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The person charging this material is responsible for its return to the hbrary from which it was withdrawn on or before the Latest Date stainped below.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.