and as the comb of the hindtarsus is homologous to the third row of spines of the midtarsus, we see clearly that the apjiearance of the comb on the hindtarsus is not due to an inner factor of development causing the comb of the midtarsus, developed for some jiurpose, to be repeated on the hindtarsus. It must have been an extraneous factor — i.e. adaptation — which was the cause of the development of the same kind of organ from different sources. The second and fourth rows of spines of the first segment of the hindtarsus, or only the fourth, do not reach the base. The fifth segment of all tarsi bears some stout and pale sensory hairs at the end on each side close to the ai)ical sjiine, forming often a brusli. There are two long bristles dorsally close to the edge, curving ventrad (PL LXIV. f !• — 15). In a few instances the number of bristles is larger in some individuals, there being occasionally as many as four or five ; but this is quite an exception. The pair of bristles is normal not only for the Sphingiilae, but also for a number of other families. In Butterilies and Skippers we find from four to ten bristles ; in Spliinqiduc, Nortiiidae, Notodoididae, La.-iiocampidae, etc., there are two ; while the bristles are absent or vestigial in I'l/ralidae, Zj/gaenidue, Aegeriidae, Hepialidae, Tineidar, etc. The character is of taxonomic value. TInjrididae with two bristles and I'l/ralidae without them can thus readily be distinguished. In Sattirniidae and Uraniidue the two large bristles are generally accompanied by several smaller ones. The claw-segment is one of the most interesting organs of insects, on account of its very numerous modifications. It does not seem to us to have been made use of in systematic works on Ileterocera. The elements of which it is composed are the claw {oti>/chiiim), the false claw (paron'/r/n'/n//), the imd {jM<li:illi(--<); and the empodium. ( Ixviii ) Tlic ciniKKlinin is in Iji'jiidojitcra a sniall tiilierclc aliove tlie pad lietwecu the claws, boariiii; one l)ristle, si'ldoiu two. It is widely distributed among the l)ad-l)eariiig species, Imt is seldom found in Sjjhingidae {Spliingulufi; llopliocnema). The claws are simple in Spfiiiigulae, not divided or toothed ; they are dilated at tlie hase (PI. XLIV.) The pulvillns (pitlr) varies much in size in Lepidoptera, and is often absent, the ])resence or absence freijuently cliaracterising genera and whole subfamilies, but forming not rarely a distinction only of specific value. Absence of the pulvillus is of common occurrence among Acheronfiinae and AmljuUeiimi' ; while the organ is very seldom obliterated in the other subfamilies, Arctonotus and Euproserpinus being tlie only instances. The paronycliium {par) is membranaceous, ventral and lateral in ]iosition, finely ciliated on the surface, and provided with a more or less long fringe at the edges. Its special shape varies much in insects. It is generally deeply sinuate mesially, and each half ' may be again divided (or perhaps ])roduced) into two lobes. The ])aronychinm of Sphingidae possesses in its most generalised state two slender lobes on each side (PI. XLIV. f. 9. 10), of which tlie upper (or lateral) lobe is generally the longer, very rarely the shorter one. The reduction of the paronychium begins with the ventral lobe (PI. LXIV. f. 11. 12) ; this is always the first to disappear, there being no Hawk Moths which possess the ventral lobe and are devoid of the lateral one. The remaining lobe becomes also gradually shortened, and disappears too, the paronychium then being represented by a small ciliated* membrane, which is often less distinct than in f. 14. 15 of PI. LXIV. The reduction and, practically, obliteration of tlie paronychium occur very often in Acherontiiiiae and Ambulicinae, rarely in the other subfamilies. The diflerence in the paronychium is often employed by us in the definition of genera. However, there are genera {IIi/loic>i!<) in which the absence or presence of a paronychial lobe indicates only sjjccific distinctness, and sometimes not even that. Ilgloiciis perelegans consists of a black-backed and a grey-backed form, which are found in the same country, and are most likely seasonal forms. The black-backed form possesses, like most other Hi/loicKS, a long lateral lobe (PI. XLIV. f. 13), which is represented in the grey-backed form by a short triangular flap (PI. LXIV. f. 14. 15). This difference within the same species is of considerable significance. The case teaches us distinctly that the same kind of character is by no means everywhere of the same taxonomic value. No morphological character is a priori of specific or generic value or indicates a higher category than genus. The taxonomic value of a character has to be studied in each group of s])ecies ; a difference which is insignificant here may be very important there. The reduction of the claw-segment — or, rather, of the paronychium and pad, for the claws never disapjiear, nor do they lose the ordinary shape — begins in Sj)hingi(hie with the obliteration of the paronychial lobe. From this second stage lead two paths to the abolition of the pulvillus and paronychium. The ordinary ]>ath is that on which the third stage is represented by a segment without (or with strongly reduced) pulvillus, and with well-developed lateral paronychial lobe ; while the fourth stage is attained when this lobe has also ( Ixix ) disappeared. We meet with this development, which is illustrated on PI. LXIV. by ligs. "J — lo, in Acheroittiiiiae and in Ambulichiae, also in the other subfamilies, but here less often. The intermediate stages are numerous ; but these are generally of little systematic importance, as it is scarcely possible to distinguish diagnostically, for instance, between a long and a moderately long lobe. However, intermediate stages in the development of the claw-segment are often wanting between one species and its nearest ally, or between one genus and the next. The second line of development is represented by the two closely alliel genera Kentrochrysalis and Sj///in(/ultus. The former possesses a complete paronychium and a pulvillus ; the latter gejius has only a pnlvillus. Here the paronychium has disappeared before the pulvillus. The intermediate stage between the two genera is not known ; in this the paronychium would have one lobe. Nor is the fourth stage without pulvillus, tlie one following Sphiinjulas, as yet known. The fact of the pulvillus having been preserved in Sjikinguliis and the paronychium lost is imjiortant. 8uch an exception from the general rule demonstrates that an external cause determined which of the jwssible lines of development tliat single species out of many had to follow. A similar development occurs among the Ambtdicinae. Tlie scaling of the tars'is and tibia is sometimes a distinguishing character of genera. As a rule the mid- and hindtibiae appear strongly compressed, owing to the scaling being longer above and below than on the sides. The scaling between and near the spurs is mostly long and tuft-like, and so is that ventrally at the base of the first mid- and hindtarsal segments. In Pachi/gotiia the proximal segments of tlie hindtarsns appear triangnlai'ly dilated, owing to a higli crest of scales. The mid- and hindtarsi of Macroglog.stim and some other genera are compressed, especially the first segment, which has become strongly asymmetrical, its first inner row of spines being ventral, while the first outer row is almost subdorsal. The singularly meagre success attained in the definition of genera of Sphinyulae by the authors of the old school — meagre even from their point of view — was mainly due to the wings seldom offering in the nenration such obvious distinctions as are found in other families of Lepidoptera. The nenration is indeed surprisingly constant in the main features, apart from some exceptions like Cephonodes with a very short cell to the hindwing, and Daphnusa with SC^ and R' of the hindwing on a long stalk. In consequence of this relative constancy, the nenration of a Sphiugid can scarcely be confounded with that of species of ether families. On the forewing (PI. LXV. f. 1) we find a five-branched subcosta, S(' ; the first branch, SC, arises between middle and end of cell ; SO'' and SC are on a long common stalk branching off before the ujiper angle of the cell ; from this stalk SC'^ turns towards the costal margin not far from the tip of >S(_'\ In very many species, or in many individuals, SCI- is not present at all, it being generally either very weak or absent ; the vein is of no constancy whatever, and consequently of no taxonomic value within the iamily. SC^ ends in the distal margin just below the tip of the wing, and joins here soipetimes ( Ixx ) SC (Miirnylo.s.'onn; Si's/'a ; etc.); SC and SC'' are on a .stalk, and separate at about one-fonrtli tlie way from the cell to the apex of the wing {subcogfal for//). The radial vein, H, has the three branches situated as follows : R^ at the nj)i)er angle of the cell or from the stalk St'^-'', but never standing far from the angle of the cell ; IJ- nearer W than R', but never at the lower angle of the cell, sometimes central. Of the two liranches of the median vein, M, the proximal one, M-, arises mostly before or in the middle of the cell, seldom beyond ; and M' stands proximally of the lower angle of the cell, which it approaches sometimes without reaching it. SM' is absent ; it will be found designated as (SM') in the descriptions of this Revision. SM'' is supported basally by a remnant of SM', the two forming a fork ; .SM'' is rather long, disappearing generally in SM- at a point where the wing is about as broad as SM^ is long. The fork is mostly distinct, but in Macroglossum and some allied genera SM^ is so close to SM'- and is so weak that one can scarcely speak of a fork ; while in other genera {Manimba, for instance) the fork is very obvious. The cross-vein D^ between R' and R" is very oblique ; D^ is mostly somewhat curved ; D* is a direct continuation of M. The costal vein of the hindwing ends costally of that point of the wing which is farthest from the base ; it follows on the whole the curve of the_ costal margin, and is accordingly bent backwards and again forwards in Degmuptera with sinuate and lobed costal edge. The first branch of the snbmedian vein is generally designated as the "bar," connecting SO with C ; it is the • vein forming together with C the small snbbasal cell of Butterflies. This bar or SC branches off in or before the middle of the cell ; in the latter case it is mostly weak and long {Macroglossum, for instance). SC^ and R' are generally on a short stalk, seldom on a long one, or come from a point or are slightly separate. R- varies much in position ; it is more or less central, standing sometimes nearer the upper angle of the cell, sometimes nearer the lower one. R' comes from the lower angle of the cell, and is nearly stalked with M', this vein arising in all genera, except Cephonodcs, from before the angle. Its position is occasionally constant within a genus or a species, and different from that which it occupies in the next. M" stands mostly beyond the middle of the cell, approaching M' occasionally. The first submedian vein is absent as on the forewing ; SM^ and SM^ are present. The cross-vein D^ between R' and R- is transverse or oblique, straight or curved ; D' is mostly obliqne, seldom transverse ; while D' appears as prolongation of M. When D^ is transverse, the lower angle of the cell is W or obtuse {Amphion ; 0 recta ; Rhodoprasina ; Cypa ; Lijcosphingia : etc.). The frenulum and retinaculum are j)resent in all generalised forms. There exist, however, quite a number of genera in which they are reduced, vestigial, or absent. Such genera are found only among the Acherontiinae and Ambulicinae (= Spliingidae asemanophorae), whicli therefore may be called frenulum losers. The frenulum is especially often lost in slow-flying species, or such species as have a tumbling flight like Butterflies, and not the darting flight of the greater portion of the Spkingidae. The wings of the frenulum losers are generally ( Ixxi ) broiul and ol'teu leat'-sbaped. However, the lireadtU of the hiiidwiu^^ or the dilatalion of the costal margin of tlie hindwing i;s not the cause of the loss of the frennlnin. The absence of the frenulum and the weak power of flight are tlie result of the same tendency to retrogressive development, but they are not dependent on one another. There are quite a number of Sphingidae with broad wings and heavy body {Marumba and Lojihostethus, for instance), and such with strongly dilated costal area of tlie liindwing (^StoUdoptera, llj/paedalia), wliich have preserved the frenulum. The reduction of the wing-power, the moditication of the shape of the wings, and the loss of the frenulum and retinaculum occurring together is to be understood in the same sense as the 0('curreuce of such wing-characters as those together with reduced moutli-parte, for instance. The glossy patch of modified scales found on the underside of the forewing near the base before the hinder margin does not extend beyond SM^ and stops generally at SM'. It is elongate and more or less j)ointed distally behind SM^. The scales of the j)atch are broadly rounded at the end, entire (not dentate), sometimes subtruncate, ofteu somewhat asymmetrical, and never elongate- pointed. The non-scaled area which is found, for instance in Notodontida,e, within this jjatch or' proximally of it, and which is covered with fine short hairs, is not met with in Sphingidae. In front of the patch of scales just described there is another, separated from it by hair-scales. This patch is not distinct in all species, but sometimes rather couspicuons — for instance in Herse concohuli and Pholus labruscae. It is situated beiiind M, extending occasionally into the cell, and consists of entire, more or less lanceolate scales, which are ofteu directed obliquely costad. The glossy and sharply defined basal costal patch on the upperside of the hindwiug is composed of scales similar to those of the corresponding patch of the forewing. The generalised forewing of the Sphingidae is elongate-triangular in shape, about 2h times as long as broad, with the a])ex acute, the hinder (or anal) angle distinct, the outer (or distal) margin entire and shorter than the hinder (or inner or abdominal) margin, and the latter slightly convex proximally and slightly concave distally. Departures from this type are very numerous. There occur triangular entire wings which are three times as long as broad, and others (rarely) in which the distal margin is as long as or longer than the hinder margin. The opposite development results in a short and broad " bombycine " wing similar to that of broad-winged Notodonts or Lasiocampids, with obtuse apex, a form which is found in a number of Acheront.iinae {Lapnra, Hopliocnema, etc.) and Ambulicinae (several Polyptychus, PI. I. f. It). 11), which have lost the typically Sphingid aspect. The apex is sometimes jiroduced into a hook {Lijcospltingia), and the hinder angle is often very obtuse, in one case completely rounded {Pln/lloxiplna, PI. I. f. 4), and produced backwards in the rather numerous species with more deeply sinuate liinder margin (PI. I. f. 5. 6. 9). The distal margin is entire, concave, straight, or convex, bisinuate, dentate, scalloped, or strongly lobed, The irregular distal margin does not ( l.xxii ) iifciir ill AclirrdiitliiKir, ami is rarely iiit't with in ('/i(),'nic(iinjiiiini: { l'//iai()j//la /ii/s(riu) ; among the other subfamilies this form uf wing is often found. The lobe at R' is generally the most ])rominent ; but this rule does not hold good in every case— in P.seudein/o, for instaiu'c, there is no distinct lobe at K'', bnt a prominent lobe at R-. The hindwing is scarcely less variable witliin tlie family than the forewing. The apical angle, the point farthest from the base, lies at .SM- or (rarely) at R' ; the anal angle at SM-. The distal margin is less irregular than that of the forewing in the lobate sjjecies — as a rule convex, sinuate before anal angle or not; the apex is very often completely rounded (PI. I. f. •). 0. 8. 10 11. 12). The costal margin, ordinarily straight proximally and slightly convex distally, undergoes very conspicuous modifications. It is dilated into a proximal lobe {Macroglossum aquila ; Giganteopaljjiis : Air///v//) ; or the costal area before vein C is widened and, when the insect is at rest, overlaps the forewing, recurving and enveloping the costal margin of the latter {StolUIopfera and Ili/paedalin) ; or the costal margin is dilated distally, and either sinuate proximally of the lobe or not {Df(/iHaj)tera) ; or it is dilated proximally and distally and sinuate between the lobes {Gxrelca). Though the opinion to which Butler gave expression in 1875, that an obvious difl'erence in the shape of the wing is a generic difference in this family, is not correct in its entirety, the wings of close allies and of the sexes being sometimes very different (PI. I. f 1, c? ; 2, ?), a special wing-form is, however, in many cases accompanied by other distinguishing characters, and restricted with these to an assemblage of allied species constituting a genus. In such cases the shape of the wing is a very convenient character to recognise a genus by, or to be used in a key. Frotamhuhjx (PI. I. f. 3) and Ox)/ambulyx (PI. I. f. 2) illustrate this point, all the species of the former genus and its close ally Amply jder us, which two genera are purely American, having the apex of the forewing truncate-sinuate, while all the species of the purely Oriental genus (>j:</ambiibjx have the apex acute. Ox)/amhalyx and Protambulyx can, therefore, easily be distinguished by the shape of the wing. However, this distinction alone would not justify us in treating the insects referred to as generically different ; the justification lies in the whole organisation of the species proving them to belong to two sharply defined groups, each containing material which is inter se more closely related tlian with the members of the other group. The abdomen of Lepidoptera is composed of ten segments. The ninth and tenth of the S and the eighth to tenth of the ? are modified, so that the last one of ordinary appearance — the last of descriptive Lepidopterology — is the eighth in the ,$ and the seventh in the ?. The first segment is short and generally overlooked in ordinary descriptive work ; it has no free sternite, and the tergite lies often close against the thorax (PI. LXIII. I 1. 2). The base of the abdomen of a number of families of Moths exhibits very distinctive features, here being found organs of various kinds, most likely all ( Ixxiii ) of a sensory character, the metaiiotum occasionally taking part in the peculiar structures. The SphingUlae have nothing of the kind, but possess other sjiecialisations in the abdomen, of which an armature consisting of spines is the most notable one. The sj)iues stand at tiie edges of the segments, and are found on segments 2 to 8 in the c? and 2 to 7 in the ? . The dorsal s])ines are stronger than the ventral ones ; the latter are often very weak, while the former are strong ; or the ventral ones are only vestigial or are absent, while the dorsal ones are numerous. The longest spines are on the seventh tergite. We tind three main types, illustrated by PI. LXII. f. 0. li). 11. The most frequent type is represented by f 1(». Here we have several rows of elongate, flattened spines, those of the proximal row being (as is always the case) the shortest, and those of the last row the longest, so that the upper ones cover the ne."ct only partly. The spines of this type vary in the different species (resp. genera) from being -very strong (P/iri/xus lineata ; Atemnorci) to being very weak, disappearing in some Ambulicinae altogether, or rather developing all into scales {Li/cosphingia ; Crei^soiria ; Langio'). Where the spines are weak, the skeleton of the segments is also weak. The reduction in the degree of chitinisation of the spines is generally accompanied by a decrease in size, and often by an increase in number. In some instances weak spines are distributed all over the tergites, this being due to the scales of the under layer having become spiniform {Ambulicinae ; Marumba ; Clanis ; etc.). Where the spines are longer, more conical, and more strongly chitinised, their number is generally smaller (Goci/tias, for instance). This variable type is found in all subfamilies, the strongest reduction of the spines occurring in Achevontiinae and Ambidicinae. In the second type, representetl by PI. LXII. f. 11 the spines are similar to those of the strongly chitinised form of the first type, except in the sjjines of the first row being short and rounded. The spines are all flat, black and glossy, being very strongly chitinised. The type is very constant in itself, and there are no intergradations between it and the preceding, except on the proximal sternites. This kind of spination makes the abdomen smooth and sli})])ery, the spines lying very closely upon the following segment. The spines of the sternites are the same as those of the tergites, only the proximal ones of the proximal sternites being a little longer. The type is confined to Se»ia and allies and Maeroglossurd and allies — i.e. to the most specialised genera of Se.siinae and I'hilampelinae. The basal sternite (= sternite of second segment) is spinose in the species with this type of spination (PI. LXIII. f. 2). In the third type, which is connected by intergradations witli the first, there is only one series of spines (PI. LXII. f U), which stand often well apart, and are long, conical, and very strong. Tiie form of this type without any small spines occurs only on the last tergites ; on tiie more proximal tergites and on the posterior sternites smaller s])ines ajjpear between the long ones, while on the basal segments the spines are all sliort and fiat, the uniserial ( Ixxiv ) arraiijii'iiiciil, liowi'vor, beiiij,' jirescrvcd. Tlic miiscrial spiMatioii is I'duiiil only in l'.te}i(losjthinx and allies, Fachjilin, and allies, a number of Xyloph'Dic^, and in Vhohia — /.('. in generalised forms of the S/thiiiyidae semanophorae. The second type of spination has arisen from the first ; there can be no doubt on this point, if one compares Atemnora and MacrogloMum, and the members of the Sesia series of genera ; it represents the highest specialisation in one direction. Which, however, is the original spination of S/ihiiHiidae ? Does the weak spination of the higher AmhiiliciiKie and AcJieronttinne represent the generalised state, or is a stronger chitinised t} pe more ancestral ? There is a complete gradation from Cimso/iia, Laiigia, etc., without spination, through the reduced SpluiKjuhie with weak spines, to the strong uniserial spination of I'liolits, l'sei(clo»jiliiiix, etc. Sjtination is certainly a specialisation. It is restricted to the SphiiKjidae, the only a]>proach to spination we know of being found in the American Castniidae. Therefore one cannot seriously doubt that the ancestral Heteroceron from which the family Spldngidae has arisen was a Lepidopteron without abdominal spination. But was the abdomen of the early Sphingid also spineless ? or was the acquisition of spination among the first specialisations by which the early Sphingid deviated from its allies? Where there is a series of ([nantitative gradations from a to z, the student is easily inclined to consider a the beginning and z the end of the develop- ment. The presence of such a series is, however, no proof whatever that the quantitatively lowest degree is the j)ylogenetically oldest stage. The mistake has been made, and will often be repeated no doubt. It is a very natural and very convenient conclusion. If this view were correct in our case, the Hawk Moths most generalised in spination would be the few species which have no spines {Cregsoniu juglandis, Langia zenzeroides and Lijcosphingia hamatus), and next would come a great mass of genera with weak spines. All these weakly spined Hawk Moths are weak also in other respects ; they are reduced and otherwise modified forms, showing reduction and modification in the mouth-parts, scaling, legs, wings, and sexual armature ; they are decidedly developments from less reduced tyjjes. It is true, a sj)ecialiseu insect may have preserved one or more generalised characters ; but it would be very singular, — and is surely improbable— if snch a number of species of various subfamilies had all preserved the same ancestral spination, and become in other respects so diversely modified. On the contrary, one is bound to conclude that the probability is altogether in favour of the weak spination being also a character accpiired — i.e. that the weak sjiination is the result of the reduction of a stronger spination, as the short tongue, the short autennal end-segment, the small palpus, the short spurs, etc., etc., are the results of reduction. The first abdominal segment {(W and )>p, V\. LXII. f. 6. 