UCSB LIBKART CAMBRIDGE ZOOLOGICAL SERIES General Editor : — Arthur E. Shipley, Sc.D., F.R.S. MASTER OF CHRISt's COLLEGE, CAMBRIDGE THE BIOLOGY OF DRAGONFLIES (ODONATA OR PARANEUROPTERA) CAMBRIDGE UNIVERSITY PRESS C. F. CLAY, Manager ILouDou: FETTER LANE, E.G. Etlinburgf) : 100 PRINCES STREET ILontion: H. K. LEWIS AND CO., Ltd., 13G GOWER STREET, W.C. ip.tbo lorit : G. P. PUTNAM'S SONS Uombao, (Talcutfa anft iJlaliraB: MACMILLAN AND CO., Ltd. CToronlo: J. M. DENT AND SONS, Ltd. IBrlbournc ant SgUiuu: THOMAS C. LOTHIAN, PROPRIETARY, Ltd. STokgo: THE MARUZEN-KABUSHIKI-KAISHA 411 rights reserved Plate I Fi<;. I. Anvtropctfilid jKilricia Tillyard, $ {Aeschnidae, New South Wales). Fig. 2 Xaiinoplii/(i (iH-slntli.s Br., cT (Libdlnlidae, S. Queensland). Fig. 3. Rhjiothonis rcsplciuhns Sely.?. o (LibcUulidae, N. Queensland). Fig. 4. Rhinocypha quadrlmaculata Selys, ^ [Calopkrygidae, New Britain). Fig. 5. Ehiiiocypha tinda Ramb., ,^ {Calopterygidae, Papua). Fig. (5. Nturohasis chineii-sis I/inn.. o {Calopterygidae, Burmah). Fig. 7. MatroHo'ules cyanfipennis Foerster, rj (Calopteryijidae, Borneo). Fig. 8. Megaloprepus coerulatus Drury, o (Agrionidae, Central America), Fig. 9. Agriocnemis ruhricauda Tillyard, ^ {Agrionidae, N. Queensland) (All figures natural size.) (Original water-colour drawings.) THE BIOLOGY OF DRAGONFLIES (ODONArA OR PARANEUROPTERAJ BY R. J. TILLYARD, M.A. (Cantab.) B.Sc. (Sydney), F.L.S., F.E.S. Macleay Fellow in Zoology to the Linnean Society of New South Wales Late N.S.W. Government Research Student in Zoology in the University ot Sydney; formerly Foundation Scholar of Queens' College, Cambridge; and Assistant Mathematical and Science Master at Sydney Grammar School Cambridge : at the University Press 1917 Haec est mei LlBELLULA libri : TWN KOpAyAoON (})pONTl'zOa TOJN TYnToycooN TWN rOM0O3N TOJN nHrNYNT03N TOON Al'xMOJN T03N AAMnOMeNo:)N nANTo:)N KAAAinTepo:)N TT AN TOON AfpiOON. A living fiash of light." Tennyson, The Ttvo Voices. TO MY GOOD FRIENDS AND MENTORS IN ODONATOLOGY M. REN£ martin and Dr F. RTS, I DEDICATE THIS BOOK, IN MEMORY OF MANY KINDNESSES RECEIVED FROM THEM a'.i PREFACE rpHE principal object with which this volume has been written is to present as full and complete an account of the Biology of the Odonata as it is possible to offer in the present state of our knowledge of these insects. Thus the book is written primarily for biologists rather than systematists. More than ninety per cent, of the papers so far published on the Odonata have been chiefly systematic in their aim. The remainder are principally morphological, and form a valuable though somewhat small collection of facts on which a book of this kind must necessarily be based. During the last three years, I have fortunately been able to undertake a considerable amount of work on the internal anatomy of the Order, thereby not only confirming many points which were more or less in doubt, but also adding some new dis- coveries, and here and there rectifying errors. Most of this work has not yet been published, and will first see the light in the condensed form in which it appears in the present volume. It is hoped that the method of treatment followed in this book, by which the morphological, phylogenetic and physiological view-points have been correlated, in so far as our present know- ledge allows, will enable students of the Odonata to take up any line of research on this interesting Order with a full know- ledge of what has already been achieved. To this end, the Bibliography has been practically confined to those publications bearing on the biological study of the Order. It would have been impossible to include a complete list of the enormous number viii PREFACE of purely systematic publications on these Insects. Only those have been included which contain material of sufficient biological interest to require a reference in the text. All papers, however, dealing with the larvae, life-histories or faunal studies, have been included. With the exception of two or three which it has been quite impossible to obtain, all the works listed have been carefully read and studied by the author. References to the Bibliography are given within square brackets. A complete scheme of classification down to tribes is given in Chapter XIV. It is hoped that this, combined with a very full account of those external parts which are of use in classifi- cation, will make the book of value to systematists. For the use of British entomologists, a chapter is included dealing ex- clusively with the British species (Chapter XVIII). At the request of several entomological friends, a short glossary of biological terms has been introduced, for the use of readers who are not biologists. In a previous volume of this series\ the author, Dr 0. H. Latter, in dealing with the Insecta, selects three types of this enormous Class for study. These are the Cockroach, the Dragon- fly and the Wasp. The Cockroach has for many years served the purpose of a "type" for the Insecta, in all general courses