8 ; PI. LXIII. f. 1) consists of a tergite {at^) and a more or less triangular lateral plate, the paraplenra (/i}>). The tergite varies much in length, and offers distinguishing characters, being, for instance, reduced to a very narrow strij)e in Macroglossum and allies, while it is about half the length of the second tergite in Haemorrhagia, ( Ixxv ) Sesia, etc. There is no trace of real spines on the first tergite, or on the parai)leurae. The latter are always simple, flat, not raised in front to a flap ; in short, there is nothing recalling the special structures of some other Heterocerons families. In front of the parapleura we iiud the first abdominal stigma (sfi) : it lies free in the membrane. The stigma of the species in which the abdomen is pressed closely against the breast is not visible without separating the abdomen from the metanotum. The second to sixth tergites are essentially of the same strnctnre, the spines of the posterior ones becoming stronger, however, as said above. The seventh tergite is longer, with the sides more strongly converging anad in most species ; it has a quadrangular shape in Sesia, ^[acro- glossum, etc., or becomes more or less conical or elongate-trapeziform (most genera). The eighth tergite is small, and jiartly (c?) or completely (?) concealed by the seventh. This segment will be described below. The parapleurae of segments 2 to 8 are membranaceous and bear the stigmata. The second stigma, however, is situated upon the tergite (PI. LXIII. f. 1), and the third one half upon the tergite and half upon the parapleura. The eighth segment and following have no stigmata. The sternite of the first segment is absent, or perhaps vestigial and merged together with that of the second. In the special part of this Revision we understand under second, third, etc., sternite, the sternite of the second, third, etc., segment. In Sphingidae, as well as other families, the structure and shape of the sternites of the first and last segments require special attention, as they undergo sometimes remarkable modifications. The second (= basal) sternite of Sphingidae touches the merum of the hindcoxa, with which it is connected by a short membrane. In by far the larger unmber of species it is slanting (PI. LXIII. f. 1), transversely impressed in front, the impression ending at each side in a small but often deep groove, and is mesially carinate in front, the carina fitting in between the coxae. The sternite of Macroglossum differs from the normal type in the narrower frontal part being vertical, and the main portion of the plate on a level with the following segments and with the posterior trochanters ; the low mesial carina is found on the vertical portion of the segment, and the anterior edge of the horizontal portion is smooth and evenly curved. Haemorrkagia, Cephonodes, and a few allied genera have the frontal part of the segment also vertical, and so incurved that the edge of the horizontal part projects forward. This edge is not simple as in Macroglossum, but is mesially produced into a conical obtuse tubercle, which fits in between the posterior pair of trochanters (PI. LXIII. f 2 ; in this figure the interspace between abdomen and coxa is purposely drawn too wide, in order to show the process of the sternite). The sternite of the seventh segment appears in the female sex of Spliingidae in two principal types. Tiie ordinary type is that illustrated by PI. XIX. f. 11 and PI. LXIII. f. 3 and 4. Here the apical portion of the sternite is more or less broadly membranaceous ; the stronger chitinised plate is short, broadly rounded, or sometimes elongate-trapeziform, with the apex faintly sinuate. The ,( Ixxvi ) jilatf i^ nut .so sliiiriily liiuited iis it iii)iK'ars to be in the figures. It doc-s not bear any spines. Tile second type is represented liy I'l. LXIII. 1'. ."> and G. The membrane connecting tlie steriiite (vii. v) with the tcrgite (vii. t) is very small; the sternite is not membranaceous apically, tlie strongly chitinised jdate extending right to the apex. The plate varies in form and size, occupying sometimes tlie whole ventral side of the segment, the tergite being scarcely visible in a ventral as))ect of the abdomen (f. 6); or it is smaller, trai)eziform, the tergite encroiichiug ujiou the ventral side (f 5). The apex of the sternite is either totally spinose {Sesia, Cej>hoiioiles, etc.), or at least spinose laterally {Pacl/)/gonia). This tyj)e is a derivation from the first. The seventh sternite of the male of S/)htniii(h.ie is normal, being similar to that of the sixth. The eighth sternite, however, is modified ; it is always without spines, deeply sinuate as a rule, and occasionally incrassate mesially or produced into a j)rocess (PI. XXII. f. 18 — !29). In the Sesiinaf and Philampelinae with a stumpy abdomen (Sesia, Fachygonia, Macroglossum, etc.) it is very short. The eighth tergite of the male is spinose. It varies in size and shape like the seventh. It is long, and appears conical in dorsal asj/ect in Acherontiinae, for instance, and short and flat in Macroglosstim, Sesia, and allies. The abdomen taken as a whole is mostly elongate-conical. Deviations from this form, which represents the generalised type, are numerous. AVe mention that the abdomen is ovate in some Uaemovvhaiiia, flat and broad in Jlypaedalia, broad and stumpy in Pachi/gonia ; that it is horizontal in most species, and curved upwards in many Amhuliciiuie and some Acherontiinae. The scaling of the posterior segments exhibits sometimes striking features. The scales at the ventral apical angles are occasionally prolonged to tufts (many Scsiinae and Pliilampelinae : Oxyamhulijx ; Poliana ; etc.). The long scales of the seventh and eighth segments form tufts especially liable to modifications. In Pacliyyonia the abdomen is triangularly truncate witli a short lateral tuft; this tuft is prolonged in Himantoides, the " tail " appearing divided. The expansible fan-tail found in numerous Sesiinae and Nepkelicae is generally tripartite ; it occurs also in Cypa and allies, of the subfamily Amhulicinae. The modification in the skeleton of the fan-tail segments consists in tiie plates being short and amjily movable. There is no additional division of the segments besides tlie separation into a tergal and a sternal plate. We have not studied the muscular apparatus. The modifications of the scaling of the abdomen are very instructive. The upper scales are generally elongate, more or less hair-like in. appearance, and the underscales large, mnltidentate. Among the Sphingidae asemanophorae we find a good number of forms which have instead of the broad underscales long hair-like ones, which are often modified info lanceolate spines resembling the weak spines at the apices of the segments in the respective species. Kellogg drew the conclusion, from his studies of the scales of Lepidoptera, that the hair-like scales represented an older type than the broad mnltidentate scales. ( Ixxvii ) We liav'e arrived at the 0{)j)0site opinion. The shagginess of the scaling of Hacmorr/iagia, many Amhidicinae, and Achrrontiiiuie is a specialisation derived from the more smootli scaling preserved l)y the majority of Sphingidae. We find shagginess only in also otherwise mucii specialised forms. Roughness of the scaling and development of broad scales into long ones accompany nearly always the retrogressive development of other organs. The copulatory apparatus of the male is composed of the ninth and tenth segments, in some cases also the eightii segment taking part in the formation of the apparatus. The accompanying diagram (Fig. 3) will make clear the relative position of the various elements of the clasping-organs. The ninth segment is a strongly chitiuised girdle, broadest above, and here sinuate basally (PI. XLII. f 2. 3. in.t). This belt is ventro-laterally dilated into a large flap ((Jl), the clasper or valve, which corresponds to the pleura and sternite of the segment and bears the harpe (//), the stei'uite being divided mesially, as is mostly the case with the ninth sternite of insects. The pleura is attached to a proximal stripe of chitin (Fig. 3, jjI) and to the sternite. The tenth segment (Fig. 3, X.t and X.r) stands in very close connection with the ninth ; there is no intersegmental membrane between them, except occasionally a remnant on the npperside. The tentii tergite is strongly chitinised like the ninth segment, while the sternite is occasionally weak. The tenth tergite is movable in a vertical direction, or, if comj)lctely divided, also mesiad. Between the sternite and tergite we find the anus {A), and between the tenth sternite and the ninth the jienis-funnel {I'-F), from which i)rotrudes the penis-sheath {P). None of these organs are constant within the whole family of Sp/iingidae. The variety in tlieir structure is enormous. But in spite of the seeming eccentricity in the development, the lines along which modification took place ( Ixxviii ) are liuic suid tliore plainly visil)lo, and thorc is not rarely a consjiicnons hoiuoirencity preserved in more than one gronj) of species or of genera, and in more than one particnlar structure, as we shiill see by reviewing shortly the mass of specialisations described in the systematic part of the Revision. The tenth tergite bears stiff hairs, wliich stand either singly or form a more or less dense covering on the njiper and lateral surface. The hairs are generally directed frontad, not anad, and are occasionally so close together that the ])roximal outline of the tergite cannot be ascertained without removing the covering of hairs. There are two principal forms of the tergite, it being either diviiled mesially or simple. The symmetrically divided tergite represents the generalised state of the segment, "^f tlie two halves are divided down to the junction with the ninth segment, they are a little movable against each otlier ; but as a rule tlie two processes are not separated proximally, and therefore move together and only vertically (PI. XXVI. f. 2S. 33; PI. XLII. f. 1— 2;")). We find the divided tergite preserved in some generalised Ackerontiinae, some Ambuliriimc, and the Sesiinae. It aj)pears in many modifications. From the divided tergite the undivided one has developed in two ways : — (1) One side of the tergite becomes obliterated, a type whicli is found only among the Sesiitiae, and of which we shall have to speak later on. (2) The tergite becomes narrower, the processes shorter, and the mesial ])ortion correspondingly longer, resulting in a narrow sinuate tergite and then in a non-sinuate one. This is the ordinary type of a simple tenth tergite. Here the single, symmetrical process is generally more or less compressed, convex above, the apex curved downwards to a hook and mostly pointed. The variety in this type is very great, as a look at tlie figures will show (Pis. XXV., XXVI., XXVIL, XLIV., etc.). The tenth sternite is a belt running from tlie base of tlie tergite ventrad, encircling a membranaceous area, from wiiicii projects the anal cone, the end of the gut (A). The ventral transver.se part of the sternite is in by far the larger proportion of Hawk Moths as strongly chitinised as the vertical side-parts, and j)roduced into one or two processes or lobes of various shapes and sizes. In some cases the transverse portion is very feebly chitinised and does not form a distinctly separate plate (PI. XXIV. f. 10; PI. XXV. f. 12. 27) ; in other species it is a very low ridge without lobe or process (PI. XXV. f. 1. 24 ; PI. XXVI. f. 1). As a rule the sternite is shorter than tlie tergite. However, in some Sesiinae (PI. XLII. f. 11. 25; PI. XLIII. f. 1. 4. 0) the reverse is the case. The sternite is simple or divided ; it may agree in this respect with the tergite (PI. XLIII. f. 1), or it may be divided when the tergite is simple (PI. XXV. f. 20 ; PI. XXVI. f. 30), or simple when the tergite is divided (PI. XXVI. f. 1 — 14). It is never hairy like the tergite, but the apex is often granulose, the granules bearing fine Itristles, most likely of a sensory nature (PI. XXIII. f. 1. 2. 22). At the inner side of the lateral part of the sternite we find, especially often in Ambulicinae, the membrane swollen to a hump beset with sensory hairs (PI. XXVI. f. .O. 9), An asymmetrical develojiment of the sternite will be referred to below. ( Ixxix ) The clasper of Spliingidae is normally sole-shaped (PI. XXXVIII. f. 5), with the dorsal and ventral margins ronnded. The size is variable. The jirincipal modifications are as follows : — (1) Modifications by rednction. AVhen only shortened, the clasper assnmes the stnmpy form as illnstrated by Pi. XXXII. f. 18. 10. When only narrowed, we have the narrow clasper of CaUamhuh/x (PI. XXXIV. f. 19. 20), a common type among Ambulicinae. The longest and jiroportionally narrowest clasper is fonnd in Hinvintoides and Sf.iia (PI. XLVII. f 1. 15). The broad clasper tonches the tenth tcrgite, while the narrow clasper stands apart from it, if the rednction has taken place in the dorsal portion of the clasper. The interspace between clasjier and tenth tergite is then covered by the long fringe of hairs of the former. (~) Modification by division and the develoi)ment of a special armatnre (apart from harjie). Lobes, processes, teeth appear very often in connection with the rednction of the clasper. Compare PI. XXXII. f. 12. 13 ; PI. XXXIII. f. 6. K). 21. 22. 23 ; PI. XXXIV. f. 3. 2s ; PI. LI. f. 22 ; PI. LII. f. 1. 6. The clasper is not solid, bnt consists of an external and an internal i)late, between which the remnant of the matrix is found. It is an outgrowth from the segment, and can best be likened to the wings. The inner sheath is more or less covered with hairs, which are often very dense, very stiff and long. The hairs of the distal half of the clasper are directed dorso-basad (PI. XXXVI. f. 9 ; PI. XXXVII. f. 15). They are not generally drawn in the figures. The inner sheath is in most species raised into special armatures. There is very often a consjiicnons subdorsal longitudinal setiferons fold (PI. XXXVI. f. 18. 19. 2n). A dorso-basal ridge or process is commonly found among Ambulicinae; it is seldom spiniforra {pdh, PI. XXXIII. f. 21 ; PI. XXXVIII. f. 7), mostly almost membranaceous, compressed and beset at the edges with setiferons tubercles {pd, PI. XXXII. f. H — 11. 15. 17. 18). The ventral armatnre of the clasper is termed the harjie : it begins at the ventral basal (hollow) incrassation of the clasper, which is the principal part of the sternite proper, and extends more or less distad and dorsad. If the harpe is reduced, the line of sejiaration between the triangular sternite and the main part of the clasper (= pleura) is plainly visible (PI. XLV. f. 26 ; PI. XXXL f. 2 ; PI. XXXV. f. 20), but as a rule the ridges or processes of which the armatnre consists are continued over the pleura, the inner sheath of which takes part in building up the structure. The harpe is so diversified tliat it is impossible to give a general description of this armature. It is on the whole more complicated in the Sjjhingidae usemaiiojtkorae than in the Spkingidae semanopkorae, but we find also among the latter sometimes species with formidable-looking harpes (PI. XLVIII. f- n — 27). The harpe of Choerocampinae is as a rule a simple, slightly curved, slender process (PI. LII. f. 8) ; an armature of the same type is not rarely met with among Philampelinae (PI. L. f. 32). As a rule the processes of the harpe lie upon the inner surface of the clasjier ; bnt occasionally the sternal part alone forms the armature, and is situated below the main part of the ( Ixxx ) olasj.er (I'l. XXX II I. f. I. 14 ; I'l. XXXIV. f. G). In liy far the greater mimber of species tlie liarj)e is siiorter tiiaii tlie clasper. A dense patch of scales is found on tlie harj'c of nearly all Protoparce and in many other Acherontiinae. The right and left clasper and harpe arc generally the same in ontlinc and structnre. However, there are exceptions, asymmetrical development occurring also in these organs, as will be detailed below. The external jilate of the clasper is divided by a longitudinal rib-like incrassation into a narrow dorsal and a large ventral portion. The dorsal part is generally concave above the rib. Ui>on and near the rib there is a peculiar modification of the scaling found in a great number of Splnngidae. We consider it an organ of friction, by means of which a sound of some kind is produced. ^Ve do not know whether a similar organ occurs anywhere else among Lepidoptera. There are two types of tliis organ, the one contined to the Spliiiu/idae ast'ma)iojthorae, the other to the Spldnyidae semanophorae. In the former group of Splnngidae it is a patch of scales lying more or less flat upon the clasper (PI. LVIII. f. 29—32). The patch as well as the scales vary in the different species (respectively genera) in shape and size. The patch is as a rule restricted — being, for instance, narrowly halfmoon-shaped in Protamhulyx strigilis (PI. LVIII. f. 3(1), triangular in Protambuh/x euri/alus (PI. LVIII. f. 32). But in some cases it is very large^Ampli/pterus (PI. LVIII. f. 29). The scales of the patch are rounded and entire, or elongate and dentate (PI. LVIII. f 35) ; the middle ones are half erect in Poliana bachholzi and micru. A peculiar modification obtains in Polyptijclnis trisecta and P. ortho- graplms (pp. 243. 244). Here the scales are obliterated for the greater part, the dorsal area of the clasper being naked and rough with some grannies. The friction-organ of the Splnngidae semanophorue consists of lanceolate scales which are half erect. The scales are very numerous and small (PI. LVIII. f. 3G), or are reduced in number and have increased in size (PI. LVIII. f. 37), the smallest number, namely one, being found in Ilippotion rebeli and isis. The scales are strongly striped, have generally a prominent midrib, and are more or less obviously asymmetrical (PI. LVIII. f. 38) ; they are more often truncate than pointed. In Ort/ba kudeni (PI. XLVI. f. 4) the organ is entirely diftVrent, consisting of a large mass of long, wire-like, brittle hairs. Among the Ambulicinae with friction-patch on the clasper we find a number of species which possess a corresponding organ of scales on the inner side of the eighth tergite. This organ is almost confined to the Ambulicinae, being found elsewhere only in some Acherontiinae, not in any Splnngidae semanophorae. The organ consists of one single row of large scales, or of several rows. The scales are ribbed, and are so closely applied to each other, or are even soldered together, that they form a kind of ribbon at the apical and lateral. edge of the tergite, the flat ends of the ribbon being directed towards the clasper. This ribbon is either continuous from one side to the other (PI. LIX. f. 1), or it is interrapted apically (PI. LVIII. f. 39. 40 ; PI. LIX. f. 2). The scales of the ( Ixxxi ) ribbon are homologous to the nmlermost scales of the teryite, which stand in other species at the ver}' edge of tlie segment, and are also here often turned internad. The organ is as a rule visible witiiont dissection, bnt has as yet escaped observation, as lias also the friction-patch of the clasper. There can be no doubt that tiie ribbon of the eighth tergite and the patch on the clasper form together one (stridulatin":) ajiparatns. Is the friction-patch a now acfjnisition, or have tlio six'cies which are withont it lost this organ ? In the Spliinyidae asi'm'iiwpliorae the organ is found in the more generalised forms and never in the strongly specialised (reduced) species. In the Sphingidae semanophorae the organ is absent from many of the most specialised genera ; it consists of a multitude of small scales ill several generalised genera (P/iolus, for instance), and its development cul- minates in some Ilippotion with one large scale. From these facts we conclude that the ancestral Sphingid possessed a friction-patch wliich consisted of numerons small scales not very different from the ordinary scaling of the clasper, a patch perhaps similar to that of Pl/olus and Pseuclosphinx (PI. LVIII. f 30). From tliis indifferent patch the two modifications arose, — in the one type the scales remaining numerons and becoming closely packed together, as in Piiilogromma, Meganoton, Protambuli/.r, etc. ; in the second type the central scales becoming enlarged and seriated, and assuming a half-erect position, with the broad sides turned more or less dorsad and ventrad, as in Ncphele, Deile- pJdla, Tlieretra, etc. In the Sphingidae sfinanophorne, where retrogressive development is comparatively rare, the friction-scales are found in the greater number of species. They are lost in the reduced or otherwise strongly specialised forms like Sj/hingonaepiopsis, Proserpinus, etc. ; tiicy are also absent from modified claspers like those of Ilaemorykagia, Perigonia, etc. Among the HpJdngidae asemariophorae the friction-patch is comparatively less frequent. Of Arherontiinae only Old World genera have preserved it, not one of the numerous American species possessing the organ, while a large proportion of Old and New World Ambulicinae. are provided witii it. The area between the two claspers and the tenth sternite is more or less membranaceous. There is a central hole, of which the edges are more or less raised and chitinised, forming what we have termed* a penis-fnnnel (Fig. 3, P-F). This penis-funnel {P-F, PI. XXIX. f. 38. 39. 47 ; PI. XXX. f. 41 ; PI. LI. f. IT — 25) is vestigial or distinct; it has often a special shape, and has an armature of its own which is of great help in the discrimination of species in more than one case (see Ni/n'ri/x, p. 414, PI. LIII. f. 44 — 48 ; compare also Pi. XXX. f. 33—38; PI. LIII. f. 13. 14). Above the penis-funnel there is in a few cases a further armature consisting of a pair of processes, one on each side («, PI. XX^'. f L'). From the penis-fnuKsl projects the penis-sheath, of which the former is a snjiport. The penis-sheath is provided at the end, or near it, with processes and teeth of various sliapes and sizes, this armature being as diversified as that * Xov. Zool. V. p. 561 (1S08). / ( Ixxxii II, <>t of tin- claspi'i'. Tlif ]ir()ocssfs are cidier tiniily (■oiiiu'ctcd with tlu' slioatli which they ai"e an uutgrawth or a proloiijiatioii (PI. XXX. f. 'J^!. 4s), or they arc more or less movable against the sheath, the base of tlie proce.ss at the juncture with the sheath being partly membranaceous, forming a kind of joint (PI. XXX. f. 31 ; PI. LIV. f. 1—12). The long and slender whijj-like process found ill I'tTiijonia, Si:\/k, and some allies bears two sensory bristles at tlie end. We can distinguish two kinds of jirocesscs : (1) the process is a prolongation of the apical edge of tlie slieath, and is accordingly flat or concave on tlie innerside (PI. XXIX. f. 9. 12); or (2) the jirocess is an ontgrowth from the surl'ace of the sbeath, an enlarged tooth, and accordingly more or less conical and hollow, at least at the base (PI. XXIX. f 47 ; PI. XXX. f. 38). The processes are short or long, single or double, simple or dentate. The most remarkable is that of E. sagra (PI. LIV. f. 13). The teeth on the processes and on the sheath are small or large, single or in patches, irregularly distributed or seriated, forming serrate ridges. They are in most cases solidly connected with the sheath, not bre.iking oft" easily. There is, however, a second kind of tooth, which is easily detached from the sheath. This tooth, which has a central and three or four lateral branches, is found in 'rherelra latreillei and allies (PI. LVIII. f. 4—7). The spine-like teeth of Xylophams nechus (PI. LVII. f« 18. 19) also break oif easily. The same obtains in centroctena rather ford i. Within the penis-sheath we find the membranaceous penis proper, the duct of the sperma. This duct can be pushed out, and enters the vaginal orifice with the help of the accessory aj)pliances. The duct has in most cases an armature of its own. It is either partly beset with small teeth (PI. XXIX. f. 34. 41. 42 ; PI. XXX. f. 12), or the teeth are enlarged to stilettos (PI. XXIX. f 40 ; PI. XXX. f. 44). In other cases we find two or three slightly more strongly chitinised denticulate processes of various lengths and shapes (PI. XXIX. f. 7. 8. 9. 21 ; PI. XXX. f. 16—19). Or the love-daggers are strongly chitinised, acute or club-shaped (PI. LV. f. 44 ; PL LVI. f. 41). The armature of the penis-sheath and penis serves a double purpose. The firm processes and teeth of the penis-sheath are grasping-organs like the harpe. But the processes are often far too long for that purpose alone. Now, the fact that the peculiar dentition of Therctra latreillei and allies is easily detached, and the detached teeth are found in the vaginal cavity of the females, makes it evident that the armature of the jienis and penis-sheath is a means to stimulate the female, analogous to the apparatus found, for instance, in some Rodents. The se.xual apparatus is of great taxonomic value, and we have taken special care to dissect as many species as we could. In a number of cases the apparatus is the only safe guide in the recognition of species. From this point of view the Sphingidae can be divided according to the sexual armature into the following categories : — (1) Species which are not diiferent from their nearest relatives in these organs. — Nephelc is an example of this kind. Some Ghoerocampinae allied to gallii ( Ixxxiii ) and tyijihorhiiii' ulsu do not .show any apparent ditt'erenccs. Protoparce jlorestan and Ch laciio(/ nrmma jasniinearum, thougli genprically distinct, have the same sexual anuatnre. Tiiis kind of identity shonld not be contonuded with similarity arising out of reduction. It happens that members of ditferent subfamilies become similar in one or the other organ in consequence of the loss of special structures. Such species are, however, never the same in all parts of tlie copulatory apparatus, as is the case in the insects mentioned lu'fore, the list of wliich is by no means exliaustod l)y tlie few species used as illustrations to show that it is (juito erroneous to maintain tliat all specifically distinct Lejiidoptera exhibit diti'erences in the copulatory organs. (~) Species which are difi'erent from the nearest relatives, but do not show any marked geographical variatiou in the sexual armature. — Here belongs by far the greater proiiortion of the Hawk Moths. The specific differences are very slight or very conspicuous, or intermediate iu degree. They may lie found in one single organ, or in more, or iu all. The differences may be sliglit in some species, and great iu others of the same genus. Two species may difi'er strongly in external features, and little in the sexual armature, or the reverse may be the case. There is every conceivable gradation in this respect. There is always some individual variation, as a matter of course. AVhen examining the armature of only one individual, one is sometimes induced to cousider a certain outline or structure as specific, while it is in fact only an individual character; if possible, a number of specimens from different localities should be examined. Individual variation is most obvious in species with complex structures, but it is here far less easily misleading than in the case of simple structures, because the s])ecific differences are as a rule more conspicuous there than here. (3) Species which differ in the copulatory orgau from the allies and vary in themselves geographically. — Geographical variation is most often met with and is most conspicuous in those forms which are sedentary iu habits. Sluggish species with functionless mouth-])arts and reduced power of flight, species of which the sole function as imagines is ])ropagation, are especially liable to develop into geographical races with differences in the sexual armature. The phenomenon occurs often among Ambidicinae. The geographical differences in these organs, which may or may not be accompanied by differences in external features, are t)Ccasionally surprisingly great. In Pseudoclanis postica from \Vest and East Africa (PI. XXXII. f. 1:2. 