on Zoology. Now there are three essentials which a "type" for general use ought to fulfil. Firstly, it must be obtainable without undue difficulty, and easily preserved for use. Secondly, it should illustrate as generally as possible the morphology of its Class. Thirdly, it should present a fairly typical life-history for its Class. The Cockroach completely satisfies the first two conditions. But it fails in the third, since it offers no illustration of the very important phenomenon of metamorphosis, which is associated with the great majority of the Orders of Insecta. It ^ The Natural History of Some Cotnyion Animals, by 0. H. Latter, M.A. Cambridge, 1904. PREFACE IX seems, therefore, to be in the highest degree advisable, that Dr Latter's lead should now be followed in all Zoological Laboratories, by the selection of two other Insect "types," which should illustrate respectively the phenomena of incomplete and complete metamorphosis. Of these, one should undoubtedly be the Dragonfly, the other either the Wasp, Bee, Blow-fly or Butterfly. The metamorphosis of the Dragonfly is easily observed and easily understood, yet it is as striking as that of any other Insect. If we add to this the further advantages that both larva and imago (but particularly the former) provide dissections of com- parative ease and unrivalled beauty : that both closed and open tracheal systems can be studied in the one insect: and that the adaptations to the aquatic way of living in the larva are of an interest not to be equalled anywhere in the Insecta, it will be seen that a very strong case exists for the recognition of the Dragonfly as a second biological "type" for the Class. If this book helps towards the general acceptance of this view, by pro- viding a text- book suitable for the use of zoological students, the author will feel more than repaid for his labours. Little need be said as to the design of the book. I have made original text-figures in every case except where the material was not available. In such cases, the authorship of the figure is gratefully acknowledged, but the lettering has been altered to suit the text of the book. Of the original drawings, figs. 23, 30, 33, 35, 37, 38, 125, 128, 139, 146 and 154 have already appeared in my contributions to the Proceedings of the Linnean Society of Netv South Wales [167, 169, 171, 172, 176], while figs. 79-83 are taken from a paper contributed to the Journal of the Linnean Society of London [179]. Almost all the specimens studied and figured have been collected around Sydney. Readers will therefore not fail to discover a very strong Australian "flavour" in the book. Those X PREFACE who know how rich and varied the Australian Dragonfly-fauna is, will look upon this rather as an advantage than otherwise. While care has been exercised in selecting, for general description, well- known genera of wide distribution, yet there has been no hesitation in making use of the more archaic Australian forms, when these latter could throw new light on the phylogeny of the Order. Throughout the book, the names Gotnphus, Aeschna, Calopteryx and Agrion are used in the Selysian sense. In common with practically all other students of the Order, I regard these names as nomina conservanda. Their alteration would not only lead to immense confusion, but would in a sense vitiate the whole of the classical or Selysian literature. Odonatologists are more concerned than anyone else in the retention of these names, and they will never consent to their arbitrary removal by committees formed of men who know little, and care less, about the require- ments of this very special branch of entomology. The MS. of this book was completed on March 19th, 1915, and was forwarded to Cambridge in July, 1915, with the Bibliography brought up to that date. During the first correcting of proofs, opportunity was taken to bring the work on the Respiratory System (Chapter IX) into hne with the recent researches of Calvert and myself; but it was not considered practicable to add any new records to the Census and lists of genera. The Additional Bibliography on p. 373 was added in July, 1916, when the first proofs were returned to Cambridge. The sending of the original MS., figures and plates, three sets of galley-proofs and three sets of page-proofs (involving in all a total of some fifty packets and letters), to and fro between England and Australia in these troublous times, without the loss of a single item, is an achieve- ment which I cannot let pass without offering my grateful thanks to all who were responsible for it. In conclusion, I desire to acknowledge with very sincere thanks the help received in the preparation of this work. My PREFACE XI thanks are due to my wife, for the execution of figs. 25, 167, 184, 185, 187, 188, and for her valuable help in indexing; to Mr F. W. Carpenter, M.A., of Sydney Grammar School, for much arduous work in the field, for advice and help in the preparation of wing- photographs, and for the two beautiful photographs reproduced in Plate IV; to Dr F. Ris, of Rheinau, Switzerland, for the loan of publications from his library, and for much useful help and criticism; to M. Rene Martin, of Paris, for an account of the Ethiopian fauna ; to Professor P. P. Calvert, Ph.D., of Philadelphia, for an account of the fauna of the New World, and for other in- formation; to Dr F. F. Laidlaw, of Uffculme, Devon, for an account of the Oriental fauna; to Mr R. A. Muttkowski, B.Sc, of Madison, Wisconsin, for information on the Nearctic fauna; to Mr K. J. Morton, of Edinburgh, and Mr Herbert Campion, of Ealing, W., for specimens of, and information concerning the British and European fauna; to Dr S. J. Johnston, Lecturer in Zoology, University of Sydney, for assistance in drawing up the glossary; to Mr B. G. Engelhardt, of Hornsby, for geological information; and finally to those many other correspondents, whose exchanges of publications and specimens have indirectly helped me in the completion of this work. R. J. TILLYARD. "KURANDA," Mount Errington, Hornsby, N.S.W. September 'Mh, 19 IG. CONTENTS CHAPTER page I. Introduction ...... 1 II. The Imago — External Features and Skeleton 9 III. The Wings 38 IV. The Larva or Nymph .... 67 V. The Alimentary and Excretory Systems . 100 VI. The Nervous System .... . 120 VII. The Sense Organs ..... 137 VIII. The Circulatory System .... 157 IX. The Respiratory System 166 X. The Body-Wall and Muscles . 202 XI. The Reproductive System 212 XII. Embryology 229 XIII. Coloration 243 XIV. Classification ...... 258 XV. ZOOGEOGRAPHICAL DISTRIBUTION . . 281 XVI. The Geological Record .... 301 XVII. BlONOMCS, ETC. ..... 321 XVIII. British Species ..... 338 XIX. Collecting, Rearing and Biological Methods 353 APPENDIX A. Bibliography 363 B. Glossary ....... 376 c. Some Important Synonyms 381 Index of Illustrations .... 382 Index of Text ...... 386 Table of Entogenic Genera of Odonata To face page 282 Map of Zoogeographical Distribution „ 300 I. II. III. IV. PLATES Frontispiece To face page 44 243 321 ERRATA Plate I, Fig. 4. For New Britain read India. Fig. 5. Fur Papua read New Britain. CHAPTER I INTRODUCTION Historical. In his subdivision of the Class Insecta, Linnaeus placed all the Dragonflies known to him in the single genus Libellula, forming the family Lihellulidae of the very heterogeneous Order Neuroptera. His pupil Fabricius, in re-arranging the Orders of Insects by the structure of their mouth-parts, constituted the Dragonflies as a separate Order under the name Odonata^, because of the form of their mandibles. His general scheme quickly fell into disuse. But the name Odonata persisted, and has been adopted by all subse- quent writers. Latreille merged the Odonata with the May-flies {E phemeridae) in his Order Subulicornia, based on the form of the antennae. To these Gerstaecker added the Stone-flies {Perlidae) forming a group Orthoptera Amphibiotica, characterized by the possession of aquatic larvae. Erichson, in splitting up the old Order Neuroptera, formed an Order Pseudoneuroptera to comprise the Odonata, Evhemeridae, Psocidae and Termitidae. Later workers have almost unanimously rejected these unnatural groupings, and have regarded the Odonata as a distinct Order. The name Paraneuroptera has been suggested by Dr Shipley as a substitute for Odonata, with a view to maintaining the termina- tion "-ptera" for all Orders of Winged Insects. Recent work on the morphology and phylogeny of the Dragon- flies has not only tended more and more to conflrm the view that they form a distinct Order of Insects, but has even gone so far as to increase our sense of their isolation. Thus Handlirsch [07], on ^ Greek obovs a tooth, stem oSofr-. The word should therefore more correctly be Odontata. 2 INTRODUCTION [CH. phylogenetic grounds, raises them to the rank of a Sub-Class (Libelluloidea). This seems to be going rather too far, particularly now that the very important character of the radial sector has been shewn to hold only for the Anisoptera, and not for the Zygoptera. We shall therefore be content to treat the Dragon- flies as a very distinct and isolated Order of Insects, under the name Odonata or Paraneuroptera. Little attention was paid to the Odonata during the first half of the nineteenth century, and the total of described species remained small, despite the excellent work of Rambur [127] and others. The darkness was dispelled by the appearance on the scene of the great Belgian master, rightly called the "Father of Odonatology," Baron Edmond de Selys-Longchamps. His work on the Order began just before the middle of the century, and lasted until his death in 1890. His whole life was devoted to the study of Natural History. But the Dragonflies soon became his favourite group, and remained so to the end. He not only built up a magnificent collection of these insects, but published mono- graphs of every subfamily except the Libellulinae. His collections, which contain an enormous number of types, were placed at his death in the Natural History Museum at