13), and Poli/jAj/chus trilineatus'Jrom Ceylon, North India, and the Philippines (PI. XXV. f. 2; PI. XXXIV. f. 7—10), we have extreme cases of geographical variation in the copulatory organs, which cases are the more instructive as the races exhiliit only slight external differences. Oxf/ambuli/x substriiiilis (PI. XXX. f. 3. 4. 5) is also an example to the point. It is quite erroneous to say that differences in the sexual armature are always of specific value. Geographical races may be difi'erent or not in these organs ; aud the diflerence may be minute or consjiicuous. It is idle to maintain that geographical representatives are sjiecifically distinct, if the sexual armature shows obvious differences. What one investigator considers obvious ( Ixxxiv ) ill these (irgans (us well as in (itiiers), a K'coml student, ddcs not think worth noticing-; and wliat appears to ns to he an insigniticant distinction may actually lie an aecessory character of a very important distinction so far bidden from us. Moreover, even if only quantitatively considered, there is no line of separation Ix'tween conspicuous and inconspicuous differences ; and there being geograjiliiciil races whicli are not constantly different, the differences being found only in a larger or smaller jiroportion of the individuals, tliere is also no line of sejiaration between geographically separate portions of a species which are different in all individuals and geographically separate portions of a species wliich are not different. Tiierefore to say which geographical differences are sjiecific and which not wouhl be entirely dependent, not on tlio facts, but on the personal view of the investigator. Specific distinctness, however, as we conceive it, is a phenomenon in Nature independent of the student, and therefore independent of all individual opinion. A view of specific distinctness making it dependent on personal opinion must, therefore, necessarily be erroneous. The taxonoraic value of the sexual armature is not confined to subspecies and species. In many cases there are peculiarities in these structures which relate to all the species of a genus, and only to these. Or a group of genera possesses in some or in all members a certain structure not found elsewhere. In fact, the genital organs are very often of great help in locating a genus or a species correctly. We will give some illustrations. The series of genera from Hemeiojilanes, Pseudofjjhinx, etc., upwards to Sesia, Huemorrhagia, and Cephonodes, has the tenth segment (c?) of the same type ; we find here only modifications of one type of anal segment, which do not recur among all the rest of the Spkingidae semanophorae. Within this series we find in several genera a very peculiar whip-like penis-sheath (PI. LIV.), not met with anywhere else. Ni/ceri/x hypostieta with this whij) will at once be recognised as a relative of coffeae, etc., and not of Protambulyx, with which it has been associated. The clasper of Nyceri/x is always different from that of Perigonia. The tooth of the penis-sheath of Protoparce is always directed towards the riglit side, of Cocytius towards the left. The harpe of llerse convolviili and the other species of the genus is different from that of Acherontia, but the harpes of botli genera are built up after the same type, confirming the conclusion arrived at from other characters, that Ilerse and Acherontia are near relatives. The friction-patch of the clasper consists in Spkingidae a semanophorae of a great number of mostly small scales lying more or less flat on the clasper; such patches as represented by PI. LVIII. f. 29 — 34 do not occur in the Spkingidae semanophorae. On the other hand, this section possesses in many members a friction-organ of large erect lanceolate scales which is not met with among Spkingidae asemanopkorae (PI. LVIII. f. 37). The ribbon of friction-scales on the inuerside of the eighth tergite is found only in Amhtilicinae and some Acker ontiinae. This may suffice to show tiiat the sexual apparatus should not be studied merely with the view of finding differences between species. The general results arrived at with respect to the male sexual armature ( Ixxxv ) liolil good also in tlie case of tlie cnjnilatory oi'saus of the female sex, with the exception of the differences being uiore often iiisiguificant (qnantitatively) or apparently absent. This is due to the vaginal armature being on the whole not so strongly chitioised, and therefore offering, where more or less mem- branaceous, far less obvious differences than in the case of the stronger-chitinised male organs. However, there is an abundance of species, especially among Sphing'ulae asemanophonte, in which the sexual armature of the female jiresents very striking characters (PI. XVll. f. 1. 2. 3), and the differences between the species, where these organs are less complicated, become in most instances also apparent when the structures are closely coaij)ared. While the organs of copulation of the male can be dissected from the body without destroying the seventh and eighth segments or cutting oti' the hinder part of the abdomen, it is necessary to remove the seventh to tenth segments of the female in order to relax them, and then draw tiiem apart, which can be done with the help of a pin. The vaginal armature lying hidden in a cave in most species must be pushed outside by pressure from the inside to become plainly visible. The ninth and tenth segments are fused into one (/A '+ A', PI. XIX. f. 11. 12) and covered with bristles of various length, whirh are doubtless sensory in function. Between the two halves of tiiis double segment lies the anus, and ventrally of the anus the aperture of the oviduct. The segment is always short in Sphinyidae, and there is no ovipositor. The vagina has tiie same position as in other Lepidoptera — namely, between the seventh and eigiith sternites. The vaginal area is more or less snnhen between the two segments, but not always — the vaginal orifice of Coo/tins, for instance, lying on a level witli the seventh sternite, and being plainly visilile after the removal of some scales. The vaginal orifice is surrounded by more or less obvious folds, ridges, processes, grooves, whicli are doubtless adapted to the cdasping-organs of the male. The orifice is mesial and central, sometimes more proximal, sometimes more distal. In the species with asymmetrical male clasi)ing-organs it is sjiifted towards the left side (Pi. XLI. f. 1 — 3). The armature of the vagina is also asymmetrical in Ckuds bicolor (PI. XVI. f. 7), of which the male is not known, and in Hippotion li/cetus (PI. XLI. f 17), the male armature of which is symmetrical. It is perhaps necessary to add that the armature is never absolutely symmetrical in either male or female, the two sides ditfering ?'«fe/- se just as much as one individual does from anotlier, and the apex of tlie penis-sheath being seldom the same right and left. The vaginal area is generally naked, the scaling being restricted to the more or less completely membranaceons eighth sternite. But in some species the scaling extends into the vaginal cavity (PI. XXI. f 12. 15). We find also occasionally species which have the vaginal armature provided witii dispersed (sensory ?) hairs (PI. XX. f. 22). The part of tiie vaginal area in front of the oritice is termed in tliis Revision antevagi nal plate, and the posterior part postcin/itud jdatc. ( Ixxxvi ) Hcsides tlic Vil.uiLial vfiitral area there is tlic eiglith tergite wliicli forms jiart of the ciiiuilatoiy apparatus of the female. This tergite, covered by the seveutli, is never spinose ; it varies in size anil shape (Fl. XV. f. 9. 13 ; IM. XVI. f. (i. 9— 11. lO: IM. XVII. f. .■)— s), and is in many cases different in tiie sjiecies of tlie same genus, occasionally even in geographical forms of the same species (PI. XVIIl. f. .5. 8). The most remarkable eighth tergites are those rci)resented by PI. XVII. f. 10 and PI. XVIII. f. 7. In the former there exists ajiparcmtly a gland (o), from which a groove leads anad to a deep cavity. Tlio geographical variability of tiie sexual armature is very instructive for tlie beginner in the study of comparative morjihology. The similarity between the structures of the various subspecies is evident in sjnte of the differences ; the modifications are easy to perceive, and there is no difficulty in recognising the homology. The meaning of evolution and of relationship is very clear wiien a series of subspecies is compared with one another and with the nearest allied species. No less instructive is the asymmetrical development of the armature found in a uuuiber of Sphingidae. From one jioint of view the asymmetry is even more interesting than the geographical variability. For it is in tlie case of geographical modifications not rarely impossible to say with certainty which of tiie modifications is the oldest, while there can be no doubt about the asymme- trical armature being a development from a symmetrical one. Therefore it is easy to make out by comparison of all the allied species which was the ancestral form of the armature. We liave here not only the line of development, but also the direction followed (see p. l.xxiv.). Apart from tlie penis, the armature of which is nearly always asymmetrical, we meet with asymmetry in the claspers and harpe and in the tenth segment of the male. There is outside tlie Sesiinae only one case of obvious asymmetry in the liarpe, which occnrs in a species of Poli/pti/chus, a genus oi Ambidirinap. In Pohfptyckus dentatus (PI. XXXIV. f. (j) the harpe has a long ventral process {pp) ; the processes of the right and left harpe are the same in length, and both harpes are quite separate from one another. In a clo.-;ely allied species, Poli/pti/chus trilineatus, a remarkable modification is observed. The ventral processes (pr) of the right and left harpe are fused proximally in the Philippine form of trilineatus (PI. XXXIV. f. 7) ; the process of the right harpe (left in figure) is small and slender as compared with the process of the left harpe. In the North Indian form, P. trilineatus itndatns (PI. XXV. f. '1, pr), the two processes are fused together to a fishtail-sliaped structure, ami iu the Ceylonese subspecies, P. trilineatus luteatus (PI. XXXIV. f. lU), the process {pv) appears as a single asymmetrical one, which does not show a trace of its being the product of the fusion of two processes, except at the base, where tlie two halves e.xteud upward on each side, surrounding the penis-sheath. Tiiis combination of geographical variation and asymmetrical development is very interesting. It will be seen, moreover, from the figures quoted, that the upper, shorter, process {pm) of Poli/pti/cltns dentatus is of different length right and left in P. trilineatus (pdl and /jdr). ( Ixxxvii ) The tenth seg-meut is apparent!}' always symmetrical in the Sjjlumjidab with tlie exception of Sesiinae, where it is asymmetrical in a great number of species. The tenth segment of the Si'siinae is normally divided into a right and a left lobe, the tergite {Xt) as well as the sternite (A>) (PI. XLII. f. 1—25). In most species of Sesii/tae the tergite remains symmetrical ; in the highest members of the subfamily it undergoes, however, peculiar modifications. The nearly symmetrical nintii and tenth tergites of Haemoirliagid, and Cephonodes (PI. XLIII. f. 0. 21) become twisted (PI. XLIII. f. 8. 11), and the left half finally nearly obliterates, being represented only by a broad semi-detached swelling {I) at the base of the riglit process (left in figures), as shown in f. 15 — 18 of PI. XLIII. The sternite of llaemorrhagia 'ceinild and Cephonodes janus has two unequal processes (PI. XLIII. f. 12. 14. 2ii) ; tiie riglit one of them disappears, as more fully exfjlained on p. 4(31, there being scarcely a trace left of it (PI. XLIII. f. 25, r), the sternite having a single i)rocess. This ventral process becomes shifted towards the left side in Cepho)iodi's, lying finally in a plane with tlie tergite, and forming with the latter a horizontal forceps (PI. XLIII. f. 8. 9). In Ein/o japix (PI. XLIV. f. 7) the sternite has become simple, but the left process has not entirely disappeared. In Himaiitoidea, also with a simple tenth tergite, tlie sternite has ]>reserve(l a remnant of the aborted left process (PI. XLIV. f. 8), which remnant appears as a tubercle on the left side of the sternite beyond the middle of the latter. The sternite of Epistor (PI. XLIV. f. 10. 11) is simple but asymmetrical at the end ; it is doubtless a development from a symmetrically divided sternite, but there is as yet no species known which shows an intermediate stage. In Calliomma pnircf and Leucorhampha ornatus the left process of the sternite is shorter than the right one, wliile in Pnch)/Ua re.'otmens (PI. XLII. f. 6) the right one is a little longer than the left. In all these cases tlie steriiile is in advance of the tergite in the asymmetrical reduction. The right and left harpes are conspicuously different in several Si'siinae. Fach/Ua darcefa (PI. XLV. f. 26. 27) has on the left harpe a large curved process, which is represented on the right harpe by a small tooth only. In Aleuron neglectum and Epistor lugubris we find the right harpe produced into a long conical, elbowed process (PI. XLVI. f. 13), while the left harpe is very short (PI. XLVI. f. 12). 'Differences which are not quite so conspicuous are observed in the allied species and genera. Sometimes the claspers are obviously different right and left — as, for instance, in some Epistor. The most remarkable cases are again found in the genera llaemorrhagia and Cephonodes. The latter genus has no free harpes ; the right and left claspers are always dift'erent in size and shape (PI. LI. f. 22 — 25 ; PI. LII. f. 1 — 3), the contrast being strongest in (.'. trochilus. In all Cephonodes, as well as in all llaemorrhagia, the left (right in figures) clasper is reduced or modified. The right and left harpes of llaemorrhagia venata are practically the same (PI. LII. f. 5) ; in the other species the left is always shorter than tiie riglit, the latter being often long and club-shaped, whereas the left one is vestigial (PI. LI. f. 17—21 ; PI. LII. f. 4). ( Ixxxviii ) Asyiuimtrv is iilso oliscrvcd in the armatiive of the jieiiis-funiu'l — for instance, ill yi/rrnjx e.rimi (PL LI 11. f. 44). Another kind of iisymuietry occurs in the legs and antennae. Here tlie ilill'cieiicc is not observed in tlie appendages of the left and right side of the iiody, liut ill tiie right and left side of tlie appendages themselves. The antennae and legs arc on the whole cylindrical, but in most instances the internal and external sides are more or less different. Tiiis is especially noticeable in pectinated antennae, and in tlie spines of the tibiae and tarsi. l^csides the differences in the genital armature and in the frenulum and retinaculum, the two sexes of the Spltinyidae show very obvious distictions in other organs. Tiiese distiiulions are the result either of tlie different develop- ment of one or the other organ in the two sexes, or of the ajipearauce of special structures confined to the male. The sexual dimorphism is accordingly relative or absolute, the former being far more frecjueut among the Spldngulae thau the latter as regards variety in the differences. Oue absolute sexual distinction occurs in all species of Hawk Moths. This is a scent-organ * situated at the base of the abdomen. It is found iu all species. The orifice of the organ lies in the pleural membrane above the upper edge of the basal sternite (Fl. LXIII. f. 1. 2, (jl). It is a cavity from which protrudes a bundle of long scale-hairs, which serve as distributors of the scent produced by the scent-cells. A groove or fold runs backwards from the orifice of the cavity over the pleura of the third segment, ending on the fourth. The tuft is sometimes retracted into the cavity, but frequently projects in the dry cabiuet- specimens. Its colour is occasionally carmine or pink (Temnora). The organ is essentially tlie same as that found iu some Notodontidae. Another scent-organ is found on the hinder side of the anterior coxae.* It occurs in all groups of Sphinyidae, but is very frequently absent or vestigial. The difference in the degree of development is often very pronounced in closely allied species. In Coelonia fnlrinotuta the organ is very strongly developed, while it is vestigial in Coelonia soluni ; and C/iroinis erotiis and heliodes differ in a similar way. The organ is more often preserved and on the whole more prominent in the Sphinyidae semanopliorae than in the Sphinyidae asemanophorae {Acherontiinae and AmbuliciiKte). A third scent-organ is met with iu one species only of I'ldlampelicae — namely, in Macroglossum yodejf'roiji. In this insect the mid- and hindtibia are the bearers of the organ, which consists of a subbasal groove, situated on the outer side, and of a large tuft of hair-scales. In the i of tiesia fadus there is a black tuft of scales at the base and another near the apex of the foretibia, not met with in the other species of Sesia. We do not know the nature of these tufts. In the genus Epistor we find a peculiar scent-organ on the forewing, reminding one very much of similar organs observed iu Lithosiinae. The cell of the forewing of the males of Epii^tor yorgon, E. taedium, and E. cari/er is ' llaasc, Iih i. p. 159 ff. (1887). ( Ixxxix ) distorteJ, being covered on the npperside by a fold i)rojecting backwards from the subcostal vein and ending beyond the apex of the cell between li' and R'-. The width of the fold is different in the three species, being greatest in Epistor t/oryon. Underneath the fold— /.«. covered by it — there is a mass of rather large, short, broad, uon-dentate, creamy white scales and a bundle of woolly scale-hairs. The distortion of tlie cell and the neuration is best visible on the underside of the wing, where the cell bears short scaling, contrasting with tiie more woolly scaling on the rest of the wing. In the c? of Eijistor carifer there is, besides, aiiotlier strnctnre on the forewing, cjnsistiug of an oblong and rather large cavity on the npperside in the cell near the base. The cavity bears erect scale-hairs ; the wing bulges out on the under surface, and is here covered with short anl broad scales. All three species have a very woolly abdomen in the male sex, the hair-scales, wliich are long subdorsally, being directed obliquely dorsad (exce[»t on the first tergites) and massed together in flakes. Tiie hindlegs, inclusive of the tarsi, are also very woolly in tbese males, appearing compressed and broadened. Such sexual distinction in the legs like this occurs also elsewhere, tiie forelegs oi Chromis erotus, the hindlegs oi F/tolas anc/temolus, and the hindtibia of Hiinnntoirles undata, for instance, being broadly scaled in the males and simple in the females. The sexual differences in the antennae have been noted above. Rhopalo- psyrlte is the tiuly instance where the antennae are the same in the two sexes. The palpi are on the whole smaller in the female than in the male ; the most obvious differences occur in the Ambali'^in'n'. The tongue is not rarely shorter in tlie female than in the male (I'rotambuli/x, for instance). The abdominal end-tufts are also very often different in the sexes, tlie female possessing in such cases the more simplified tuft or nj tut't at all (J'J//istor ; Spkecodina ; Cijpa ; Uiloplionotn : etc.). The abdominal side-tufts of Oxynmhidyx are present only in the males. The sjiurs of the mid- and hiudtibiae and the tibiae themselves are occasionally longer in the male than in the female (some Polyptychus), while the hindtarsus is longer in tlie ? than in tiie cJ of Pachylia Ji'^uti and Pliolas aiichemolas ; and the armament of the tarsi is also not always the same in tlie sexes, the peculiar protarsal comb found in ('ucytins being a male character (PI. LXIV. f. «). The sensory hairs at the end of the fifth protarsal segment (ventral side) are on the whole better developed in the female (PI. LXIV. f. 1:5, s). The wings, whicli are generally more elongate in the male, show very con- sjiiciious dimorphism in some instances : Poh/pfi/rln(s pauperculu and ruiitraria ; the species of Siaeriiithidux ami Deyma.ptcnt : Oryhu. achemenides ; Ejdntor. More frequent than conspicuous difiereiice in shape is dichromatism. The sexual difference in pattern and colour is often so striking that the sexes have lieen mistaken for distinct sjiecies. Such obvious differences are observed in Epistor, Jlimintoides, Oiyba achemenidea, Vnendosphinx, IsnynnfltuSy Krinnyis, Coelo/ua /'/drh/otutd, Ilerse roiico/ndi (not cinyidnta), Mitumimus (instrulasiae, and other species. (xc) III tlic liiiily of the Hcvisiou many details of tlie niorpliology of the Sftliiiiqiddi' will be mentioned wiiich we iiave only touched upon in the fore- goin.ij Kunimary. Thoiisi'li the Hawk Moths are a compact group, the diversity of de\elo))meut within the family is nevertheless so great that the subject is not ill any way cxliansted by our researches, to which time, material, and ability have set limits. ( -lassification was the avowed object towards which the researches in tlie structure of the Sphhiyidae were undertaken. However, since classification as we conceive it gives e.Kpressiou to the blood-relationship of the forms classified, and since no sound verdict about the degree of relation- ship can be given without an inquiry into the phylogenetic development of tlie ciiaracters which distinguish each Hawk Moth, it is manifest that with tiie apparently narrow object of giving a classification of the Sphingidae issues are entered uihiii wiiich bear njion tlie broad questions of development of the animated world, of which questions we have never lost sight in struggling througli the embarrassing mass of detail i)resented by the 770 species of Hawk Moths. The value of detail for the solution of general questions has again and again been forced upon us during the jireparation of this Revision. At first sight it ajipears to be of little consequence whether 751) or 770 species of Splniiyiilde are known ; whether Aleuron ipliis is the same as neglcctum, or distinct ; whether Odontosida belongs to the Philampeliiiae or Ambuliciiiae. The distinguisliing characters upon the discovery of which so much labour is expended in classificatory research may not seem to be worth the trouble, being apparently of interest only to the collector and specialist. Indeed, if systematic work did not go beyond distinguishing and naming the forms of animated nature and jintting them into some kind of order, there would be justification for those who smile at the efforts of mere systematists. However, the discovery of distinguishing characters assumes at once an entirely different aspect, if the solution of the question "species or no species" is considered but a stejipiiig-stone towards the higher aim of understanding the How? and Why? in Nature. That Ili/loicus pere/egans has a grey form so closely resemliling Ili/loicits chersis that it has hitherto always been mixed up with it, will not interest anybody except perhaps North American Lepidopterists. But if we add that this grey form has a reduced paronychium like clwrsis, while the paronychium is always distinctly lobed in the black-backed form, a difference often distinguishing genera or even subfamilies in Lepidoptera, the rigidity of what is elsewiiere a s[)ecitic or generic character breaks down at once, and it is further evident that, tlie agreement in the paronychium of the grey-backed per- elegans with chersis being observable only under a high magnifying power, tliere are similarities which are decidedly not mimetic. Whetlier tiie right or left harpe of Fachi/lia darcetu is the stronger developed, or whether both are the same, is as indifterent a question to the non-specialist as is the result of a cricket match to a Continental. However, if one knows that iu the Sesiinae with asymmetrical development of the claspers the left side is the more reduced, and remembers the haste with which " laws " in development are often ( -^cl ) ibnnulatetl by biologists, the case of Paclii/Ua darceta Ijecomes of general imijortauce, counselling caution, since darreta has the right harpe more reduced, and hence proves either that there is no inherited tendency in tlie Seminae to a left-sided reduction, or, if there is, that the opposite development has taken place in spite of that tendency. It may be a matter of little moment to an outsider that the only species in tlie second division of the Sphingidae which has in the ])upa a projecting tongue-case, as it is found in many Aclierontiinae, is Rlii/Hchol'tba actcus ; but the matter becomes ditferent if we explain that, firstly, the presence of a free tongue-case in RIn/ncholaba and Aclierontiinae is not a sign of relationship, as the organisation of tiie insects proves, and lience demonstrates the ])0ssibiIity of the appearance of a similar conspicuous character in very distantly related groups, and, secondly, that tlie similarity again cannot be mimetic, since the pupae are buried. In chemistry and physics minute research discloses tlie secrets of Nature. In biological sciences minuteness of research is as much rec^uired as there, if we wish to understand the phenomena of life. Systematic work can do much towards that end ; it is the only means of checking off, proving and disproving, generalisations. But to serve as a reliable censor of conclusions, the observa- tions must be exact, and therefore minute, since exactness cannot be attained without minuteness of imjuiry. A simplification of systematic research by narrowing it down, as advocated in certain quarters, to tbe solution of the questions " obvious ditference or no obvious difference, hence species or no sjiecies, genus or no genus," takes the life out of it. But if tiiese questions are made subordinate, and therefore systematic work suliservient, to the liigher issues of biolugy, the dry detail of distinguishing characters and the ajiparently futile labours of the de^ciiber of species and other classificatory units become iill-iniportant in the science of life, as sui)|>lying sound criteria, where otherwise a lively imagination might run wild and substitute plausible a>sinuj)tioiis for facts. Looking from this point of view at the iliagnosrs of the genera and species of iSjiIiinyii/ue it will be found that they are a soiu'ce for information of a general character. If we compare the pedigrees (facing pp. 24, 499, etc.), with one another, and the characters there mentioned with the fuller descriptions given in the body of the Revision, the first tiling which strikes one is the frequent repetition and reappearance of the same diagnostic character (for instance, the absence of a jinivillns, or tiie spiuosity of tlie tiliiuc). AVhetiier our classification is accepted us valid or not, the fact remains that there is very frequcintly a close agreement in one or more characters between sjiecies and genera of near or of distant relationshij) in the Hawk Moths, an agreement wliicli renders the classification of the 'Spliinyidae esjiecialiy difficult. Since the agreement refers to many details of many organs, as well as to the colour and structure of the scaling and the shape of the body and wings — which give the insect its habitus — the Sp/n/u/idae furnish indeed an abundance of material for a Study of Similarity, towards which we ofter a few remarks. I ( xcii ) Altli(iu;;ii ])li_vl(ig('iictically closely allieil forms may (lifter from each other j in one or more organs to a surprising extent, they are generally similar to one ]' anotlier, if not in aspecf, at least in structure, and if not at one stage, at least i at others of the metamorjiliosis of the individuals. The same kind of similarity V or identity we meet again in sjieci.'s or genera which are not nearly related, but A have preserved one or more cliaracters of the common ancestor. It is evident I that this kind of agreement stands in contraposition to similarities whicli are l the outcome of evolution. The preservation of four wings in Lepidoptera, J Orthoptera, Hemiptera, etc., aud of six fully developed legs in most insects, i the occurrence of ocelli in many Heterocera and other insects, of filiform or J setiform antennae in most orders of Arthropoda, and the retention of a pad to i the claw-segment of the tarsus in many Butterflies, Moths, and other insects, ,■ or of a frenulum in one Butterfly and the greater proportion of Moths, etc., »: etc., establish a similarity in respect to the said characters which is manifestly i of quite a difterent kind from the similarity established by the acquisition of i clubbed antennae or of a sucking-apparatus, or by the loss of the wings or \ the mouth-])arts, or by the adajjtation to bark in colour and pattern, or to leaves • and sticks in shape. j Though the distiuctiou between a resemblance preserved and a resemblance | acquired is obvious, it is not so sharp as it appears to be at first sight, I since the two kinds of similarity are connected to a certain extent by that i acquired resemblance whicli is the direct outcome of inherited community of « characters. We find, for instance, that the distribution of pale and dark ] colours on the legs of Arthropoda stands normally in close cjunection with the i articulation, the segments being very generally pale-coloured at the joints. | White-tipped tarsal segments, and tibiae with three pale rings are of common % occurrence among insects, as are white-ringed antennae and white-ringed abdomina. ] Where there is a break in the surface of the skeleton of insects, there is normally ' also a break in the coloration. Thus we find a stripe along the eye aud lines t at the edges of the thoracical plates in members of nearly all orders of insects. | The wing is perhaps the best example of a conspicuously coloured, regularly ^ interrujited surface, wliere the connection between the pattern on the one hand and the outline and the division of the surface by the veins on the other is very obvious in a great many insects. The original pattern of the insect-wings \ (outstretched) is symmetrical to the body and symmetrical to the veins. On the triangular wings of Liqjidoptera the modifying influence of the cross-veins and the oblique distal margin is especially conspicuous. The position of new markings is dependent on the position of the old ones, new lines appearing often along i tiie old ones and rings around spots. The external lines of a vast number of \ Heterocera are distally concave between the veins according to the undulation i of the distal margin of the wing, while the proximal lines are distally convex j (respectively dentate) between the veins. The median space of the forewing j extending from the costal to the inner margin and including the cross-veins j seems to be greatly influenced in its pattern by these veins, a large number i ( xciii ) of Li'})iil(i[itcra rxliihitiug the same kind of iiu'iliiiu sj)iK-e, outrastiu;^ with the rest of tlie wing either by being different in colour or by the absence of lines transversely to tlie veins. The aj)jiearance of a ''stigma" or a kidney-shaped mark on tiie wings of so many Lepidoptera is tiie direct consequence of the possession of cross-veins. Since the division of tlie wing-membrane is so similar in the various groui)s of Lepidoptera, it is only natural that there are many joints of agreement in the pattei'ii of sjiecies belonging to not nearly related gronjis. It is not the occurrence of this kind of similarity wiiich is astonishing, but tliat it does not occur everywhere, and tliat — bearing in mind the remarkable phenomenon of " Mimicry " — Nature has abandoned in most cases similarity, though it offers such great protection to the species, as we must assume it to do, judging from the great number of cases of " protective " resemblance at which Nature has arrived in a roundabout way. However, notwithstanding the difierences — in the first instance the outcome of segregation — outweighing tlie similarities, the agreement in some such detail as referred to above, which resemblance is the result of a general agreement in the structure of the insects, is the starting-point for a clevelo[)ment towards more conspicnous similarities. The pale superciliary stripe, for instance, is continued in many Sphiiigidae over the pronotum and along the sides of the mesonotuin : a mesial stripe begins between the two lateral ones on the occi{)ut and runs over the thorax, end'ng on the abdomen. The lateral lines converging frontad, the tliree together are similar to an arrow. The same arrow-pattern is met with in other orders of insects — for instance, iu (Joleoptera {Currulionidae, Cerambi/cidne, etc.) ; and the agreement extends even further, the mesial vitta being either simple or divided into two lines, and these lines being either parallel or divergent behind, in the Hawk Moths as well as the Beetles. Moreover, tlie mesial vitta of the head and thorax becomes a sutural vitta on the elytra of C'oleoptera, and similarly the hinder edge of the forewing of Lepidoptera, which corresponds to the sutural edge of the elytra, assumes often the pale colour of the thoracical mesial vitta in those Heteroccra iu which the wings have a roof-shaped position over the abdomen when the specimen is at rest. Tlie conspicnous colouring (scaling, pubescence, or pigmentation) of the mesoscutellum found in many insects and the three rows of spots so frequently occurring on the abdominal sternites belong to the same kind of resemblance as the thoracical vittae, the respective markings being homologous in the various insects, evolution starting from similarity and resulting in higher resemblance. In structure the same jihenomenon is no less frequently observed. Here we may conveniently dis- tinguish between similarities arrived at by retrogressive and by progressive development. Kednction of organs occurs very cdmnionly among Sjjlihu/id'ie, and leads often to the loss of the organs. The disappearance of the horn of the larva iu some Fhilampelinae and Ambulicinae and in Lapara (Aekeronfi/nae) ; the reduction of the tongue-sheath of the pupa iu Ambulicinae and several Ac/terontiinae ; the absence of an externally visible sheath of the anterior femur of the chrysalis in ( xciv ) ^|iccit's (if viirions suht'iLiiiilifs of Sphnnjidac ; tlic loss (if tlic touijue in (sunie Amhiiliriiuie iiiul Arhcrontihtue, of tlie proximal pnir of simrs, of the pnlvillus and ])aroiiycliium, of the frcniilnm and retinacnlnni ; the rednction of the mid- and hind-tarsal comb and of tlie paljii ; the rednction of the tarsal spines in nnmber, of the tibiae and tarsi in length and of the head in size ; the loss of the fascicnlated ciliae in the male antenna of l{liop((Iojj;/sclu\ of the .abdominal spines in various Amhiilii-iniw, of the ]iatch of sensory hairs at tlie base of the jialpi in \\w Spliiiigulae asemaiiflpJionn', and Gnrelca and allies among the Sfihingidue scmanophorae, of tlie bristles on tlie jiilifer, etc., establish between these Spkinqidac, as well as between them and other Lepidoptera, a similarity in the jioints mentioned which is of importance for the jirojier comprehension of the origin of agreements in structure, since there is no doubt that, at least in most cases, these negative characters are independently acquired in the various gronjis of Sphiiiyidae and other Lejiidoptera. The loss of the wings in a number of Lepidoptera, and of the hindwings in Diptera and many Beetles, may be mentioned in this connection. The independent effacement of differences in consequence of loss occurs also in pattern. Uniformity in colour may be arrived at either by the exaggerated development of one colour leading to the suppression of a pattern, or by the markings disappearing on account of the non-development of the pigment (or of the particular structure in case of structural colours). The pattern is in most cases only partially effaced. The partial transparencies of the wings as found in Cocijtius and some Haemorrhagia have no more to do with one another than have the entirely white anterior wings of some Beetles (elytra) and Butterflies. Tiie loss of markings in Mimas tiliae is as independent from that of Amorpha popiili, as is the disappearance of yellow abdominal side-patches in certain Acherontiiime and some Macroglossum. The pale colour frequently exhibited by a large percentage of specimens which are obtained by breeding in and in— for instance, in Amorpha populi — is due to the meagre production of dark pigments, and reminds one of the pallid cave-insects. The weakening effect of breeding in and in and the result of a life in dark caves are, therefore, similar in so far as both lead to the loss of pigment, producing clayish ochraceous forms. The causes are different ; the result is similar. Indeed, if the numerous cases of the disappearance of distinctions in structure, colour, or jiattern are inquired into, it will be found that the causes of the loss of distinctive positive characters and the reasons for their disappearance are as different as the starting-points of the retrogressive development which residted in the effacement of the positive characters, replacing the ])irture which was variegated in pattern and structure by a clean surface. The reduction of organs is often accompanied by the appearance of certain ])ositive ciiaracters in Spliingidae. The small head is frequently crested ; the small eyes become lashed, and the reduced tibiae often spinose. The abdomen, whi(Oi has very weak sjiines at the edges of the segments, is often spinnlose all over the tergites or assumf^s a woolly appearance. Weakened power of flight and broadened wings are frequently associated. The positive and negative characters ( xcv ) are not necessarily deiiendent, on one another. Spinosity of the tibiae, tur instance, occurs in very many insects without a rednction of the tihia Imving taken place, and eye-laslies are not always a sign of a reduced head. A conspicuous specialisation which is of common occurrence in insects is the pro- longation of the foretibia into a pointed thorn. We iind this thorn or claw among all subfamilies of Sphingidae except Choerocanijnnac. The i)rolonged outer spines of the first protarsal segment (generally three in number) found in many SphiiKjiclae and otlier Heterocera are also a s[)ecialisation which does not necessarily indicate relationship, and to are tlie clubbed antennae of Sphinyiilae, Butterflies, Argeriidae, some Noctuidar and (reonvtridae, Agaristidae, etc. The ])ectination of the antennae is of special interest. AVe find such antennae in several orders of insects. The remarkable point is that there are antennae which are similar in aspect to the pectinated ones, but are not pectinated. They are of two kinds. The one type is that commonly found in the males of Sphingidae where the seriated cilae look in a dorsal view like lateral expansions of the segments ; and the other type, not met with among Hawk Moths, possesses a heavy subdorsal bristle on each side, which Entomologists have often enough confounded with a true process of the segment. Here we have an obvious resemblance not based on homology. Likewise, the pectination of the antennae of (.'ressonia and Ceridia, or of Suturniidae and Enpterotidae, though the antennae are similar to one another in aspect, are not homologous. Tibiae mottled with single dark scales have very often the appearance of being spinose. In the place of the before-mentioned foretibial claw there is in some other Sphingidae (for instance, some Polgti/chus) a heavy s])ine homologous to a hair, not to the claw. The horn-like projection of the palpus in the Acherontiine genera Cocijtius (America) and Codonia (Africa) is the third palpal segment, while the equally conspicuous projection found in the Sesiine genera Aleuron and Eni/o (America) is a process of the second segment. The same segment has acquired a j)rojectiou similar to the latter in the very distantly related African genus Ht/paedalia. Similarities analogous to the above occur also in pattern. Lines, streaks, and spots on the wings, and spots and belts on the abdomen, superficially but strikingly alike in different insects, often prove on closer study not to be homologous, the belts being, for instance, basal to the segments in one insect and apical in another, and the wing-streaks standing here upon the veins and there between them. Such similarities, which mostly do not affect the general habitus of the insect, are nevertheless very instructive. There is another kind of equally remarkable resemblance in some details met with among Sphingidae which appears at the end of more or less distantly related branches. It is a priori conceivable that a certain ciiaracter of structure or pattern present in closely related forms is preserved when these forms develop divergently further and further. Now, just as this certain character may be lost in the one branch, or in some members of it, at one period, and in the other branch at tlie same or anotiier time, so the character may at one time or the other develoii progressively in either branch, and this progressive ( x.vi ) (Icvi'ldinuciit Iriiil imlcpeiidciitlv In iui identical siieciiiiisatioii in the two divergent liranches. The spination of the abdomen furnishes a noteworthy illustration. In the higher ASesii/iae the spines become very strongly cliitinised on the tergites and sternites, and those of the ])roximal row assnme a short and broad form in Senia, Cepkonodes, and llaemorrhaijin \ the same form of sjiination is found again in the highest genera of the tribe JS'epIirlicne of tlie subfamily Philatnjicllnac — namely, in Afac/0(/lo.'<sum, l{hoi>alops>/clie, and LcKcostrophus — and only here. In the Pliilampelinae there appear here and there characters which are met with again among the Sesiinae — for instance, the very strongly clnbbed antennae, the angulate mid- and hindcoxal raerum, the close connection between abdomen and thorax, the fan-tail, etc. : wliile, on the other hand, Fhilampelinae have also arrived at similar specialisations as CJioerocampinae — for instance, in the shape of the chrysalis, the sexual armature, the end-segment of the antennae, and the ocellated spots of the caterpillar. The American (Jhoerocampine genns FJianoxijla is a derivation from Xi/loplianex, from which it has become different in acquiring an apical tuft to inner surface of the second palpal segment, a tuft which is present in many Old World Clioerocampinac. The peculiar cavity at the end of the first segment of the jialpns, on the onter side, occurring in all species of the Choerocampins genus Tlieretra and the derivations from this genus, is acquired also by a few species of the allied genus Hi}ipotioii, and it is highly surprising that the same peculiar sjieciali.-^ation, which does not seem to occur outside the Sphingnlae, is as conspicuously developed in one solitary species of Achprontiinae, in Megacorma ohliquoi'-, the relationship of wliich with the Choerocampinae is very distant. This similarity reminds one of the projecting tongue-case of the pupa of the Choerocampine genns Rhjncltolabu and many Acherontiinae, a Sphingid character confined io that one genus and the one subfamily. The pattern of the Hawk Moths offers also obvious illustrations of the kind of resemblance under discussion. The most striking is the similarity in the pattern of the abdomen of the lower Acherontiinae and the higher Pliilampelinae. Here we find those conspicuous yellow side-patches bordered by black which are restricted to these insects. The white abdominal belt of the American genns Sesia and the African Philampeline genus Leucostrophiis, which belt Entomo- logists have erroneously considered to indicate a close relationship between the insects, may also be mentioned, as well as the black-bordered yellow hindwing occurring in many Sphiinjidac and other Moths, and the wliite fringe of the first abdominal tergite of the Oriental Deilephila hijpot/ioiis and the American Aleuron iphis and allies. The antemedian and discal blackish brown bands on the forewing acquired here and there in Spliinc/idae are the result of the inter- 8})aces being filled in with blackish brown scaling. These examples of similarity are a kind of belated expression of relationshi[). It will be observed that in a few of the instances mentioned the agreement in some character affects tiie aspect of the insects which have the character in common. There is a complete gradation from similarities in some small detail to similarities of the ensemble. ( xcvii ) Th(! Otif^es of miitiuil agrociuent in asipect may lie classified in two groups — similarity as tlie outcome of adajitation towards the same or similar extraneous objects, and similarity as the jimduct of adaptation to each other. The colour and ])attern of tlie upperside of tlie body and forowinL;" of Sphinqiihie are very generally imitative of the bark of trees, variegated witli algae and lichens. The bark-jnittern is common to a great many Moths, Beetles, Ortho]itera, etc. The similarity is often eniianced b\' the appearance of projections — as, for instance, the thoracical tufts of Eriniii/is. I'seudosjthinx, Hetneroplanes nomn/s, Kj/it^tor, and Notodordichte. The silvery marks of the fore- wing of Xepht'le, Hemeroplanes, Madori/x, etc., and, among JS^octuidae, of IHusia, break the uniformity of the wing, as do white lichens that of the bark. The subapical butttsh jiatch of Eiiri/pteri/x niohwca aud some Notodonts has the same effect. The shape of the distal margin conforms very often to the irregu- larity of the bark. A dentate, scalloped, or lobate distal margin may be actpiired almost everywhere. In Spliingidae a very irregular distal margin occurs, commonly among Amhulicinae, Sesiinae, and Fhilampelinae. In Acheroiitiinae the margin is at the highest undulate, while there is one (American) species of Choerocampinae which has an irregularly lobed forewing looking like a piece of wood {Phanod-'/la). A resemblance to leaves is also frequently met with, especially among Amhulicinae. The leaf-like forewing may either be entire and have a dark apical line indicating the midrib (<7««/.5, some Poli/pfi/cluis), or it may be irregularly lobed. Phi/lloxiphia (PI. I. f. 4) has perhaps the most leaf- like entire forewing among Spliingidae, the midrib being here represented by a line running from the apex to near the base. The clayish colour of this insect, agreeing with the colour of dry leaves, occurs abundantly in Hawk Moths. A leaf-imitation after the well-known pattern of GaHtroparha quereifolia, in which the widened hindwing projects beyond the costal margin of the forewing, is found in some Amhulicinae, Sesiinae, and Pliilampelinae — namely, in Amorpha, Calasi/mholus, Plii/llosphingin, StoUdoptera, and Ihjpaedalia, perhaps also in Dcgmapiera. Besides the imitations of leaves, bark, and other non-animal subjects, there are striking cases of resemblance to other insects among the Spkingidae. The imitation of humble-bees and humming-birds by some Sesiinae is well known. And here again similarity between the mimetic Euproserpiniis flavo- fasciata and the mimetic Haemorrhagia diffinis is purely adventitious, both imitating the same or similar models, and therefore resembling also each other. The resemblance of two species to each other acquired indirectly in the way indicated is of frequent occurrence. And it is easy to perceive that, since the similarity iias independently arisen, insects similar to each other in this way may occur in widely separated countries. Though very many Sjikingidae are similar to one another, there does not seem to be a species which mimics another Sphingid — i.e. of which tiie pattern or shape has been developed directly in relation to the other, ( xfviii ) Tlie various kinds of siuiilaritics sliortly rclV'rred to are met witii attain in other groups of insects. Tliey demonstrate, we think, not only that there are various ways to arrive at and different causes for resembhiuce, but also that there is abundant material of similarity in details upon which selection may seize, and modify and augment the incipient resemblance, and i)roduce those wonderful likenesses whicli illustrate that fascinating iihenouienon called Mimicry. A studv of resemblance from this ]K)int of view is incidentally also a study of relationship, and therefore an essential part of classification. It is well known that the older students of Entomology were fret^uently taken in by similari- ties which do not indicate rehitionship, associating often widely different species on account of superficial resemblance. The older systematic works, and old collections which have not been disturbed, testify to this abundantly. Bnt even modern work, or rather work of recent date, is not free from striking blunders of this kind, although the flourishing study of Mimicry has, or should have, made everybody suspicious of mere resemblani^es in aspect. We well remember trying in vain to convince a famous Lepidopterist, now dead, that the streaked Indian Papilio mararci/s, .rrnocles, etc., are not nearly related to the streaked form of Papilio clytia ; and we notice in a recent number of Lepidoptera Indica that the mimetic Papilio rhett'iiof and its models are housed in the same Moorean genus. The association of the Sphingidae with the Aegeriidae and Zi/gaenidae from Linne down to recent times was due to a certain similarity in the shape of the antennae being erroneously interpreted as meaning relationship. The older writers were the more convinced of the correctness of the association of the Aegeriidae with the Hawk Moths, since the clear-winged Aegeriids ai)peared to them to be connected with the ordinary Sphingidae by the equally clear- winged species of the Sphingid genera Cephonodes and Haemorrkagia. The fallacy of the conclusion has now become evident to every student of Lepidoptera, though perhaps not to every collector of " flies." The similarity of Pseudonphinx tetrio to species of Protoparce, the " Bombycine "' appearance of Arctonotus lucidus and Lapara, the resemblance of Deidamia inscriptum and other Philampelinae to certain AmbuUcinae (= Smerintkinae auct.), the agree- ment in appearance between Proserpinus Jiavofasciata and Haemorrkagia, of Ni/cenjx In/potsticta and Amplgpterus, of Akbesia davidi and Proserpinus, of Sesia and Leucostrophus, etc., etc., have freijuently misled even modern classifiers of Sphingidae. The student who tries to build uj) a classification based on relationship has to beware esi)ecialiy of two kinds of similarities referred to above : (1) similarity as the outcome of reduction, and (2) similarity as the result of adaptation to similar extraneous objects or to each other. The first point is the more frequently overlooked in classification, and re(^uires some further explanation. If a, b, c, d are four organs of a s]:ccies or a genus, and «', b', c', d' the same in their most reduced state ; if, further, the species or genus develops C xcix ) iuto four others Ly one of the organs becoming reduced in eacli, and these four again into more in the same way, we have tlie following dingram (for the sake of convenience in printing the middle part of the diagram has not been completed) : — V. al/c'd' a'b'c'd' a'b'c'J' <,7/cd' aUcd' alicd' IV. a'b c'd ah cd' a he'd' a'b'c d a l/cd' a lie d I I I I I I III. II. I. n'h'cd (I'lic'd A : abed I „:bt;l' ah'cd a'b'c'd' a'b'c'd' a'b'c'd' a'b'c'd' a'b'c'd' a'b'c'd' . I , , I , , . I , I , , . I , . I , , a b cd a'bc d a b cd ab c d a be d' ab'c'd' I I I I I I ,1 , abed' abed abed ab'cd' I , abed' : D. abed' The development, starting from I., results first in a series of four derivations (II.) differing from one another in two organs. The second series of derivations (III.) contains forms deviating from each other in two, three, or four organs. In the third series (IV.) tlie forms differ only in two organs, or are identical. And we arrive finally (V.) at a series of identical forms. There is divergency from I. to III., and convergency from III. to V. Considering now the phylogenetic relation and the similarity of the various modifications, it will be found that the derivations of A in series IV. are more similar to the derivations of D in the same series than to A in series II., and that the derivations of A in series V. are indistinguishable (in organs ahal) from the respective derivations of D in the same series, while they are difierent from the forms in series IV. from which they are derived. It is therefore evident that tlie classifier who judges from these organs abed alone