Brussels, where, we most sincerely trust, they have remained untouched by the greedy hands of the Despoilers. The life of de Selys marks the Classical Period of the study of Odonata. It was he who first saw the importance of the wing- venation as the basis of classification, and who started the study of the Order along the right lines. To him we owe the facts that the Odonata now stand as one of the best-understood of all Orders of Insects, and that their classification is based on a natural system supported by an ever-increasing weight of morphological and phylogenetic evidence. After his death, the work of de Selys was continued by his sons, who arranged for the publication of a series of catalogues of the various groups of animals in their father's collection. These catalogues became expanded into complete and magnificent mono- graphs, dealing with every known species. In the case of the Odonata, the Corduliinae, Aeschninae and Calopterygidae were en- trusted to M. Rene Martin of Paris. He has faithfully completed l] INTRODUCTION 3 his task, though the work on the last-named family had un- fortunately not left the printer's hands by August 1914. The immense task of dealing with the Lihellulinae was entrusted to Dr F. Ris, of Rheinau, Switzerland. As a result of more than ten years' labour, he has now completed his work, in a series of brilliantly conceived and executed fascicules which mark the high- water level of systematic achievement in Entomology. The Gomphinae, Lestidae and Agrionidae were unfortunately entrusted to three German authors. Not one of these has so far made any serious attempt to deal with his obligations, accepted now over thirteen years ago. The personal influence of de Selys reacted directly upon the next generation of students of the Odonata, and shewed itself in an era of great activity and advancement in our knowledge of the Order. There were many, including Martin and Ris, who could proudly claim that they had sat at the feet of the master, and had been directly fired with his spirit of enthusiasm. The younger generation of Odonatologists, amongst whom I am proud to include myself, owe to these men what they in their time owed to de Selys. For the past twenty years, with unfailing courtesy and industry, they have placed their unrivalled knowledge at the disposal of all who sought it, and have lit afresh the flame of devotion which we trust will now be passed on to succeeding generations. Mainly owing to their endeavours, the systematic study of the Odonata has been brought up to a level of excellence not exceeded in any other Order ; and thus the way is now paved for the biologists of the present and the future to pass on to the deeper studies of the morphology and physiology of the group, not by selecting, at a chance, any common type that happens to be convenient, but with full knowledge of the phylogenetic position of their regional fauna, and an added certainty as to the degree of specialization to be expected of the types they choose to study. Chakacters of the Order. Habits. CarmvoroMS in all stages. Amfhihiotic: the larva or nymph being aquatic, breathing by means of tracheal gills; the imago being aerial and breathing air directly through spiracles. Hemimetaholoiis : the larva active, campodeiform, with no true 1-2 4 INTRODUCTION [CH. pupal or resting-stage, but with a considerable change of structure at metamorphosis. Imago. Head with biting mouth-parts, the mandibles thick and strongly toothed; antennae reduced, filiform; three ocelli, two large compound eyes. Thorax with small movable prothorax, and large obliquely-placed synthorax (formed by fusion of meso- and metathorax) ; the legs placed well forward and of little use for walking ; the wings placed far backward ; two pairs of thoracic spiracles. Wings, two equal or sub-equal pairs, unfolded, richly veined; the six typical longitudinal veins all present, but R -f- M and also Cu + A fused basally for a short distance; a secondary anal vein {A') developed; Sc stopping far short of the apex, at a thickened cross-vein, forming a joint, the nodus; a thickened membranous patch or pterostigma nearly always present between C and R, near the apex ; R either simple or once-branched ; M branched either three or four times; an arcidus and discoidal cell always present. Abdomen w^ith ten complete segments, and ter- minal appendages; in shape, narrow', elongated, the pleurites membranous, the large tergites enfolding the reduced sternites; eight pairs of spiracles. External Genitalia : — ^ with genital pore on ninth sternite; complicated copulatory apparatus developed from the second and anterior part of the third sternites. 