cannot possibly find out the true phylogenetic connection between the various forms. He will easily mistake the forms which are equally reduced for forms wiiich are closely allied. To him the members of series V. will appear to be very near relations, while they are in fact the end-products of different jihylogenetic branches. Let us take as illustration the retrogressive development so often observed in Sphinyidae of the tongue, the midtarsal comb, and the claw-segment. If we start from a form with long tongue, with midtarsal comb, and with fully developed pulvillus and paronychium, and assume as Ijefore that one of these organs becomes reduced in each derivation from tliat ancestral tyi^e, we arrive finally at forms which agree with one another in having all the four organs reduced. Therefore the reduction of the tongue and tlie absence of the mid- tarsal comb, of the pulvillus, and of the paronychium in those species or genera are not an expression of close relationship. Looking at the pedigree of the genera of Sphingicae (facing p. 30), it will be noticed tliat a reduction and loss ( f- ) of organs like tliosc ofciir iiidi't'd in many genera wliieli stand far apart. lioss, reduction, and fusion may occur independently everywhere. An instructive illustration of erroneous classification of this kind outside the SphliKjidae are the Ifkoniiindt' (= Xeotfopinac), wliich are divided by Salvin an<l (lodman and later on by Schatz into two groups, the one containing the forms with strongly reduced forelegs, the other those with less reduced forelegs. The groups as they thus stand comprise ({uite lieterogeneous material ; they are not a division based on relationsliip, but on similarity in that one organ, each group containing genera wliich have their nearest relatives in the other, the species or genera with reduced forelegs being derivations from such with more fully developed forelegs. The striking resemblance in pattern between certain members of the two groups is explained, at least in some cases, by the similar species being nearly related, and not by having developed convergently. A form which has lost an organ becomes not only similar in tin's respect to another which has equally lost it, but also to those which never did possess the structure. The abdominal spination, wliich is ajiparently confined to the Sphiiigidae, is lost in several genera, which therefore agree in the non-spinose abdomen with the Notodontidae, for instance. The loss in many Sphingldae of the midtarsal comb, of the forecoxal scent-organ, and of the friction-organ of the clasper constitutes again similarity with 2\ofodo)itidaf', SatHrniidae, and other Moths, and not relationship. There is anotlier point illustrated by the pedigrees of Sphingidae which is of equal classificatory importance. A genus and a species are a composition of many organs ; and since each organ becomes in most cases modified independ- ently from the other, it is evident that a species or genus may be specialised in some and generalised in other organs. It is indeed a rule holding good in nearly all Splungidae — we say " nearly " in order to be on the safe side — that a species or genus is further advanced than its nearest allies in some res])ects and at the same time left behind in other details. This being so, it is clear that a genus or species is not (or rarely) a direct derivation from another. Though we have directly connected the genera with one another in the pedigrees, we do not mean to say that, for instance, Proserpinus is a derivation from Rhodosoma as this is now, but wish to indicate only that BhodoKoma is of all recent genera the one which comes nearest in the characters mentioned in the pedigree to the ancestral genus of which Proserpinus, Ampliion, and Sphecodina are developments. Each genus is, so to speak, the end-development of its own branch. In the pedigree facing p. 499 the genera Enpitianga, Rethera, Cizairr, and Rhodosoma are thus connected :— I I Rethera Enpinanga ( ci ) We have adopted tliis kind of connectiou for the sake of simpHcit}- of printing Tlie pedigree should read as follows : — Hh'iiloxoi/ia Cizara Rethevd Eiiiiininuia If couiieeteLl like this, each genus may have aeipiired its own specialisations after the point of separation, and each may also liave preserved generalisations lost by the nearest allies after separation. GEOGBAFHICAL DISTBIBUTIOIT. The Hawk Motlis are an essentially trojiical family, the number of species existing in the temperate regions being comj)aratively small. Very few Sjjhingidae. extend into the Arctic Regions, and then only as occasional visitors, no species being known to breed beyond the Arctic Circle, though some {Amorpha pojinli and amnrcnais, SjJiht.t oeellat i, Celcrio ii<UUi, Ili/loicus pinastri) may be expected to occur as far nortli as their food-plants go. As the greater number of species have a powerful flight, and are, moreover, often wanderers, covering wide distances, the area inhabited by some of tiie s])ecies is very extensive. The proi)ortion of geographically uniform species of Sjtl/inyiddi^ is large as compared with other families of Lepidoptera. Methodical researcli has jn-oved to the systematist tiuit countries like West and East Africa, India and the Malayan Islands, Central and South America and tiie Antilles, tiie Andes and Eastern Brazil, etc., etc., are inhabited by sj)ecial races of nearly every species of Hutterfly there occurring. The stiuleiit is, a priori, certain in most cases that a species of Butterfly not yet discoveretl in some such district will, if occurring, exhibit some kind of <listiuction. In mpluryidne a priori conclusions like this would mostly be wrong. Since geographical variation depends on geographical isolation, it is evident that those Lepidoptera which are stationary are far more variable geographically than those which are not prevented by geographical barriers from constantly crossing over to other districts. The l)Ower of flight, however, is as such no i'actor elfacing the geogra{)hicaI barriers. The swift-flying Ckaraxcs and I'apilio vary geographically as much as do slow-flying Butterflies ; and in birds — which may be called stationary from the point of view of a student of geograpliical distribution in so far as they breed in their native country — geographical variation is also tiie rule and not the exception. On the contrary, the more powerful the flight, the more is an animal enabled to flee from or resist the factors of passive distribution (water and windj, to counteract their influence. But if powerful flight and the haliit of wandering during the period of propagation coincide in a species, the ( eii ) geogriiiihiciil liuriiLTs becouu' ineffective, and the inflnence of isolation (se<;'i-ega- tiou) is annulled. The wandering Hawk Motlis, like Acherontia atropos, IJerse coi/eoln/// and cingidata, Ilijjpotion cclerio and osiris, Deilephila nerii, Theretra nessus, etc., do not vary geograj)liiciilly, or only in a slight degree. However, considering what we have said alioiit the jiower of flight and isolation, it cannot be wondered at that a good many species of swift-flying genera iiave developed into geograjdiical varieties (= subspecies) in spite of their power of flight. Notwithstanding tliis fact, in the Spliiin/iclae the rule holds good tiiat, on the wlioh', the sluggish sjiecies are more often and more distinctlj' varialde geographically than the swift-fliers, and that, further, the areas of the geographical races of slow-flyers are mostly smaller than those of the subspecies of less slnggisji Hawk Moths, as exemplified by AmbuliciiKte and Clioerocavipitiiw. There is only one cosmopolitan Sphingid, Celerio lineuta \ it consists of three subspecies, one inhabiting the AVestern Hemisphere and two the Eastern. Herse conwlvuli and Ilippotioii celerio occupy practically the whole Old World, while Herse clmiulata and Protoparce sexto, occur from Canada to Patagonia. One species {Celei-io gallii) is common to the Nearctic and Palaearctic Regions. Ce.phonodes In/las is found, in three subspecies, from Sierra lieone to Mada- gascar and eastwards to Australia, with a wide gap in the Malay Archipelago. DeilepJrila nerii inliabits the wiiole Aethiopian Region, and extends far into Europe and south-eastwards into AVestern India and Ceylon. A number of species occuir all over the Aethoi)ian Region without being split up into subspecies {Ilippotion eso/i : Busiotliia medea ; Coelonia faU-iiwtdtu : etc.); others are of wide distribution in tlie Palaearctic or Oriental Regions (Macro- glossum utelldtdrnm : Hippotion boerliaviae, velox ; Theretra ?iessus ; Cephonodes picus ; etc.); and others again occur all over the Neotropical Region or go even far into the northern Iialf of tlie New World (.\///oj>//nnes tersa, pliito : Psei(dospkiii.r trtrio : ]Crinrri/is (dope ; Kpistor //fc/id/r/.s : etc.). The distribution of such common species is very instructive in one respect. Although their range is wide, it is nevertheless restricted, and we ask ourselves, What antagonistic factors are there preventing these Sjihinfiidne from occupying the whole globe ? A country must be accessible to a sjiecies. If tiiere is a practically unsurmountable physiogra{ihicaI barrier, the species have little chance of passing over it. The swift-flying Spkingidue are not able to cross the Atlantic and Pacific Oceans, none of the New and Old World species being identical, not even Celerio lineata and gallii being the same in the Western and Eastern Hemispheres ; and the Indian Ocean is an equally effective barrier. But there is no such barrier to prevent Macroglossmn stellatarum, which ranges from Morocco and Ireland to Japan and North West India, from going farther south and east in India ; no physiographical chasm faces in Spain the Aethiopian ihppotioii osiris, which extends from Madagascar to the Pyrenean Peninsula ; no geographical barrier keeps the African Deilephila nerii back from settling ( ciii ) farther east than Sontli India and ( 'eylon, and jirevents Profopdrcr ochus, SO coiuiuon in Mexico, from cxtcndiiig sonthward beyond Venezuela ; no ocean, no higli mountain-range, and no wide desert restrict Oreefa to Southern Brazil and Argentina, Tlcretra capensis to South Africa, Hi/loici(s lagens and allies to Central America, Hi/IoIcks cIu'i-.s/h and allies to North America (and Mexico), and Celerio vespeiiillo to Central and South East Europe. The range of these Sphingidae is restricted because the conditions of life (temi)erature, food, com])osition of fixnna, etc.) are not suitable outside their present range. It is therefore evident that tlie limits of the range of a species are determined by two kinds of factors : physiograpliical barriers, beyond which the Sfjecies is piiysicaily prevented from going ; and biological barriers, beyond which the species is not able to exist. This lieing so, it follows that it is erroneous to conclude that the limits of the range of species indicate always that a physio- graphical barrier has formerly existed, that there was in the Continent or the chain of Islands a discontinuity barring the way. One has to carefully discriminate between those facts of geographical distribution which allow of conclusions being drawn as to the former configuration of the earth's surface, and tliose facts which are the result of tlie action of biological causes.* Since anything strange attracts more attention than the normal, it is not rarely the unexpected upon whicli the student of geographical distribution lays most stress, often exaggerating the significance of single cases and drawing conclusions from them which are contradicted by the otlier, normal and hence neglected, cases. Some small percentage of an Indian element in tlie fauna of Madagascar misleads many a student to treat Madagascar as standing faunistically closer to India than to Africa, and some small but conspicuous difference in the fauna of Bali and Lombock, which difference is in insects not larger than that between most other adjacent Malayan islands, gave rise to the famous but (in Lepidoptera) arbitrary Wallacean line separating tiie Miday Arcliij>elago into a western and an eastern district. Among the Sphingidae we iind a number of sjiecies and genera with a ])eculiar]y striking distribution which miglit easily give occasion to similarly fallacious and misleading conclusions, ('eplionodes hylas consists of three subs])ecies, which occur one in Africa and Madagascar, another in India, China, and Japan, and the third in tro[)ical Australia, no representative being found in tlie Malay Archipelago. Celerio lineata has two subspecies in file Old World : one inhabiting continental Asia, Africa, and Eurojie ; and the second tropical Australia. The genus Nephele, so common in the Aethiopian Region, has bnt one species each in India and the large Sunda Islands and Australia. All these insects are so common that it is not likely that they have been missed by the collectors in the Malay Archipelago. Further, the genus Cephonodes has four species in the Malagassic Subregion, six in the Papuan Subregion, and not more than two anywhere in India and the Malay Archi- jiclago. Looking at such cases, the Antarctic Continent most conveniently comes on the scene as a dfiis ex machina to exjilain the peculiarity in the distribution, * See also Xov. Xool. iii. p. .''lO.') (1S9C,). ( civ ) while the true cause of the absence or scarcity of those insects in the Malay Archipelago is probably of i|uitc a (liflerent kind, as we shall see later on. The North American Spln-cudiita ahhotii is confined to the Atlantic side of the Continent ; tiic only other species of tlie genus (tiie only near ally, in fact, of abbotti) is restricted to the Pacific side of Asia (Aiuurland, (Jhina). Tiie Indian genera Ajjocali/iisis and I'soidoifolbiiia stand quite isolated in the Old World, their nearest (and very close ally) being the Neotropical Euri/glottis. The South African genus Rkodafra comes nearest to the Palaearctic Perf/efia. If we compare such cases with tlie distribution of Cclcn'o, of which one species is cosmoi)olitan, one Holarctic, several Palaearctic, and one each Malagassic, Argen- tinian, Chilian, and Cuban, while two are confined to the Sandwich Islands, it becomes obvious that the interrupted range of genera or allied genera is the result of the extinction of the insects (or allied species) in the now non-inhabited intermediate districts rather than the result of great geographical changes. We have to do here with the remnants of once more uniformly distributed gronp.«, remnants existing where by chance tlie conditions were favourable for their preservation. However, it is not our intention to speculate on the aj)parently abnormal distribntion of single species and genera, fascinating as it is, but to lay before the reader a short summary of the actual stale of the distribntion of the ISjiIiingidae, from which he will better understand the comfiosition of the Sphitigid fauna ot the varions areas tiian from a speculative elncidation of single cases. Accepting the Wallacean division of the globe into five Regions, we have a distribntion of the 770 species of Hawk Moths, as illnstrated in the table on the op{)osite page. There are only 2 species common to the VV'estern and Eastern Hemispheres, each represented in America by 1, and in the Old World one by 2 and the other by 1 subspecies. Of the remaining 768 species, as yet known, 293 are ]>ecnliar to the New AV'orld, inclnding the Sandwich Islands (3 species), and 47;") to the Eastern Hemisphere. Since the Papuan Snbregion and Africa luirbonr without doubt a great many undiscovered species, i)ro])ortionally more than Soutii and Central America, we can estimate the proportion of the species inhabiting the New and Old Worlds as being 1 : 2, or presumably 32U-odd to 650-odd. The superiority of the East over the West of the globe in the number of species of Hawk Moths is quite intelligible, since tropical America is far smaller than the tropics of the Eastern Heniispiiere, does not offer so mnch variety in the conditions of life, and is not so broken physiographically. A heteromorphic country has a heteromorphic fauna. The paucity of the SpJiiiu/idae in the temperate Regions is very apparent in the table. North America, Europe, North and Central Asia, and Palaearctic Japan, where few new species will be discovered, if any, jwssess only 100-odd species peculiar to them, as compared with 620-odd peculiar to the only partially explored tropical and subtropical countries. New discoveries will alter ttie ])roporti(ni so much to the disadvantage of the northern temperate Regions that the proportion will ultimately bo about 1 : 9, according to our calculations. ( ev ) a a "i ■r, 5 Op:; I CO I, I I lit ( t-vi ) TIu' (liliiTi'iict's ill tlic number of species between the three tropical He^ions are not considerable, tlie Neotropical Region possessing altogether 236, the Aetliiopian 170 (many new ones will be discovered), and the Oriental 2.")0. The Palaearctic and Nearctic Regions iiave each 75 species. The differences in the distribntion of the species of the five subfamilies into which the Sphiinjiiha' fall in our classification are remarkable. The Aclierontiinae and Sesiinae are for the greater part American, while the Ambulicinae, Philnmpelinae, and Choerocam.pinae have more representatives in the Old than in the New World. The difference is especially large in the case of the Sesiinae on the one hand, and Amltitlirinae and Plulumpi'Unae ou the other. The projiortion between western and eastern species of Acherontunae (84 : 52) is almost exactly reversed in the Clioerocampinae (56 : 86). The Ambulicinae are most numerous in Africa, nearly half of the Philampelinae are Oriental, and about one-third of them Aethiopian, while the Neotropical Region has scarcely one-twelfth of the total. North America has no single species of Choerocampinae of its own, but Central and South America possess more species than auy other single Region. The number of species of Ac/ierontii»ae is much larger in the Neotropical Region than anywiiere else. However, the numbers of species which inhabit a country in themselves cannot be trusted when forming an opinion on the diversity of its fauna. Comparing, for instance, the number of Clioerocampinae peculiar to the Neo- tropical Region with the number of purely Palaearctic species (48), it would appear from these figures alone that the Palaearctic Sphingid fauna was not more diversely developed than the Clioerocampinae of South and Central America ; whereas, in fact, these Neotroi)ical Sphingids belong to only three allied genera, i.e. are very homogeneous, and the Palaearctic species to more than twenty genera of five subfamilies. The diversity in tiie development of a family is better illustrated by the genera of the country, since a genus is a category higher than species, rejiresenting as it were the generalised state of development of wiiich tlie various congeneric species are modifications. Each Region has one or more genera ci ntaining a large number of species, while the nearest allied genera comprise as a rule few or single species. Protoparce, Krinni/is, and Xi/lophanes are large Neotropical genera ; Uijloici(» has nnmerous species in North America, Celerio in the western half of the Palaearctic Region, Macroglossum and Tlieretra in the Oriental Region, and Poli/pfi/clnis in Africa. Such genera and the com- monly occurring species give a' country its special feature from the point of view of a collector, while the number of genera and their diversity are the more important feature for the classifier and for the student of the origin of the fauna, relying as they both do principally upon the affinities presented l)y the genera. We have grouped the 770 species of Spkingidae in 167 genera, of which the table here following gives the numerical distribution. Since genera are groups of species, and hence generally of wider distribution than single species, the number of genera common to adjacent Regions is proportionally very much larger than the number of species common to them. ( cvii ) i L. tZ ^ 3 p- t» « «, 1-1 1-4 S N CO CO. 1H II 1-f-, a •• -.-H .. -^ 1 .. ■• Wl ■• .-1 •• >l:^ a o : J ^^ J J J -2i a. m 00 tH oo' CO •*^ T-t e* >a II ffi c s o a 1 1 1 i>) o g o O c ea . ea T-l ^"^ «5— > §?- J gj ^ ^ ■SM -1J r ■^ L - ... 1 3 f= c .2 •2 S= .« c > 4 .^ T^ < — 1 -^e N .-. « ... 5« ri -v: i -re ■r> .^■: .2 o a S'Sj N -:^ c* — -^ 1 •-< ci _. 1 —J t^--^ l« 1 .^ •M '-m _u _. "S o . I J> «'i I> ^^ «? J CO 1 tH >-< ;^03 1 •M i~ L -^ .'^^ S ■ ■2 c Co . £-a CO — J ■»ii ^ ^ tH §-^ g« iz; 1 . ■* 1-" 3! 3! 5! 1 kj g ■^ g ,| CO lH ? 1 C4 &. c -S !< 1 i o t; "^ <^ ^ o ( cviii ) The New 'World lias 54 genera which do not occur outside it ; 10(i are peculiar to the Eastern Ilemispliere ; while 7 are common to both. The pro- portion between genera and species is,- therefore, nearly the same in either hemisphere. Taking into account the origin of the genera as shown in the following pages, and leaving aside the truly cosmopolitan Celerio, there are 53 genera of American extraction, against 113 of Old World origin. Some significant discordance will be found in the numerical distribution of the genera and species when comparing the two tables (pp. cv. and cvii.). The Western Hemisphere has one and a half times as many species of Aclterontiinae as the OKI \Vorld, but they are more uniform, belonging only to 16 genera; whereas the s]iecies of the Eastern Hemis])here are groujied in 27 genera, representing many more ste{).^ in the development of the AcheroHtiinae than do the Trans- atlantic forms. On the other hand, the 26 Old World species of Sesiinae belong only to 3 genera, while America possesses 22. In Choerocampitme America, with its 3 genera, of which 1 is cosmopolitan {Celerio), stands far behind the Old World, which has 12, though the proportion of the species is 53:85. The difference in the pro]>ortion of tiie species and genera is instructive in another resjiect. Looking at tlie table of genera, it will be manifest that the student of geographical distribution who bases his conclusions on the Sesiinae alone would have only one centre of development, which has given off a few branches to other countries ; tliere would be no tpiestiou of " Regions." If he took the Vhoerocampinae as tiie basis of his zoogeograjihical division of the earth, Aethiojiia and Indo- Australia might become Regions and the rest of the globe appendages of them. If lie based his exposition on the Fhilampelincie, Africa, Indo-Australia, and Nortli America would rank as Regions, and South America, Europe, and temperate Asia ai)pear to be zoogeographical appendages of them. While in tlie case of tlie Arheroiitiiiiae there would be four Regions. It is evident fiom this that the division of the globe into zoogeographical Regions is different according to wliicli group of animals is taken as the sole basis of the division, or, in other words, that the division which is correct in one group of animals does not necessarily apply to every other group. That the Neotropical, Aethiopiau, and Oriental Regions are nevertheless natural zoogeogra[)hical districts in all larger groups of land and f'lesh water animals is not to be wondered at, since these Regions comprise each the subtropical and tropical ])arts of a continent and its satellites of islands. America north of Mexico and Europe and temperate Asia are by no means so well characterised as, and co-ordinate to, the other three Regious. In tlie above tables of distribution we have included the Papuan Sjthingidae in the Orieutal fauna. Australia, New Guinea, the islands farther east, and those westward to the Moluccas have 26 genera of Hawk Moths, of which 0 liave not been found elsewhere. The Palaearctic Region has 7 peculiar genera out of 3(J. These figures alone would show, therefore, that the Papuan countries were as much entitled to the rank of a Region as the northern temperate Old World zone ; or, on the other hand, since the proportion of genera peculiar to ( fix ) eacb is so very small, that iieitlicr the Papuan iinr the I'al-iearctie ooniitries can be considered to be a Region in the distribution of the S/j'/in(jida\ But if we sink them both to the rank of a Subregion of the Oriental Region, witli wliich the Palaearctic conutries have 18 genera and Papuasia 19 in common, the most characteristic feature of the temperate Old World would be swallowed up ; the Oriental Regiim would receive foreign elements, and hence the relation of the faunae of the Old and New Worlds be very ranch obscured. The Palaearctic countries being particularly important as a connecting-link between the Old and New Worlds, having many atHnities with North America, as we shall see further on, we give tliem the rank of a Region, in spite of the close relationship with the Oriental famia. Here again we see that the purely numerical consideration of a fauna does not give its true position in respect to geographical distribution. In order to understand the composition of the fauna of a district, the geographical origin of the various members has to be inquired into. A point essential towanls this object is manifestly the knowledge of the relationship of the various species. And as the grouping together of species into genera is meant to be a grouping of species of the same origin, it is further evident that the correct composition of the genera and their correct position in the classification are necessary premisses for sound zoogeograjihical research. E.xponents of zoo- geography who rely on catalogues — in which the distribution is far more erratic than in Nature — having themselves not enough knowledge of the animals to detect mistakes, cannot go beyond the numerical stage iu the treatment of the distribution, and, if they do, will as often be wrong as right in their conclu- sions, constantly finding in the faunae discrepancies and similarities which are nothing but the result of an insufficient study of the animals iu question. If we took, for instance, the classification of the Sphingidne in Kirby's Catalogue, Butler's Ref/sion, Druce's Ileterocera of Central America (iu Biol. Centr. Anier.), and Hampsou's Motlts of India as the basis for the exposition of the geographical distribution of these insects, we should find a far greater similarity between the Neotropical Region and the tropicus of the Old World than there really exists, numerous genera being said to occur iu both tlie Eastern and Western Hemispheres, while closer inspection proves these genera to be quite " unnatural " (heterogeneous), and the various heterogeneous components to belong to different genera ])eculiar either to the New or to the Old World, Protoparce, " Diludia,''' '■'■ Triptogoii,^^ '■' ^[mh/di/x" Pneudosphina-, Daraps'i, Pcrgesa, Macroglossum, '^ Aellnpus" " (7/oerocampa,"' Theretra, etc., etc., being names employed erroneously for mixtures of Old and New World Spluntjidar. Proceeding to examine more closely the composition of the iSi)hingid fauna of the various Regions, we take them iu the order as they come iu the tables of distribution on pp. cv. and cvii. The DTeotropical Region is inhal)itedby 33 genera, containing 'ZZ~ species. Two of the genera (Celerio and 1 terse) and one species (Celerio lineata.) are cosmopolitan. The remaining 37 genera may be classed in two groups : 22 which are confined to the Neotropical Region, and 14 which occur also outside it. We will examine the latter first. They are : Protoparce, ( <-x ) lli/loicu.Sj ('/d'(<')hy/ramma : I'rotiinihnhix : PscKdo.sphinx, Krinin/is, (ir.imniorlia, Pac/n/lia, Ejiistor, Cautetkia, I'eriyonia, Sesia ; Pliolus ; Xylophanes. In addition we must mention Sphinx, Pac-li>/t<plnnx, sind ArctonotuH, which extend from North America into Mexico, and are not trnly Neotropical. Ili/Inlats lias many species outside the Neotropical Region ; but the section of the genns which is Neotropical does not occur northward of the Southern United States. ( 'lilncno- (/ramma consists of two si)ecies, one inhabiting the Atlantic Subregion of North America, the other Argentina.* Protopnrce has two truly Nearctic species {quiiKjiiemaculatus and scxt'i), besides a great number of Neotropical ones ; the same applies to Pliolus. The remaining lU genera are Neotropical, extending in one or a few species into tlie Southern States of the Nearctic Kegion, or going farther north without having developed Nearctic sijecies. Among the 22 genera confined to the Neotropics there is only one {Monnrda) which is not of Neotropical origin, belonging to the branch of the Ambidicinae to which Cressonia, Sphinx, Amorpha, etc., belong, and of which the Oriental Callambnlyx is the most primitive genus. All the other 21 genera are trnly Neotropical. Hence it is clear that the only foreign element in the Neotropical Region is Monarda, and this is known only from Mexico. None of the other Regions have such an almost exclusively autochthonous Sphingid fauna. The Neotropical genera of Ackerontiinae are all developments from the same stem, of which Cocijtius, Amphimom, and Protopnrce are the most generalised divisions, coming near the Old World genera Xanthopan and Mec/anoton, which are, however, still lower in organisation than Coo/tins. None of the New "World Acherontiime have preserved the organ of friction of the claspers and eighth abdominal tergite, which a number of the Old "World genera have retained. The 4 Ambulicine genera of the Neotropical extraction are Protainbuhjx, Amplypterux, Orecfa, and Trogolegnnm. The last two are clearly modifications of Amph/pteniti, and tliis and Protam.balyx are near allies of the Oriental Compsoqene, which is the most primitive of all recent Ambulicinae. It is worthy of note that both derivative genera occur in the border districts of the Neotropical Region, Orecta in the South and Trogoleynuni in the North. The Scviinae are a special feature of the Neotropical fauna, the very few genera (:i) occurring outside it being derivations from the Neotropical stock. The Philampelinae are very poorly represented ; the two genera occurring are confined to the New World, and closely allied to one another, Tinostomn being found on the Sandwich Islands only, while Pholus is widely distributed over America and contains many species. The Choerocampinae are rather uniform in develoj)ment, there being only three genera, of which one is cosmopolitan. The other two are truly Neotro])ical, Xylophanes with fifty species and Phanoxyla with one, the sjiecies of Xi/lopliatiea which are found in the Nearctic Region being Neotropical ones extending northward. Phanoxyla is a specialisa- tion of Xylophanes, and this is a derivation from the cosmopolitan genns Celerio — /.('. from an ancestral genus of which Celerio is the nearest recent representative. * The locality Costa Rica is doubtful. ( exi ) As !i niattpi- of course tlie Ncotropiciil Sj,jn)uji(l<tr juv not iniitoniily distributed witliin the Region, though tlie greater ]>n>iioiiion is found nearly throughout the tropical part of South and Central America. The commonest species of which large numbers of individuals seem to occur everywhere are : lli'ise ciiHinlata, Coci/titis duponchel, Protoparce sexta and rustica, Protambah/x afrigiliH, Amplifjjterus gannasdis, Psendosphin-v tetrio, some Erinn.i/is, Panlii/lia Jims and ri'sumeng, Epistor liKjuhris, Pergem Ittsca, Xglophan's pluto, chiron, and tmsa. The temjierate South of the Continent, which is much poorer iu species and genera than the tropical districts, is cliaracterised by the s])ccialised genera Neogene and Orecfa, which, tliough penetrating into the tropics, do not extend very far northward. There are further in this southern district, which readies on tlie east coast to Espirito Santo and in the La Plata region to Tucuman and Paraguay, a number of species which have not been found farther north, besides some subspecies. We meet there with I'rotopatec bergi, P. tucunmnn, P. (Htfiss'i petuniae and diffissa (/ij/imi, P. ma>id//ro>des, Hi/loiats mauva and 'ptsticiac, several Ni/cergx, Chlaenogramma nKdafa, Alcxron j/roininens, A///ojd//iiii'.s depuiseti, X. xijlohotes, X. scknitsi, X. pistacina, X. isaon, X. aglaor, Hemeroplanes grisescp.ns, Celerio euphorbianim, Pliohis sateUitia analis, Plnyo japijx discri'pans. Chili has only two species of Hawk Moths — Celerio (duiri and Protoparce sexta caestri. Ilerse cingulata may also occasionally put in an ap})earance, but is not recorded. The discrepancy in the Sphingid fauna of Central and South America, though many of the apparent differences in the comjiosition of the fauna will most likely disappear on further research, is not inconsiderable. No species of Fjiryijlottis has as yet been found in Central America, notwithstanding the species not being of rare occurrence iu Colombia; Prutoparcc st//arti, triniucnla, sautdtft, pctiddae, Ilyloicxs maurd, JHsficiac, Pliolus cissi, obliqit/it<, several Pro- tambtdg.r and Ampli/teriis, and a number of Xi/lophnnrs, Xi/ceri/x, Madori/x, etc., have not turned up in Central America ; while Protoparce dilncida, occulta, KCgq/iiplex, muscosa, coralline, lanuginosa, crocala, several Hijloicus, Pholus fi/phon, Amplgpterus donijsa and ypsHoii, and some Xi/lop'nines are not known from South America. Such differences as these are found everywhere between districts of wide extent, areas differing physiograpliically like the Andes and Brazil being naturally more or less different in the composition of the population, whether there ever has been a geographical barrier or not between the districts. The lesser Antilles are rather im])erfectly ex])lored. The few species of Sphirigidae known from there are widespread. It is, therefore, the more remarkable that the common I'mfopan-e nisficu, which ranges from Argentina northward to North America and tln^ larger Antilles, has developed into a subspecies on the lesser Antilles as it has on the Galapagos Islands, while it does not vary geographically on the Continent and on the large West Indian islands. The occurrence of Protoparce rustica harterti on the lesser Antilles (inclusive of Bonaire and Cura9ao, but exclusive of Trinidad) is an indication that more species may be represented by special races on these islands. ( .•xii ) Si'veral nf tlic Ncot rojiii'iil Spliiix/idde were first descriljed i'rom tl]e small islands of St. Christopher, Antigua, and St. (!ruz. We have not seen any specimen from there. The four larger West Indian islands, ( !nlia, Haiti, Porto Rico, and Jamaica, are not very difl'erent in their Sphingid fauna i'rom Central America, and agree closely witli tliat of tiie Bahamas and Florida. They form together a West Indian dis- trict, cliaracterised by the absence of many Neotro))ical species and the occurrence of one genus and several species peculiar to them, besides several subspecies. Apart from Florida, which has a number of North American Sphiiujidae, there is no Nearctic element in the district. An exjiloration of the Lejiidojiterous fauna of Cuba and Haiti is a desideratum, the material in collections being very scanty. The species (occurring are distributed as follows, those j)eculiar to the district lieing marked with an asterisk (*). The subspecies in brackets are given in the list to show the distribution of the species outside the district. The North American species occurring in Florida are left out. C3 s s s i Occurrence outside the •r; 1 .g ■5 0 0 '3 s 03 district. E m 6 w eu ^ Acherontiinae. Herse cinguluta X X X X X X N. and S. America. Cocytius clueutius X X X X Neotropical R. — duponchel .... X X — X „ ,, * — antaeus untaeus . X X X X ( medor) .... — Neotropical R. *Protoparce sexta januiicensis . X X X X X ( sexta, etc.) . X — Nearctic and Neotr. R. — afflicta X X X — Amazons. — rustica rustica X X X X X X Neotropical. ( harterU) — Lesser Antilles. * — broiites hrontes — — — X * cubeiisis .... X — X X X — *Naiuioj)arce poeyi poti/i . — — X — — X ( hateriu/i) — — — — — — Yucatan. Ambulicinae. Protfiiiihiili/j- sir/ij/l/s _ _ X X X X Neotropical R. * — carteri .... X X — Amph/pteno: gaiiiiasrns . — X — — X Neotropical R. Sesiinae. Paeniloaiihiiix letrio . X X X X X X „ „ * hngiKithuis rinidud riiiinsa — — X X X — ( hwlitiis) Central Am. ( pediUiHthi) . Venezuela. * — ctDic/rdtiila/ix X Erinin/is ithipe . X X X X X X Neotropical R. ~ luxaaiixi f. iiteiiaiHic X X X X Central America. - ello . . . X X X X X X Neotropical R. — oenotrus X — X — X X — crameri — — X X — X ,, „ ' - ohsciira X — X X X X „ „ — (loiit/ngnnis . — — X X — — V 11 * — guttidorix . — — X X — — Grammodia caiciis . X X X Neotropical R. Puchylia ficus . X X X X X X * — syces instdarin — — X X . — X ( syces) . — — — — — — Neotropical R. — resumens X — X X — X ,1 1. ( cxiii ) Sesiinae — continued. *.]failiiri/.r pseudothi/reus . llrijivi-opUines calliomenae EpixliD- liiiiubrin hi(iuhi-is. * latipeniiin — or,,pctf *flliiuuit(iidcs undata *C(iiihthia noctuifuriiiix * — ijrotei .... *Fcrigonia divisa — luscaf. luscci f. hiterriipta . *- hfihmei . * ■ • — JIIIIKllfPllSlS . * — (i/iiiireacetis . Eitpi/rrlw(/lnsstiin sagni *Sesia hlaiiii * — tantalus zoiuita . ( tantalus) ( — ■ — clavipes) — fadus .... Philampelinae. *Phulus salellitia satellHia * imsticatus ( licaon, etc.) . * — strenua — riiis i-itis * ht'speridum — fascintus — lahruscae Choerocampinae. Xijlophiini's phitn * — irrurafii * — gmiiHiir/ii . * — rhndocera . * — piircus povcus ( (•(iiithientiil/s) — chiron rh/rou nechus . — tcrsa . * — suana . * — rohiiisoni *Celerio calverlepi — Vtneata tineata ( livomica) ( livornicoides) Occ'urrcnco outside the disti'ict. S. America. Neotropical R. Neotropical II. Neotropical R. Neotropical R. S. America. C. and S. Amerii a. Neotropical R. Neotropical R. Neotropical R. Neotropical R. Neotropical R. Neotropical R. Neotr. and Nearctic R. Europe, Asia, Africa. Australia. Among the 22 genera and 60 species occurring are 1 genus, 20 species, and 11 subspecies peculiar to tbe district. The genus Himantoides is confined to Jamaica ; it is closely allied to Cautethia, which is not found on Jaiuaica, hut occurs on the other islands, as well as in Florida and Mexico. While Torto Uico, Haiti, and Cuba do not apparently ditTer except in the number of species, Jamaica has some specialities restricted to it. Besides Himnntoides undata, we find in Jamaica Protopurce bronfes brontes, Epistor htgubris lafipennis, Perifjonia jamaicensis, P/iolus vitis /lesj'i'ridi/m, Pholus salellitia satellitia, h ( exiv ) Xylojilianoi r/i/ro/i rZ/iro//, which iiro t'ither absent i'rom the other islands, or represented by other subspecies. The Lepidoptera of the Galapagos Islands are almost entirely unknown. The fauna is very poor in Butterflies ; but the Moths, especially the smaller kinds, seem to be fairly well rejiresented. Dr. Holland, when describing the Gala[)a<;;os subspe;'ies of Protojiarce rmdea* said tliat he had received larvae of five species of Sphimjidae. In the Tring Museum there are six species of Hawk Moths from this group of islands : Hersc ciiKjuUita, Protoparcc rustica calapagensis, a new species of the same genus, a new subspecies of Erimujis obscura, E, ello, and Celerio lineata. Some species of Xi/Iojihancft, and one or the other genus of Sesiinae — Perigonia, for instance — must be expected to occur. The Sandwich Islands have six species of Sphuu/idae. Protoparce quinque- maculatus blackburiii is a well-marked race of the common Nearctic "Potato-worm." Tinostoma smtrvagditis is a very peculiar specialisation of the Neotropical genus PIiolus ; as yet only one imperfect specimen is known of this remarkable insect. The other three species belong to the cosmopolitan genus Celerio. They are Celerio lineata lineata, identical with American specimens, and Celerio calida and wilsoni, which have their nearest relative in Celerio annei from Chili and Peru. Tlie affinity of the Sandwich Island Sphingid fauna is, therefore, absolutely with America. The Spldngidae of the Nearctic Region offer, we think, more interesting points in their distribution and relationship than those of tropical America. As said above, quite a number of Neotropical species extend into the warmer districts of North America, some going regularly or occasionally to New England. If we exclude this purely tropical material, which is foreign to the Nearctic Region proper {Protambulgx strigilis, Pachglia Jicus and resumens, Erinngis alope and oenotrus, Pseudosphinx tetrio, Xylophanes pluto, Perigonia lasca, etc.), and deduct also Ilerse cingidata and Celerio lineata lineata, which occur everywhere in the New World, there remain 29 genera with 65 species. Among these there are again several which are Neotrojjical, but may be called inhabitants of the Southern States. Pholus vitis, labruscae, Erinnyis ello, Epistor li/gubris, Sesia /adus, and Xylophanes tersa have a similar position in the North American fauna, as have Acherontia atropos, Herse comokuli, and Hippotion celerio in the fauna of Europe. They are immigrants from the south. Two other Neotropical species have developed into Nearctic subspecies : Pholus satellitia pandoras and Protoparce sexta sexta. Among the rest there is one Holarctic species, represented in North America by one of its two subspecies {Celerio gallii intermedia). All the others are Nearctic, but all are derivations from tropical genera. We can group them in two sections, according to their origin — the one section containing specialisations of the Neotropical stock, the other the genera derived from the Old World stock. We add to the Neotropical section the Neotropical species which have settled in North America, north of Florida, in order to make the list complete : — • Proc. U. St. TV. Mils. xii. p. 195 (1889). ( cxv ) NEOTROPICAL SECTION. Acherontiinae. *Prot'iparce senla sexta. t * — ijuhiquemac. quiiiqiiemaculalus. *Chlae)iofjra m ma Jasin iiica rion. *Dolba hylaeus. *f>!ogramma hageni. *Cerutomia amyntor. * — undulosa. * — catalpae. *Isoparce ciipressL *Dktyosoma elsa. *Atrmts plebeja. *Hyloicus eremitus. * — eremitoides. * — separalus. *— chersis (3 subsp. and 1 in Mexico). * — vancouverennh. * — libocednis (2 subsp.). * — - perelegans. * — caiutdenxis. * — fraiwhi. * — kaliniac. * — gordius (2 subsp.). * — luscitlom. * — drupiferaruia (2 subsp.). * — dolli (2 subsp.). * — sequoiae. *Lapara coniferarum. * — pineum. * — bombyroides. * — fialicaniiae. Ambulicinae. No species. Sesiinae. Erinnyis ello. Sesia fadus. Epistor lugubris. * Haemorrhagia thysbe. * — gracilis. * — diffinis (.3 subsp.). * — brucei. Fhilampelinae. Pholus labruscae. * — sakllitia pandonis. * — acheiHOn. — vitis. Choerocampinae. X<jl<iphtiiics tfr-'ia. Besides Cdaio lineuta lineata and C. gulli inknnedin of doubtful origin, being members of a truly cosmopolitan genus. t The species and subspecies marked with an asterisk (*) are restricted to t'le Nearctic Region, or extend southward only in Jlexioo. OLD WORLD SECTION. Acherontiinae. ILrxe vingiilata. Ambnlicinae. *Sjiliiiix cevi^yi (4 sub.sp.). * — jiiinaicensis. *Ccdasyiiibol lis e.rc(ici-ala. * — myops. * — asti/Jus. *P(irhi/nphiii.i' modfxla (2 subsp. and 1 in Mexico). *Cre!>sonia juglaudis. (\[niiiinh( ory.r in Mexico.) Philampelinae. *^'l mpelocca Virsicohir. * — iiiyroii. * Da I'd psa pholus. *S/iheciidina abbutli. *DeXdamia inscriptum. *Arctotiotiis lucidus. ( — terlooi in Mexico.) *Amphio>i nessus. *Praserpinus gaurae. * — juaniki (2 subsp.). *— clarkiae. * — Jiavofascuita (3 subsp. ). * Euproserpinus phaeton. * — euterpe. ( cxvi ) II will lie observed tliiit all the AcliiTOntiiii.ae except Jlersr an; of Neotropical extraction, l)eiug derivations from I'rofoparcc. One of these genera extends into the Pahiearctic fanna (/fi/loir/ts), and has given rise to another genns in N.W. India ( 'l'h<(wiint'cha). The Atlantic Lapara is also a develop- ment of Hyloicus. The most generalised Ncarctic species of IFt/loicus are eremitoides and .^epantttin ; they are closely allied to those llyloicus which are pnrely Neotrojjical {geminim, Itigens, /star, etc.). We shall refer to this particniar relationship again when discussing the Palaearctic Spliingidac. The Amhulicine genera of North America (as well as the Mexican genus Monarda) are all of the Old World branch of the subfamily, which branch begins with Callamhulf/x, itself not very far from the most generalised Ambnlicine genns Coynpsogem, confined to the Indo-Malayan Subregion. Sphinx is common to North America and the Palaearctic liegiou, and has its older members in Central Asia (Sp/t/nx hindermanni and caecus) ; one of the two Nearctic species is strongly specialised {jamaicensis). Calasymholus is a derivation from Sphinx, and Pachi/siihinx and Cressonia (as well as Monarda) come also very near Sphinx and Amorpha (see pedigree of Ambidicinne). The only Sesiine genns of North America >vhich is not a recent immigrant from the South is llaemorrhagia with four species. It is distributed over the Palaearctic Region, and has one species in India and another on the Moluccas (Amboina, venafa). The six genera of Philampelinae peculiar to the Nearctic Region are not nearly related to the Neotropical Philampeline genus Phol/is, wliich represents an ancestral branch of the subfamily not occurring in the Old World, but are specialisations of Old World genera. Ampeloeca (rersicolor and myron) and Darapsa (ph(dus) are very closely allied to Ampelophaga of the Oriental Region and Pacific Palaearctic Subregion, being, like the Syrian genns Berutana, derived from it. Sjthecodina, Deidamia, Arctonotus, Amphion, Proserpinus, and Evproserpimis have nothing to do with Haemorrhagia, with which they are generally associated in classification. They belong to the Afro- Oriental tribe Nephelicae of the Philampelinae. North America has no species of Ghoerocampinae to itself, the three species occurring in the temperate districts being Xi/lophanes tersa, which is Neo- tropical, Celerio gallii, which is Holarctic, and Celerio lineata, which is cosmopolitan, the American subspecies C. lineata lineata extending over both New World Regions. A few of the Nearctic genera reach southward into Mexico. Dolba hylaeus is represented in Mexico by a jounger genus, Dolbogene, containing, like Dolba, only one species (Jiartwegi). One of the siibspecies of Sphinx cerisi/i, of Pachi/Hphinx modesta, and of Hyloicus chersis occur in Mexico. Uyloicus separatus of Kansas, Colorado, and New Mexico is found also in Mexico. Arctonotus terlooi is an inhabitant of West Mexico, which is geographically very closely related to the arid parts of the South-western States : California, Arizona, Colorado, and New Mexico. Not one of the genera of Old World extraction extends farther south than Mexico. ( cxvii ) The difitributioii of ilic Hawk Moths of North America illustrates a marked fouuistic division of the Continent into an Atlantic and a Pacific Subregion. The Atlantic Subregion reaches westwards to the Mississippi plains and includes part of Texas. It contains far more Neotropical immigrants than the Western Subregion, Florida offering special facilities for a northward migration of the Neotropical Splunt/iilae of the West Indies, none of the Neotropical species wliich occur regularl}- or occasionally in (ieorgia, Carolina, etc., being absent from Cuba. The genera characteristic for tlie Atlantic Subregion are thirteen: Dolba, (/I'ratoi/i/'f, Isogramina, Isoparce, Atreus, lAipara, Ampeloeca, Darapsa, De'idamia, Ampfiion, Calasi/mholus, and Cressonia, besides Sp/iecor/i/ui, which occurs, however, also in the Pacific district of the Palaearctic Region. The Atlantic Subregion is, therefore, rich in genera peculiar to itself, while the Pacific side of the Continent, from the Rocky Mountains to the Pacific Ocean, has only three genera restricted to itself {Arcfouotus, JiJiiproserjjuiiis, antl l)tct</osom((). Of these western genera, Arctonotus and Euproserpinus are derivations fiom Proserpinus, and Dlctijosoma from the Atlantic Ceratomia. The two Pacific species of Proserpinus and the only European one are more specialised than the two species which belong to the Atlantic Subregion. We have, therefore, the remarkable fact that what is characteristic for the Nearctic Region is essentially Atlantic. The Atlantic Subregion is the birthplace of the Nearctic Sphingid fauna. This is borne out by the two Holarctic genera S/ihinx and Ilaemorrluigia, which have together five species in the Atlantic and only two in the Pacific Subregion of North America. This distribution gives a hint how to explain the glaring contrast existing between temperate aud tropical America in the Sesiinae aud C/toerocampinae. One would have expected to find that, as is the case in Aclierontiinac, the large Neotropical stock of Sesiinae had given rise to a crop of Nearctic specialisations, and tliat the Choerocamjiiue genus Xi/lojihanes, flourishing with fifty species in the Neotropical countries, had some special representatives in temperate North America. Temperature and other atmospheric factors as such cannot be the cause of the deficiency in the North American fauna, since the Sesiine genus Ihiemori'hagia is almost exclusively an inhabitant of the northern temperate countries of the Old and New Worlds ; since, secondly, the Palaearctic Region has eight species of Choero- campinae to itself, besides Pergesa elpeyior, which reaches into India ; since, thirdly, temperate South America has several Sesiinae and Choeroeampinae which do not go far north into the tropics ; aud as, fourthly, the tropical plains as well as the temperate high mountainous districts of South and Central America are inhabited by Sesiinae and Choeroeampinae. However, if it is hardly possible that conditions of life are at the root of the discrepancy, the paucity of Sesiinae in North America and the al)S('nce of any indigenous si>ecies of Choeroeampinae from that Region must find an explanation in the geograpliical history of the Western Hemisphere. The explanation is rendered complicated by the fact that there are no less than seven purely Nearctic genera of Acherontiinae which are of Neotropical extraction, and therefore apt to contradict any likely ( cxviii ) cxjilaiiutidii wliicli takes cognisance only of the Sesiinae and Clioerocamjnnae. Tlip only ])lansil)le explanation is, we think, the following : Sonth America was at an early jicriod divided into an Atlantic and a Pacific Island (or Archipelago). Tlie Atlantic Islands were in connection with what are now the West Indies, and these were continuous with the Atlantic portion of North America, which was likewise divided, by a broad sea corresponding to the Mississippi plains, into an Atlantic and a Pacific Island (or Archipelago). This Atlantic country was contiguous with the land north and east, and hence stood in connection witli North Europe, perhaps as India and Australia are connected by an Archi- jielago. If we now assume that the Sesiinae and the ancestral Xylopkanes were almost entirely restricted to the Pacific Island of what is now South America, while the Acherontiinae were essentially inhabitants of the Atlantic Neotrojiical Archipelago or Island, or there represented by some generalised genera, like the early Protoparce, Ampliimoea, and Cocytius, it is evident that there was no barrier to prevent the Acherontiinae from entering the Atlantic Nearctic Islands, and crossing over the northern countries to North Europe and Asia, whereas there was little chance of the early Xi/lophanes and tlie Sesiinae reaching the Atlantic and northern countries. The only northern Sesiine genus {Haemorz-hagia) being a highly specialised one, and the tropical Old World genera GepkonoJes and Sataspes being still more modified, speak for the distribution having taken place in the direction indicated ; and the great percentage of elements of Old World descent in the Atlantic Sphingid fauna of North America corroborates the evidence adduced that there was a road of communication in the North between the Western and Eastern Hemispheres. The Atlantic Nearctic Sphingidae extend in Canada partly westward, and meet the western species in British Columbia and Washington. The Rocky Mountains and tlie ]ilateau west of them have no i)ronounced peculiarities in tlie Sphingid fauna. Tiie only distinctive features of this central district, which ranks as a Subregion in other groups of animals, are a species of Haemorrhagia. not found outside the area (hnieci) and the subspecies Haemorrhagia diffinis senta- The Falaearctic Kegion embraces Europe, North Africa, Syria, Central Asia, Persia and Afghanistan, North China, Amurland, and the three main islands of Japan. It is inhabited by 75 species of Sphingifhe, belonging to 33 genera: — 7 genera and 52 species are peculiar to the Palaeartic Region, occur also in the New World, is found in the Nearctic and Oriental Regions, occur also in the Oriental Region, occur also in the Oriental and Aethiopian Regions, occur also in the Aethiojiian Region, are cosmopolitan. 4 )) )) ^ I genus — 3 genera and 10 G „ 3 33 genera and 75 species. ( cxix ) The comparatively large nnmbi'r of genera occurring in the Palaearctic Region is acconnted for by the f\ict that many Oriental genera extend north- ward in single species. It is cnrions that this northward extension obtains here, as in the Nearctic Region, in the eastern district. The two cases are, of coarse, pnrely analogous. Tropical Cliiua being contiunous with Amnrland and India, and Formosa connecting with China the chain of islands running from Japan southward, tiiere is a convenient bridge from the Tropics to Japan and Amnrland. This close geograpliical connection between India and Japan ex- plains the contrast in the relation of the Palaearctic countries with the Oriental Region on the one side and the Aethiopian Region on the other. Numerous Oriental species of Hawk Moths range northward into Amnrland and Japan, while only tliree African ones reach Europe {Acherontia atropos, Deilephila nerii, and Ilippotion osiris), besides the Afro-Oriental traveller Ilippotioii celcrio, which comes to Western Europe most likely from the West (Joast of Africa, like Deilephila nerii. Tiie Saliara is an effective barrier between the Palaearctic and Aethiopian faunae. The Oriental species which are found as such in Japan, North China, Corea, or Amnrland cannot be counted as Palaearctic, Acherontia sU/x, Oxyamhuhjx ochracea, Clayiis bilineata, Cephonodes Jujlas, Acosmeryx naga, Macroylossum pyrrJiosticta, corythus, faro, passalus, Gurelca masuriensis, Theretra nessus, pinastrina, Uhyncholaha acteus, and also the Chino-Japanese Paruni colligata, are outside the proper Palaearctic fauna. These elements are com- paratively recent immigrants, and correspond to the Neotropical immigrants of North America. That the road by which they travelled is, however, an old one is proved by the numerous eastern Palaearctic genera, species, and sub- species which liave their nearest relatives in the Oriental Region. Five species common to both Regions are represented in each by a different subspecies : Psilogramma mcnejiliroii, I'hyllospkinyia dissimilis, Marumha sprrcliius, Ampelo- pliaga nibiginosa, and Lartgia zenzeroicfes. The eastern Palaearctic species, Callumhulyx tatarinovi, Marumha gaschkewitschi, janlwwsldi, maachi, Oxyam- bulyx japonica, schauff'elbergeri, Acosmeryx castanea, Rliagastis mongoliana, and Theretra japonica, belong to Oriental genera ; and Dolbina exacta and tanerei have near allies in North India in Dolbina inexacta and Bolbinopsis grisea, and Ilyloicus ealiginom m Thamnoerhu iiniformig. None of these Palaearctic species extend to Central Asia or Enro})e. Of the remaining nineteen genera found in the tem{)erate zone of the Eastern Hemisphere we may dismiss llerse, which is represented only by tlie wandering //. eonrolndi occurring all over the Old World, inclusive of New Zealand, whicli no other Hawk Moth has reached. Three are restricted to the Pacific district : Sphingulus, Kentochrgsalis, and Sphecodina. The last genus has only two species, one of them occurring in the Nearctic Region ; the distribution is in so far curious as the Palaearctic species is Pacific and tlie Nearctic one Atlantic. The genus is a derivation from the Afro-Indian Philampelinne. Kentochrysalia and Sphingulus, together with Dolbina, belong ( cxx ) to Mu' siimJl trihc Splii iKjiiUcar, vvliich has one more representative in North India { I lolljiiiopi^is) and tliree in Australia (Ttirae/iroa, Sy/ioi'f/ia, and Ho/j- liflcncnut'). Tlii'se seven genera are all that is Ici't of the tribe ; perhaps a few more s])eci('s than are known may be discovered in tlie Himalayan countries, China, and Australia. Tiie tribe is probably very ancient, and had formerly a more iniiform distrilmtidn in the Oriental IJegion, only remnants being now extant. The difference between the Western and Eastern divisions of the Palae- arctic ]{egion is enhanced by Central Asia, Asia Minor, and Europe possessing five genera not occurring in Amurland and Japan. Spliingonaepiopsis is Talaearctic, Oriental, and African, and is, like the West Palaearctic and Nearctic Proserpiims, a derivation from the African Pliilampelinae. Akbesia, found only in Syria, comes nearest to the Aethiopian genus Batocnema. Bcnitiina from Syria and Persia is a specialisation of Ampelophaga inhabiting India, China, Amurland, and Japan. Rrthera of Central Asia and Afghanistan has its nearest ally in the Oriental Region, and is of Oriental extraction. To these genera we must add the African Beileplula nerii and Acherontia atropos, which do not occur in the eastern parts of the Palaearctic Region ; and Theretra alccto, which has developed into a pale subspecies in Syria, occasionally occurring northward to the Caspian Sea, but rarely entering Europe. Tliere are now left to be discussed tiie genera ranging from Europe to Japan. Macroglossum has one species which is much specialised, and is nearest in characters to some Aethiopian ones, but may be a development from an Oriental sjjecies (of the group of helis). Alacroglossum stdlntarum occurs all over the Palaearctic Region, and goes southward into Nortii-western India. Ihjloicua is Palaearctic, Nearctic, and Neotrojjical ; it is of American origin, and has five species in tlie Palaearctic fanna, three being restricted to the Pacific side {(■(iligininm, obfrthueri, and crai^sistnga) and two occurring from Spain to Japan (pinasfri and ligustvi), but having developed into a western and an eastern subspecies. Mimas comes from the Afro-Indian stock of Ambulicinae ; it contains only one species, represented in the west and the east of the Region by a well-marked subspecies (Mimas filiae c/a-isfop/ti and Mimas tiliaf- tiliac). Amorplia is a development of the Oriental branch of Ambulicinae ; its two sjjecies overlap in Russia and Trauscaspia, A. amurensis ranging from Amurland to North Russia, and A. popidi in three subspecies from Morocco, Spain, and Great Britain eastward to Central Asia and Asia Minor. Sj>liin.:v is likewise a derivation from Oriental Ambulicinae {Callam- bulgu.-) ; it lias a|)pareiitly come into Europe by way of Afghanistan and Persia, Central Asia possessing the most primitive member of the genus {Sphinx kindermaniii), while Europe {Sphinx ocellata) and North America {Sphinx jamaicensis) have the most specialised ones. Europe and Japan have only one sj)ecies of Sphinx each, while Central Asia has two, Amurland three, and North America two, besides the derivative genus Calasi/mbolus. Haemorrhagia is, like Sjihiii.r, Holarctic, but has one s{)ecies in India {saundersi), and another, ( cxxi ) which deviates much from the rest of the species, in Amboiiia (vennta). II. fuciformis occurs all over the Palaearctic Region, and reaches into India ; it is se])arated into a western and an eastern subspecies ; Amurland, China, and Japan have three species {sfaiirUiigi'ri, heresowskii, and railiKiis) not occurring in the western district, and //. tityus does not go farther east than the Alai Mountains, while //. croatica is confined to South-eastern Europe and adjacent districts, ducalis to Central Asia, dentuta to Syria, and rubra to Kashmir. The genus Celer'w, which is cosraoj)olitan, gives a special feature to the Atlantic half of the Palaearctic Region, where it is more abundantly developetl than anywhere else on the globe. The genus is doubtless an old one, and had formerly more species in the tropics, the one each in Madagascar, Argentina, Chili-Peru, Cuba, and the two on the Sandwich Islands being remnants of a formerly less erratically distributed genus. The genus lias given ofl' two branches : RhodaJ'ra in Southern Africa, and Pergesa in the Palaearctic Region. Pergesa has four Palaearctic species, one ranging from West Europe to Japan, China, and North India (elpenor), a second extending from AVest Europe to Central Asia {porcellus), a third occurring in Central Asia (suellus), and a fourth in Amurland and Japan (cts/folde/ts/'x). A fifth species of Pergesa, closely allied to elpenor, is found in North India {rh'itlaris). The marked difference in the composition of the Sphingid fauna of the Atlantic and Pacific Subregions of tlie Palaearctic Region is illustrated by the following tables of distribution, in which the purely Oriental species which range into Japan without having developed into Palaearctic subspecies are not mentioned : — DISTRIBUTION OF THE GENERA OCCURRING IN THE PALAEARCTIC REGION.* »^« .mK «' P5 cs ex: pi c M § 0.2 P •2 K 'S ■s^ _u 'S. Descent - i o tn S' S ■^ s "S o of Genus. d ^ ci •S.S.S .5 .a s -^ tfi s e ■<£^ AnCCfe o <! 1^ ^ Acherontiinae. //►/■« .... X X X X X X Oriental. Arl„;;,„l>„ X X X X — — Afro-Indian. J'sil„f,,;i,niiai . — X X — — — Oriental. I/l/liiicil.s .... X X — — X X American. D'llhilKI .... — X X — — — Oriental. "^hritfochri/intl/^^. — X — *Sj,l,;>,y,d„s . . . — X — — — — „ Ambulicinae. * A /.■/,.'.•,;„. .... X Aethiopiau. (1.11/(1 III fill/ i/:r , — X X — — — O.iental. Mi'n-iiiiiha' X X X — Laiiifia .... X X *l/'/»'/.v .... X X Afro-Iniiian.' Ciill<iii,l,iilii.r . — X X Indian. S/i/uiir .' . . . X X X — J'/u/Un.ji/iini/Uc ~ X X — - ]] Tiie purely I'.ilaearctic genera are marked with an asterisk (*). Sesiinae. Hiiemnrrluigia Fhilampelinae. Lleilejjlula Anipelophaga . *Berutann . Acoanienjx *Rethera . Sphingoiiaejuojisis . Sjiheciidiiiii Proserp/itii.i Macrntjhisismii . Choerocampinae. Celerio . *Pergesa . Hippotion Theretra . Rhagastis ( cxxii ) -05 a a « Pi cc Ch \ Descent of Genus. American. Afro-Iudian. Oriental. Aethiopian. Afro-Indian. Cosmopolitan. Palaearctic. Afro-Indian. Oriental. DISTRIBUTION OF THE SPECIES OCCURRING IN THE PALAEARCTIC REGION.* .S.0 4S «<) CC Ah CB a ^ Acherontiinae. Merse coiwolvuli Aeherontia atropos . ■ — Styx Psilogramma mevephroii *Hyloicus ligunlri l/guslri * — — const ric la * — crassixliiga . * — j>in(istn pinastri . * — — morio * — calig'meus . * — oberthueri . *Dolbhia tnnerei * — exucln . *KeHtochnjstilis strecheri * — consiinUis * — sieveni *Sphingulus imis Ambulicinae. * Oxijd III hill i/.r jiipoiiica — $chiiuffelbergeri . *Marumba quercus — tperchmn * — jaiilcnwuLii . * — iniuicki * — gnsrhkewitsohi Langia zenzeroiiles . * The species restricted to tlie Palaearctic Region are marked with an asterisk (*). ( cxxiii ) v» .*^as « M o „ O y ^• a P5 "3 «3 cs ^ cS .2-? S ■3 Si S o he rt 5 ■§• •J "5. g 0 jj S 'S.o.| a t d ■<m:u ^K Oh o < •A jI Ambulicinae— foji<('n»«f/. *.\fiiiius tiliai' tll/iie ..... X — — — — — * ehrixtuphi — X — — — — *C((llambii/ipr tKliirimiri — X — — — — *Sj)hinx kimleriiiiiiini . X — — — — — * — caecus . — X — — — — * — ocellata X — — — — — * — planus . — X — — — — * Amorpha pnpiiJi X — — — — — * — aniurenisix X X — — — — Phyllosphiiifjia dissiniilh — X X — ~ Sesiinae. *Ihieiiivrrhu(/ia tityiis .... X — — — — — * — radians ...... — X • — — — — * — fuciforniis fucif. ..... X — — — — — * «^ffinis — ■ X — — — — * — beresowsh-ii ...... — X — — — — * — staiidingeri ...... — X — — — — * — croat/cii ...... X — — — — — * — rnbra t X — — — — ■ — • * — denlata ...... X — — — — — *- diicuUs X — — — — — Philampelinae. Deilephna mriiX X — X — — — .1 inpilophiKja ruhiijiiwsa — X X — — • — * Bi'futana kutschyi X — — — — ■ — *.Ai:i)snieryx castanea . — X — — — — *}i''tht'iv knmarovi X — — — — — *Sph/ng(in<n'piops/s (jnfgnn X — — — — — * — kiddjaensis . X — — — — — *Sphpcodina caudoUi . — X — — — — *Pnisi rj)inus pnisi rphvi X — — — — — *Marrinjhtssiin< stellataruin § X X X — — — Choerocanipinae. *(Merh, riipliurbiiir § X X X — — — * — gallii X X — — X — * — nicaea §...... X — X — — — * — zyyophylli ...... X — — — — — * — vespertilifi ...... X — — — — — ■ * — hippopluies ...... X — — — — — — liiieata ....... X X X X X X *Perge.'ia elpenor ilprwir .... X — — — — — — X — — — — ( fraterna) ..... — — X — — — * — ^JOJM^/f/S ...... X — — . — — — * — siielhis ....... X — — — — — — X — — — — HIjipotion eelei-in ..... X X X X — — (— osiri.i) X — — . X — — Thei-elra alerlo. ..... X — X — — — — X — — • — — ^Rliagastis ntanfjtd'utna .... — X — — ~~ 7 genera and 23 species occur iu the Atlantic Subregion. 12 „ „ .'U „ „ „ Pacific „ 12 „ „ 12 „ are conimoii to both Subregions. f Kashmir. % D. nerii extends into the Western Peninsula of India and Ceylon. § On the Oriental Region only in N.W. India. ( cxxiv ) The larger proi>orti<iii nf tlie genera and species occnrring in the Falaearctic Region are of Oriental extraction. Those of the Atlantic Subregion have come over Afghanistan and Persia, those of the Pacific Snbregion over China. Ten of the genera do not occur in the Oriental and Aethiopiau Regions, to which number must be added Uaemonluujhi, which is not of Tropical Old World origin, and OolhiiKt, which extends only into North India. Of tlie ten genera characteristic of the Palaearctic Region, two are of American descent {Ilijloietis and Iluemor- rliagia). When discussing the Nearctic Sphingulae, we drew attention to the fact that the trojiical lli/loicus are more generalised than the Nearctic ones. A similar phenomenon obtains in other genera. Tlie European Marumba quercus is more specialised tliau the Oriental species of the genus and than sperchius. The S}"rian Akbesiu is more specialised than the Aethiopian Batocnema, its nearest all}-. The youngest member of Cullamhuhjx is the Palaearctic tutarinovi, the Oriental rubricosa being the oldest. The most aberrant species of Sphinx is the Nearctic jamaicensis {=geminatus), and the genus Sphinx has given rise in North America to the specialised genns Calasi/mbolus. Of the various subspecies of Sphinx ce)-isi/i, the Mexican one is the most generalised. The American Pachijsphinx, Monarda, and Cressonia are yonnger members of that same branch than is the Palaearctic Amorpha, and this is much more specialised than tlie Oriental CuUamhulyx. The Syrian genus Berutana and the Nearctic genera Ampcloeca and Darapsa are derivations from the Oriental genus Ampelo- phaga. The southern Palaearctic sjiecies of llaeniorrhagia (croatica, dentata, rubra, ducalis) have kept the complete or nearly complete covering of scales to the wings, which the mimetic species have lost ; the Atlantic Nearctic Ilaemorrhagia gracilis and thi/sbe are also more generalised than dijfinis and brucei ; and Proserpini/s gaurae and jiiunita more than the Pacitic species of the genus and its derivative Euproserpinus. The most specialised species of the cosmoi)olitan genus Celerio are Palaearctic, and the two genera which have branched off from Celerio are South African and Palaearctic. Maero- glossum stellatarmn is in strncLnre and colour one of the most specialised species of this large Afro-Oriental genus. Orecta, inhabiting temperate South America, but extending beyond the Tropic of Capricorn, is a derivation from the essentially tropical genus Ampli/pterus. These cases show that the extra- tropical Sjihingidae are derivations from a tropical fauna, and that specialisation is generally highest in the districts farthest from the original country. There are soiue apparent exceptions which, on closer inspection, only prove the rule. Dolba of North America is more generalised than Dolbogene from Mexico ; Dolbina inexacta from North India more specialised than D. tancrei from Amurland and Japan ; Thamnoecha from North India more than the Palaearctic Ilyloicus, and Dolbinopsis from North West India more than the essentially Palaearctic genus Dolbina. In all these cases the southern Sphinx is a derivation from the northern, these insects being acquisitions from the Temperate Zone, existing only in the borderlands of the Trojiics. The Oriental Region is not larger in extent than the Neotropical and ( exxv ) Aethinjiian KeLrions ; but, lieiiig almost exclusively tn.pical ami very hetero- morphic, has the largest iuimb.;'i' of Sphingidac. The richest in Sp/t/iH/idaL' are the western districts, India and the larger Snnda Islands ; farther east there is an obvions decrease in the number of species and genera, thongh Australia possesses some genera of its own, and has, like the other Papuan countries, many species which do not reach India. ^fac■rn<Jlossl(m remains abundant in s))ecies almost everywhere except on the outlying islands. I'si/of/ramm't me/te/j/iro/i, and several species of Thrrefra, Uippofioji, and Macroylossum, are common from India to Queensland and tlie Sulomon Islands. The 251 species wliich are known belong to 04 genera. The distribution is as follows : — 3f3 genera and 226 species are Oriental. 13 „ „ 2U „ occur also in the Pa.laearctic Uegion. 6 „ ,, .3 „ „ „ ,, ,, and Aethiopian Regions. 6 ,, ,, 1 ,, ,, ,, ,, Aethiopian Region. 1 genus occurs in the Palaearctic and Xearctic Regions. 2 genera and 1 species are cosmopolitan. The proportion between purely Oriental and not purely Oriental genera and species is : genera, 36 : 28 ; sjiecies, 226 : 21. The percentage of pnrely Oriental genera appears to be small from these figures, owing to the fact of all the genera found also outside the Oriental Region being included in the 28, thongh many of them are actually Oriental, especially those extending north- ward into the Pacific Subregion of the Palaearctic Region, as detailed above. To arrive at the correct figures, we have to sepa'ate the indigenous genera, which may have sent out some species beyond the limits of the Region, from the genera of extraneous extraction. For the sake of simplicity we shall distinguish only between trojiical and nortbern temperate genera, leaving the discussion of the relation of the Oriental fauna with the Aethiojtian till we are dealing with the latter. The mountainous districts of North India and China have biological conditions favourable for species of the Palaearctic Zone. AVe find here several northern Sphingiclae and members of Palaearctic genera — for instance, Celerio euphorhiae and nicaea, Pergesa elpenor, Harmorrhagia fuci/ormis, Dolbinopsis f/n'sca, and Thamnoecha uniformis. PlnjUosphiiu/ia, Parum, and Bolbina reach from Amnrland or Japan to Assam or Sikhim. All the other genera common to the Palaearctic and Oriental Regions are tropical ones, tlie proportion of tro))ical to non-tropical genera of the Oriental Region being 58 : 6. Only 7 genera out of the 64 occurring are derivations from the American stock {Apocali/psis, Pseudorlolbina, Coenotes, Tliamnoecha, llaniiorrhaijifi, CfiphoHodes, and Sataspes). The Australian genus Coettotcs, of wliich only one sjjecies is known, is, like the North Indian genus Apocalgpsis and its derivative PseudndolhiiKi, a zoogeographical puzzle. Coenotes is very close to Neogene from Argentina and Brazil, and Apomlgpsis equally near Eurgglottis from tropical South America. The agreement between the ( c'xxvi ) respective Oriendil aiul Ncotrdiiicul geuera is so close that we must consider the relationship real. The rest of 51 genera are entlemic to the tropics of the Old World. Most of these genera are Indian. Their distribution shows that the division of the Oriental Region into a western and an eastern Snbregion is not better established in the Sphiiigidae than in most other families of Lej)idoptera. There is no line of division. The Papnan and the Indian elements, each taken as a whole, overlap. There are, however, two centres in tlie develojiment of species in the Oriental Si)hingid fauna : Continental India (inclusive of Burma and tropical China, of which the Spkinijidae are verj' imperfectly known), and trojiical Australia and the Papuan Islands. The Malay Archipelago, from the Malay Peninsula to the Moluccas and the Tenimber Islands, is exceedingly poor in species of its own, and possesses only one genus peculiar to itself, Giganteopalpus, of which the single sj)ecies is as yet found only on Java, Sumatra, and Borneo. All the other Oriental genera which do not occur on the mainland are found only in Australia, not even in New Guinea, all the Sphiiigidae of the Papuan and Pacific Islands eastward to Fiji and Tahiti being members of genera ranging to India and partly to Africa. Tlie island of Celebes has no single species of its own, and only two or three subspecies, which is very remarkable, considering that the island has a very distinctive population in many families of Lepidoptera and of other animals. Tiiough the distribution of the Sphingidae, over the Archipelago is not fully known, the following table will give an approximate idea of the extent of the Indian and Papuan species in the Archipelago. We leave out all the Indo- Chinese species which have not been found in the Archipelago, but add the Australian ones ; to shorten the table, we exclude also the species common to both Subregions. The Papuan species are marked with an asterisk (*). The column " India " embraces Continental Asia as far as it belongs to the Oriental liegion ; the " Lesser Snnda Islands " comprise the islands from Lombok to Alor ; the " Moluccas " the southern and northern Moluccas (Amboiua, Ceram, Burn, Batjan, etc.) ; "Tenimber" includes the islands between Timorlaut and Wetter; and "Oceania" means the islands south and east of the Solomon Islands: — a 1— 1 i a i a 1 Oh k i CO , i • s 1 e 1 i o o s < < 1 < 1 « o E _o .i Acherontiinae. *[ieri<efuscktla *— hu-tifem .... * — godarti .... Acheruidla sly.r — lachesi.1 .... Meganutnn ui/ct/pkumx . *Leucomuma hethia . *Coenotes eremophilaa *Tetrachroa edwardsi *Si/noecha marmorata *Hopliocnema inelanoleuca X X X X X X X X X X X X X X X X 1 X — X X X X X X X — X X X X X X X - - X ( cxxvii ) g t i 1 1 Borneo. Philippines. Java. CS c g z » 2 i s leuimber. Celebes. Moluccas. Aru. N. Guinea. Australia. Bismarck Aruh. Solomon Is. Oceania. Ambulicinae. Com psageiie pmuipus Oxj/iiiiibub/x hima . — substrigilk . • X X - • X X > -XXX < X X X X ' — - * — ii-ildei . * — xubdccllata . — semi/ervfus . * — doherti/i *Met((.mim<is austmlasiae . *0>('qwi>'a triangularis Chmit: euroa . Leucojihkhia Hnealu Polyplyrhus trilineatus . Marumba dyras * — ambiiiiiicus . — speciabilis . — ttmora Daphnusa ocellaris . Smerinthulus terranm — qiuidripunctatus . — dohrni Deymaptera olivacea CalUimbidyx rubricosa Sesiinae. *IJ(i''morrhagia vemita *Cfphoiiodes voodfordi * — janus . — hylas . *— titan . — piais . • X X > . X X ■ X — - • X X ■ X — . X X . — X - < X — X < X X X — — X — < — XX X X X X — X — — — X X — — X — — X — — XX — X — — — X — X — — XX X — — — X — - - . X X • X — . X X XX — X — — XX X — X X X — X — - < — — X — — — X — — X — XX — X — — — X * — lifurnsis Satiiyien infermdis . — tagalica — ribbei . Fhilampelinae. *Chromis heliodes *Deilephila protrudeiis Ampelophaga linigera Elibia dolichus Acosmeryx socrafes . * — miskini . X — . X — — — ^ — — - - - - - — — X • • — — X — X X X — . X X . X — X X — X X X X X — - — — X X * — splendens — malayami . — automedoH . — dohertyi — tiridates — variolosa . — mydon *Angonyx boisdurali . X — • — X XX — X — X — X -':- X — — XX — X — . X X • X X — —XX X X X X — — — ^ 1 Enpinanga vigens . — boriieensis . — lahuana *Cizara ardeiiiae *Nephele didyma ~- mbvarid . — X . — X — XX — -!- . X X — — — J ( cxxviii ) India. Malacca. Sumatra. Borneo. Philippines. Java. Lesser Sunda Is. Timor. J3 ■s . i s 4 M "2 ■ 1 3 1 fe- g ^ 1 ii| i Philampelinae - cnntirmed. S/i/i/iii/iiiKiiyi/ojis/x jiiii /■'iiri/j)ln-i/.r liliaiiii. . *- molnrrn . ' . (lifjiiiiliujialpiis mirab Afdi'i-di/loKsuiii av/cula — gil rails — vacilhi-.'s — (issiinUis — fmhsturferi * — calescens * — cdstfinriim . — pyn-hiistirta — tritijloiti/tiis . — alcedo * — ungues — sitkne * — sligina — heliiiiihila . *—m,l„s. ■;• '" . X — X X— — 'lis .— — X X — X— — • — — — — — X — — .X — X X — X X — . X X — — X — — X — — — — X — — — — — — — — — — — X — — — — — — — • X — X X X . X— X X • — — — — X X x_- . xxxxxx— — - — — — X — X XI— — — — X X— — — — — — — — — — — — X — — — — -— X — — — — — * — doherti/i . . . . — — ; ^ „ *— n,Vr,».s- . . . . *~nuhil„m . . . . *~ godrffnup . . . . ___ ~-r„n,'ga/>n„ X — X — — -~ glatiropta-a. . . .xx x— I 1 II 1 Mill II II X Mill < 1 III X X M II x> II X 1 1 > II II M II II 1 > * — joanni»/ — semlfascliita — aquila — Sylvia * — eichhorni . — mullifascia . — hemichroma — faro . * — meehi . . X X — X— - — — — — — X — — — • X X — X — X— — - ■ X— — X X X — — — •X X X X — X — . X — — — ___ — — — — — — — X — — — — — X — — — — * — phocitium . * — hiirtiensis . * — micacea * — .ydeiideiis . Choerocampin Celerio limata Hijtpolion echeclus — rafflesi * — bremms ae. . X — — — — — — — — ■ X — X X X X X — — • X — X — —x— — — — — — — — — — X — — X — — — — — — — — X — — X X X X — — — — — _ X — — — X — — — — — — — — X — — — — — — — — * — scrofa Therelra boisdtiva/i * — queenslandi — incur iiula . ' — indistim:ta . * — inornata *■ — tryimi — ahclo . . X — X X — X X — ■ — . — — — — — — X — — — — — — — X — — — — X — — X X — — — — siifftisa — lyceliis — margurita . — brunnea — turneri ^ . X X— — — X— — — ;|zE|=zZzEEz — X — — — — — — — li ( cxxix ) S c i i a 3 C i a 'E a. IS 03 > C3 •a a OS ;4 o s 1 3 H 1 3 < c '3 O .2 1 s 0 1 CO i Choerocampinae— mntiiinpil. Therrtra innignis + _-- _ X X — Jili)/nrhohiba acteii^ X X X X X X X — — X X — — — — — — Rhagosth nixlor — — — — — X — — — — — — — — — — — — aciitu ..... X X X — — — — — — — — — — — — — — alliomarghmliix . X — X X — — — — — — — — — — — — — Cfch<'iie)nt (Iff/rota . X X — X — X — — — — — — — — — — — — liiieona .... X X X X — — — — — — — — — — — — — — polliix .... ~ X ____ "" ~ X ~ ~ " ~ ~ ~ ~ f Andamans. Besides these IIT species,* there are 45 which occur from India to Papuasia. OdIv 26 s])ecies are peculiar to tlie area from Malacca to the Jloluccas, and of these 0:c)jfimbuh/x semi/ervi'nft and Sataspes rihhei are doul)tfally distinct. Judging I'rora the numlier of Papuan species described in this Revision, it is evident tliat further explorations will add substantially to tlie list of 58 Sphimjidue peculiar to the Papuan Region. Of the 45 species common to both Subregions, no less than 15 have developed into a western and an eastern subspecies. The genera restricted to Australia are T — namely, Livicomonia, Coenotes, Tefrac/woa, Si/noech(i, Ilopliocnema, Meta»ii)nas, and Coequosa. Of Papuan origin are probaldy also Cliromis and Angoiujx. It will lie seen from the above table that tiie cosmopolitan genus Ilerse has its headi^narters in Papuasia, where 4 species occur out of tlie 5 {comolcull is not in the list). It was originally most likely Papuan, and took in Australia the place of the Indo- African genus Acherontiu. Only llcrsa ronrohnli has wandered beyond the Papuan Subregion. The American llcrse cinqiihilit. may be a development from eitlier conroIniU or ijodarti. Tiiere is morphologically rather more evidence for (/oclarti than for cniicflh-uli lieing the ancestor of riiujalata ; bnt geographically ronvohuli has the advantage, since it does not require the construction of an Antarctic Continent to bring it into geographical connection with the range of ciiKjulutd. The scarcity of indigenous species in the Malay Archipelago, from Malacca to the Moluccas, and the i)ractical absence of genera restricted to these islands, and, further, the extension of Papuan species towards the west, and of Indian ones far towards the east, demonstrate that the Sphingid population lias come to those islands by comimratively recent immigration from both sides. If we call to mind the erratic distribution of Ci'phonodes //f//i/s, whicli is found (in tliree subspecies) in the Aethiopian Region, in India, Oliina and Japan, and again in Australia and on Flores ; of the two Oriental species of JVep/ieii', which occur in India, Java, and Australia, of Celerio lineata, which inhabits Continental Asia and Australia, but not the Archipelago; of the Spkingidkae, which are restricted to * Eunjj)tc)-ii.v shclfordi described on p. 813 is not included in this number. i ( oxxx ) Ansti'jiliii anil to Nurtli Imlia ami tho Pacific Palaearctic Siiliroo'inn * : and couple tiicse I'ucts with the absence of indigenous genera from the Archipelago (except tlie Malayan (i'xjanfeopalixis) and the scarcity of peculiar species, it becomes higlily ]irol)aMe that one and the same cause lies at the bottom of these phenomena in distribution. Tlie simplest explanation would be to assume that the Australian and Asiatic Continents were at an early period in the history of the Siiliinijuhic at least as closely connected as they are now, the connection allowing of the Asiatic fauna extending into Australia, and that later the inter- jacent district became so far submerged that the fauna was practically destroyed. With the separation of Asia and Australia by a wide gap, the condition for the evolution of special genera and species was given in Australia. The reason for the seven endemic genera of tropical Australia not having migrated over the Pajman Islands may be purely biological ; or it is possible that at least two of them {Lrucomonia and TetracJiron') really do occnr outside Australia, but liave not yet been found. The Indo-Malayan Subregion has twice as many species of Sphingidae as the Papuan Subregion, there occurring nearly 19U species, of which loO-odd do not extend into the Papuan countries. The cause of tlie contrast is easily perceptible. Papuasia is isolated from all Continents except Asia. Its entire Sphingid poi)ulation is Indian in extraction, perhaps with the exception of Coenotes, which has closer affinities to the Neotropical ISeogene. Since the greater proportion of the Spliiiu/idae are swift-flying and of wide distribution, and as a good many exist under adverse climatical conditions, and hence prove themselves to be remarkably adaptive, it ap])ears to us to follow that the Sphingid fauna of Australia would contain a good mixture of South American and African elements, if there had ever, within the history of Sphingid'te, existed an Antarctic (continent connecting these countries with one another, and suitable for the existence of Hawk Moths. India and the neighbouring tropical districts of Asia, on the other hand, have had an influx from the North and West, and, being mountainous, offered at the same time the necessary conditions of life for the maintenance of a great variety of species. It is especially the North East of India, and most likely Burma and (Jhina — both of which are very imperfectly explored — that harbour the largest number of Spliiiu/idac. South and West India are much poorer in Sphiiigir/ae, as is indeed the case with nearly all groups of Lejiidojitera. Though tlic absence of very many of the Himalayan species and genera from Ceylon and South India is compensated for to a certain degree by the appearance of some forms which do not extend to Sikhim, Assam, and Burma, we emphasise that the Western Peninsula and Ceylon have very few species of their own and no genus ])eculiar to themselves, the difierences being chiefly subspecific. The siguiflcance of this fact will be understood if we add that a few Palaearctic and African species reach into the Western Peninsula or the adjacent North Western districts {Mdcroglossxm sti-UatarHm, IhUcplnbi * The Australi.an PupU'io Ifostlwiics, wliich has its nearest relatives in Asia and Europe, may be mcniioncd in this connection. ( cxxxi ) nerii, Ilmmorrhagin fuciformis). Since there is no barrier to prevent these species from going farther east, it is manifest that the reason for their restriction to, and their occurrence in, Sonth and West India is biological. That is to say, the "Western Peninsula offers conditions of life suitable to those Falaearctic and African species, while the Central North and East of India do not. The differences between the West and North East are biological, most likely meteoro- logical. There is nothing whatever in the composition of the Sphingid fauna (and of other families of Lepidoptera) to indicate that there ever was another road of communication between the Western Peninsula inclusive of Ceylon and Continental Africa than South Persia and Arabia, though the road may have been less barren. The species which arc common to the Oriental and Aethiopian Regions, besides the before-mentioned Dcilepkila nerii, which is decidedly African, are only four — namely, Herse coiikoIvuU, CepJwnodes hi/las, Celerio lineata lirornica, and llipjiotion celerio. Two of them are wanderers, occnrring nearly all over the Eastern Hemisphere {Herse contohidi and Hippotio)i celerio) : Celerio linecUa livornica is also a wanderer, but does not go farther east than Continental Asia and Japan, lineata being absent from the Malay Archipelago and repre- sented again by a special subspecies in Australia {€. lineata licornicoides) ; and Cephonodes hjlas from the Aethiopian Region is snbspecifically different from tlie Oriental In/las. It is evident that, apart from the three wanderers, no exchange has taken place within more recent date. Nevertheless, there is a close affinity between the two tropical Old World Regions dating from a more remote j)eriod, evidence of which is fonnd in the genera Acherontia, ('lanis, Lencophlebia, Foli/pti/chus, Sphingonaepiopsis, Deilepkila, Sepliele, Cephonodes, Uippotion, and Tkeretra, which are found both in Africa and India. There are, Ijesides, several Aethiopian genera which are close allies of Indian ones — for instance, Pemba, J'oliana, Maassenia, Rhadinopasa ; and the Oriental genus Gurelca i s a derivation from Temnora, Rhodosoma of India comes close to Ihjpaedalia of West Africa ^ and the Aethiopian Atemnora is the prototype of Macroglossum, the tifty-one Oriental species of which are almost 25 per cent, of the total numl)er of species of Oriental Sphingidae. The African Sphingid fauna is as yet very imperfectly known. The pro- portion of undescribed species arriving from there is very large. We do not think the number of species and genera is much inferior to that of the Oriental fauna. There are as yet known u2 genera and 170 species,* of which 3S genera and 172 siiecies are peculiar to the Aethiopian fauna, a proportionally very large number. Ackeroittia utropos and DvilepJdla nerii, which extend outside tlie Region, but are also Aethiopian in origin, must be ailded, making a total of 174 truly Aethiopian Splnngidae. The se})aration of the African Continent into a Western, a Southern, and an Eastern district is not very distinct in the Hawk Moths. The faunistic differences between the West African forest region and the drier and more open districts of the Eastern side of the Continent are biological * //i/ijjutidit aurora, Uc^cribLil on p. 81-', i^ nut included in tliis muiibcr. ( cxxxii ) in origin, iind nut cuused by a geograijbical barrier having existed at au earlier epocli. Where the physiography of West Africa is similar to that of the Eastern districts, a similar fauna appears — as, for instance, iu Angola and the Hinterland of Sierra Leone and of the Gold Coast ; and where in East Africa there are extensive forests, insects of the West African type turn up. The genera I'ohiptychus and Tcmnora have very many species iu West Africa and few iu East Africa ; but allowance must he made for the fauna of East Africa being less explored than that of the West Coast. The South of the Continent, from Capetown to Benguella and the Zambesi, is physiographically and faunistically similar to that of East Africa. Tiiere are, however, some si)ecies of Lepidoptera which do not go far northwards into the tropics. Of Hawk Moths, Theretni capensis, Poli/pfi/chus mututa, and Oliijotjruplm juniperi may be spoken of as being characteristic of this area. Some other species {RhodaJ'ra opheltes and mnrshalli, Temnova iiatalis and nainaipia, Microsplunx puniilio, Leptoclanis pulchra, etc.) are too rare to be safely regarded as being restricted to the South. The differences between Madagascar and the Continent are very obvious. The Malagassic Subregion has 4 genera of its own (Fauogena, Lomoci/ma, Temnoripais, Mucmseiiia) and 20 species* and G subspecies restricted to itself, out of ~3 genera and 44 species. We append a list of the species found in the Subregion, those confined to it being marked with an asterisk : — Acheroutiiuae. Herse convolvuli Acheroiiiia alvopos . Coi'loilid fllloillollllil * — nuluiii .... *Xtiiitliiqiaii niiiri/aui jinudicla * Panotjena jamii'mi . * — liin/ens .... *l.n,inirilNl,l n,'i/nqi!ia Ambuliciuae. *Baloi-ii,enui cuqnereli. (2 subsp.) *Pxci{(lijch(ni!i (jraiididieri *Poli/jiti/i-huK iiiemider Sesiinae. Cephonodea liijhis finsixns * — iipim ..... * — tri)chilus .... * — kncogaster . Philampelinae Diilcplulit nrr/i *.\/aanse/Ha heydeni *Nc2>helc denso'i — coiniiia Occurrence outside the Malagassic Subregion. Old World. Africa ; Kurope. Africa. The other subsp. African. Africa. Africa ; Europe ; W. India & Ceylon. .\frica. * nijij/titio/t aurora, described ou p. S12, is nut include 1 iu tliis number. ( cxxxiii ) Philampelinae— CO. (///( '((■-/. ^Xfjihcle ut'iiupion uctitipimi *■ sliflica .... ^'J'f III Nora ijnimlidiiil * — iimrijiimla cuiiioraim . " — fuiiioxa jiechocvrl *~JM,lj„,lU . . . . ^ J'ciiiiior/jKiis liisli . *Sji/iinijtiiiiicjiioj)nis iibs( III IIS . I leihiioni weslenHaiiiii . *'MacivfflosisiiiH alliiaiidi . *■ — xoror ..... •■ — milvus .... '' — aexaloii .... •■ — /larlii/ceiHs .... Occuriencc outride the Malajjassic 8iibregion. The other suljsp. Africau. Tlie other subsp. African. Africa. Choerocampinae. *Celeri<i hitjiittata — lineala livoniica *EiicIiloroii iiieguera liironlnirt Btisioihia mccleii * — laticornis . */lij)piilWII lllltxclli . * — siiclacuruiH *— bullcri ... — bahamiiiae. * — ijerijoii — esiin .... *Tlii retra nrphKiin inteiisa Africa ; Europe ; Asia. The other subsp. Africau. Africa. African. Afi'ica. Africa. The other subsp. African. The Malagassic iSphingidae are thoroughly Africau, the ouly species iudi- cating an affinity with India not shared by the African Continent being llippotioii (jerjjon, which is close to the Indo-Australiau iv/ar, aud Maassetiia hei/deni, which stands intermediate between Nephele and the Oriental genus Acosmcri/x. The existence of four species of Cephonodes in the Subregion aud of live }[acro(jlosstim, while Africa has only one species of each genus, is a striking feature of the fauna, which, however, does not demonstrate a closer affinity with India or with Australia, but is the consequence of the Subregion being an Archipelago, and the wide separation of the islands offering a condition for the development of special species. The various species of Cephonodes and Xlacroylosmm, indeed, do not occur over all the islands, and most of them are ipiite restricted in range. The similarity between the Malagassic and Oriental fannae is much inferior to that existing between the African Continent and India. However, the few affinities which there are, taken together with the presence of African genera in India, and of Oriental ones in Africa, prove a rather close connection between the Aethiopian and Oriental faunae, each having iwofited from the other. There is no such agreement between Africa and South America. These Continents have no genus in common except the two ( cxxxiv ) cosmoiiolitau ones (/frn<e and Celrrio). The affinity which there exists between the Neotro])ic:al and Aethiopian Kegions is shared by India. Since low types of all tlie subfamilies of >Sphiitgidae, with the exception of the Sesiinac, occur in the tropics of the Western and Eastern Hemisi)heres, there is indeed a close affinity in this respect between the tropics of the Old and New Worlds. But the affinity constituted by the presence of allied low types has quite a difl'erent bearing on (|uestions of ojeographical distribution from that affinity which is established by the occurrence of closely related specialised genera and species. Let us take, for instance, the Acheroiitiinaf and Amhidicinae. The lowest African, Indian, and American geneia Xanthopan, Meganoton, Cocijthi», and Protoparcr are as nearly related to each other as the younger derivatives llyloicua, Ceratomia, Thamnoecha, etc. ; and the generalised Ambulicine genera Frotumbiih/x, AmjAypteras, Compsogene, and Oxi/ambuh/x are phylogenetically as closely connected as are the specialised genera Sphinx, Amorpha, Pachi/sphinx, Phi/Uosphingia, Cressonia, etc. The migration northwards and westwards into the Nearctic Region of the derivations from the ancestral Oriental A?nbtdicinae, represented by Amplypterus in the present epoch, and the extension of the derivations from the lowest Neotropical Acheroitiinae, Protoparce, or an extinct genus similar to it, north and eastward into the Palaearctic Region as far south as the Himalayas, and, further, the modification of either branch into a number of genera, have required a long time. Since not one of the younger derivations from the Oriental Ambulicinae and African Acherontiinae occurs in South America, and none of the younger developments of the Neotropical branches of these subfamilies are found in Africa or India, thougli the younger developments have become distributed northward as widely as stated above, it is obvious that there was no direct connection between the Neotropical Region and Africa and India during that period in which the evolution and distribution of the younger members of the two subfamilies took place. That is to say, the time which has elajised, since the African Xanthopaii and the Neotrojiical ('oa/tius, or the Indian Compsogene and the Neotropical Ampli/pterus, closely allied though they are, became geographically separated, is more remote than that which elapsed since the Nearctic and Palaearctic Regions exchanged a number of specialised genera. The distribution of the other subfamilies confirms this conclusion. The generalised Sesiinae occur only in the Neotropical Region, not one species being peculiar to any other region, though one single specialised— the most highly specialised — branch has crossed to the Palaearctic Region, and established itself in two genera in the troj)ics of the Old World {Sataspes, L'ejthonodes). The oldest genus of J'hilampelinac is again Neotropical {i'holns). It is sharply separated from the rest of the subfamily, and the only one occurring in the Neotropical Region, besides a derivative fmiu I'l/olas found on the Sandwich Islands {'rinostoma). All the other Pliilampclhidv are of Mid World extraction, and have parti}' migrated from the Eastern Ilemispheie into the Nearctic Region. In both the Seaiinac. and I'liildtupdiwie the gi;ogra])hical barrier between the Eastern and AVestern tropics must have been etfective at a very ( cxxxv ) early period of the history of Spliinyidae, since otherwise an interchange of the widely distributed genera, like the Old World Philampeli?iae : Dcilepkila and Nephele, and the New World Sesiinae : Pseudosphinx, Erinm/is, Pachi/lia, etc., wonld have been the nnavoidahle consequence of the presence of some kind of geographical bridge lietwecn tropical West and East. In the Choerocampinac we have again a similar kind of distribntion. The Neotropical Region has only the genus Xj/lophanes, besides the cosmopolitan Celerio and a derivation from Xjjlopfianes. All the eleven Old World generalised and specialised genera have been debarred from entering the New World, though they are quick flyers and mostly of wide distribution ; and the Neotropical genus Xi/hphain'A, tliougli very generalised, has no representative in tlie Eastern Hemisphere. Tlie aiSnity wliich exists between the Neotropical Region and the Tropics of the Old World is geologically remote, but phylogenetically close. Tl:e distinction between close blood-relationship in the faunae which means recent geographical connection, and close blood-relationship whicli indicates remote geographical relation of the countries, which distinction is clearly demonstrated by the distribution of the Sji/iingidac, is of the highest importance in studies of geographical distribution. The distinction is entirely effaced and the student led astray, if the genera are not studied and defined with the view of giving in the classification adequate expression to the phylogeny of the family. We divide tlie SpJiiiitji(/<ie into the following gionps: — A. Spliiiujidae ((semanophorar. — p. 3. Subfamily Ac heron t i i nac .— p. 4. Tribe Acherontiicae. — ]i. 4. Tribe Sphingicae. — p. 2T. Tribe Sphingulicae. — p. I -"14. .Subfamily Ambiilicinae. —p. l <■><>. B. Spkir/f/iilde t<i'nmnop/iorap. — ]>. 347. Snl.'family Sesiinae.— p. 34'.». Tril)e Dilophouoticae. — p. '■'i-i'i. Tribe Sesiicae. — ]). 37'J. Subfamily Ph i lam pcl i nae. -p. 4:."). Tribe Philampelicae. — p. 47.1. Tribe Nephelicae. — p. 4'.)8. Snbf\imily Chocrocam p i nae.— p. fi7-,\ II. SYSTEMATIC SECTION. Family 8PH1XGIDAE.— Typos : Sphiiu: ocdluta. Sphinx Linnc, Syst. Xal. ed. x. p. 489 (1758) (partim). S/>h!ii;iis Scopoli, Inlr. Hi-t. ytit. p. 413 (1777) (partim). Sjilit/inidis hsLireWle, Hid. Crust. Ins. xiv. p. 120(1805) (partim); Leach, in Brewst., Ei/hib. Eiifijd. ix. p. lao (1815). SphiiitihlcK Samouelle, Eiit. Cump. p. 243 (1«10) : Westw., Brit. Moth.^ i. p. 5 (1843); "Walk., Li.4 Lep. /n.s. B.M. viii. p. 76 (1856) (descr. of Sphing. of the globe) ; Butl., Tram. Zml. Hoc. Lnml. ix. p. 511 (1877) (revi.=. of Sphing.) ; Smith, Trans. Amer. Ent. Snc. xv. p. 49 (1888) (monogr. of N. Amer. Sphing.) ; Kirby, Cat. L,p. /let. i. p. 624 (1892) ; Hamps., in Blanf., Fauna Brit. Iml., .Voth.Hi. p. 65 (1892) ; Bartel, in Riihl, Grossschm. ii. p. 11 (1899) (descr. of Palaearct. Sphing.). Hpliiiifjiilae + Sesiidae Stephens, lUiislr. Brit. Ent., Ilaust. i. p. 104 (1828). Spliiiif/idi Boisduval, Iml. Mvth. p. 32 (1829) (catal. of Eur. Sphing.). Sphinijuahie Harris, in Sillim., Juurn. Sc. Art xxxvi. p.-290 (1839) (monogr. of N. Amer. Sphing.). Sphiniftiiae. + Orthidar, Swainson & Shuckard, Hi.st. Xat. Arran;/. 7».s. p. 101 (1840). " Sphingides'' Boisduval, Spec. Gen. Lip. Ilet. i. p. 1 (1875) (monogr. of Sphing.). Eff(/ gloLnlar or sliuhtl y flatteneil, almost smooth, without distinct excrescences, the strnctnre being microscopical. Larca cylindrical, or tapering in front, third and fourth segments often swollen; skin covered with setiferous granules, or smooth, the granules often present only in first stage ; segments riuged ; head rounded or triangular, apparently always rounded in first stage, sometimes first rounded, then triangular, aud again rounded in last stage {Ellema, see also Clanis) ; eleventh segment with horn, which is occasionally replaced by a buttou-like knob, horn longer in first than in later stages, often carried upright or leaning forward iu first stages and movable at will, bifid in first stages, or at least with two apical setae, varying in structure from being covered with large setiferous tubercles to being nearly smooth. Pupa : segments 5 and 0 of abdomen movable (three joints) ; tongue-case re;iching to end of wing-cases, or shorlened, often comjiressed, cariniform, enlarged biisally or projecting free, sometimes rolled in spirally ; sides of abdomen generally more strongly sculptured than dorsal aud ventral parts of sommites, stigmatical area mostly transversely ribbed or carinate ; cremaster very variable, bifid as a rule, ol'tcn with spines and hooks. Free iu a cell in the ground, or in a very loose cocoon on the surface between leaves, or beneath surface, seldom sjiun up on branch of food-plant (^CaU/omma). fimigo varying in lengtli of liody i'rom VZ to Sn mm., ;iud in length of forewing frum It) to 9U mm. (Sjj/a'm/onm'pioj'sis obscurus and Corijtius antacus). — Tongue varying from being several times as long as the body to being reduced to two small tubercles. — Pilifer clothed with bristles or scales, the brush even, or the bristles short at apex, or absent. — Genal process naked, mostly triangular, very • ifteu reaching to tip of pilifer. — Labrnm convex in middle, often raised into a rather prominent tubercle. — Palpus very variable in size and structure ; first segment with or without patch of sensory hairs on inner surface at base ; inner surface of second segment scaled or partly naked, sometimes excavate ; scaling B ( -^ ) at ;i|ii'.\ of lirKl, jsfgmciit cxUTiidlly .sninetiiiu^s loniiiiiL!; a kind ol' caNify ; third wifj^tiioiit. mostly sliort, always shorter than second is broivd, inxkcd and conical if very jironiinont. — Eyo varialde in size, naked ; no oceMi. — Scaling of lioad ofton raised to a crest, especially in forms with reduced head. — Antenna filiform, setiform, or clavate, }irismatically compressed or cylindrical, tapering at end, and mostly curved to a hook, sometimes pectinated ; eud-begment long or sliort, scaled dorsally, the scaling mostly dense and forming often a dorsal tnft projecting distad, or sparse and rongh, or almost absent. There are often long bristles on the end-segment, two at and near the tip, and several scattered over the surface of the segment ; the whole dorsal surface of antenna scaled, two rows of scales to each segment, but the rows quite irregular, }iroximal segments scaled also ventrally in a iQfr cases where the clubbed antenna is very thin basally. Ventral surface densidy clothed with fine hairs ; besides, there are some subdt)rsal and lateral sense-bristles, not obvious excc])t on distal t^egmeuts ; a single ventral mesial sense-cone on each segment, apical or subapical. c?: with one exception (li/iopalopsi/c/te), the segments impressed at the sides and provided basally, dorsally, and apically along tlie groove with a row of fasciculated long hairs, the proximal row of hairs of either side continuous ventrally, but the hairs always shorter in and near the ventral mesial line, the distal rows seldom extended to the mesial line. Sometimes the dorsal edge of the groove widened laterad to a process which bears the fasciculated hairs on tlie underside at the edges. In two other genera the basal and distal edges of the grooves are ventro-laterally produced into a process each — i.e. into two processes ou each side of the segment, the compressed processes bearing the fascicles at the upper and ventral edges. ?: without fasciculated hairs as a rule; however, there are many forms in which the segments resemble those of the <S, but the fasciculated hairs and the pectinations are not so long, and the grooves not so deej).— Sterna : parasternum of mesosteruum large, epi- and hy postern urn fused together, sternum and peristernum separate ; suturale of mesomerum swollen. — Abdomen : first tergite shorter thau second, a mere linear stripe in Macroylossum, tergites ::i to 8 (c?), or 2 to 7 ( ? ), and sternites 2 to 6 (c?, ?) or 2 to 5 (many ? ?) armed at end with spines, variable in number, size, and chitinisation, spines of tergites generally stronger than those of sternites, the latter often without sjiines. Some forms have the tergites spinose all over, while tlie spines are wanting in others, c? alwai/s with tuft of long (scent-distributing) hairs in groove above upper edge of sternile of second segment (steruite of first segment absent, or probably fused with that of second). — Legs : sternal (= frontal) part of coxae not carinate excejjt close to trochanter; meralpart of midcoxa reaching farther dorsad than sternal part, the coxo-meral suture slanting ; hinder edge of meral part of midcoxa, or of mid- and hindcoxa obtusely rounded, or carinate, or triangularly produced into a sharp tooth ; trochautiue distinct ; forecoxa in S often with strongly developed scent- organ ; tibiae simple or spinose, foretibia often ending in a thorn or claw ; midtibia with two apical spurs, which are sometimes much reduced ; hindtibia with two pairs, or the proximal pair absent ; the sj)urs occasionally spinose, in a few cases the spines long and arranged in a single row, comb-like ; tarsi with four ventral rows of spines, often with additional spines, especially on outer surface ; not rarely some or many of the spines absent, and some developed to prominent hooks ; ■1th row of midtarsus and 3rd of liindtarsus often basally developed into a jH-ominent comb; fifth segment with two (seldom four) long dorsal terminal bristles ; pulvilhis present or absent ; paronychium with two pairs or one pair of lobes, or vestigial ; ( - ) cImavs sini]ili'. — Frcimliim luid rctiiiuciiluiii iircsciit or iib.seiit ; an olungate suhbiisal ])at('.li of glossy, modified, scales on fdrewiiiy Ijelnw liidiind SRP, on liiudwing above before C. — Neuration ; fore wing; : no areole ; SC and SC-'^ before ajjex of cell, SG^ and SO' on a long stalk, free part of SG^ very weak and short, and mostly absent (individually variable) ; SO^ ending close to a])ex of wing, sometimes continned along edge and joining SG^ at apex, SG^ from SC* at about one-third the way from cell to apex, R^ from upper angle of cell or shortly stalked with SG''"° ; R^ from below centre of apex of cell, but always well above angle ; M^ before angle of cell, RP in or near middle of cell ; SM^ absent, SM' forming fork at base with SM- ; upper angle of cell more distal than lower angle. — Hindwing : G and SG- separate, connected with one another by a conspicuous slanting bar (= SG') in or near middle of cell ; SG- or (seldom) R' ending at farthest point of wing ; R' from ujjjjcr angle of cell, often from SG- ;