2 with genital pore in the suture between eighth and ninth sternites; ovipositor either complete or reduced. Mid-gut without caeca. Numerous Malpighian tubules. Larva. Similar to imago in general features, but with less elongated and often very broadly rounded abdomen. Compound eyes present from birth; ocelli absent during most of larval life. Antennae larger than in imago. Labium very highly specialized, forming an elongated jointed masl- for the capture of prey. Legs strong, longer than in imago. Wing-sheaths developed externally and gradually, the hind-wdng coming to overlie the fore- wing; developing wings richly supplied with tracheae, around which the main venational scheme is subsequently formed. Tracheal System secondarily closed; the spiracles present, but either closed or only partly functional. Gills developed in the rectum, or on the anal appendages ; rarely also as lateral appen- dages on some of the abdominal segments. l] INTRODUCTIOlSr 5 Position of the Order in the Class Insecta. The Dragonflies occupy such au isolated position within the Class, that it is a difficult task to indicate any other groups as having any real relationship with them. Morphologically, they have been considered to shew relationships with (a) the Stone-flies or Perlaria, (b) the May-flies or Plectoptera. The points of resemblance with the Perlaria are : (i) the pres- ence of three ocelli, (ii) the large number of Malpighian tubules, (iii) the aquatic habit of the larvae, and the fact that both breathe by means of tracheal gills, (iv) the retention of the cerci alone of all the abdominal appendages in the larva (except in a few genera). It seems very probable that not a single one of these characters has been inherited from a common ancestor. As regards the ocelli, fusion of two out of an original four to form a single large median ocellus takes place in a number of widely separated groups of Insects. The Malpighian tubules of the Dragonfly are arranged in groups of five or six together. The young larva has only three tubules. Thus the large number of these organs is again a secondarily developed character. The larvae of the Perlaria have no stigmata, and their tracheal system appears to be a 'primarily closed one. The larvae of Odonata, on the other hand, possess stigmata, and their tracheal system is only secondarily closed. This points to the supposition that the Perlaria were aquatic long before the Odonata, and that the aquatic habit is not to be traced back to a common ancestor. Lastly, as regards the cerci, these appendages in the Perlaria remain as undifferentiated filiform organs, and are so carried over into the imago. In the Odonata, they are either reduced, or else developed in the form of caudal gills. Only in the Zygoptera are they carried over into the imago. Combining this with the facts that a few Odonate larvae still possess their lateral abdominal appendages, and that the gill- systems in the two Orders are absolutely difTerent, we cannot be far wrong in assuming that the retention of the cerci has been brought about by gradual and incomplete parallel reductions of the original complete set of abdominal appendages once possessed by the ancestors of all Insecta, and does not indicate any special affinity. 6 INTRODUCTION [CH. While the true affinities of the Perlaria remain a matter of much doubt, we must maintain that no true relationships can be proved between them and the Odonata, on the evidence available to us. Turning next to the May-flies, the position is somewhat different. Their points of resemblance with the Odonata are: (i) the presence of three ocelli, (ii) the large number of Malpighian tubules, (iii) the aquatic habit of the larvae, and the fact that both breathe by means of tracheal gills, (iv) the retention, in the larvae of the Plectoptera, and in a few archaic larvae of Odonata, of the lateral abdominal appendages, in the form of gills, (v) the retention of the cerci in the larvae of both groups, and in the imagines of the Plectoptera and the Zygoptera, (vi) the reduced awl-like antennae, (vii) the presence of only one wing-axillary, and the inability to fold the wings downwards either in a flat or roof-like manner over the body. With regard to the first three points, the same argument may be applied as in the case of the Perlaria, especially as the tracheal system of the Plectoptera is as little like that of the Odonata as is that of the Perlaria. Both the latter and the Plectoptera, in fact, appear to have been aquatic in their larval stages long before the Odonata became so. The fourth character mentioned becomes rather striking when we compare the larva of Pseudophaea (fig. 87) or Cora with that of an Ephemerid. It may well be that we have here some evidence of a very ancient connec- tion between the two groups. As regards the retention of the cerci, we must repeat the argument used above. With respect to the form of the antennae, the reduction almost certainly rests upon convergence, and cannot be used as an argument for relationship. Similar antennae appear quite independently in other groups of Insects. The last character, (vii), seems undoubtedly to have been derived from a common ancestor. Thus we have in this case a certa'n amount of evidence for a connection between the two Orders, very long ago. The palaeontological evidence here helps us considerably. The Protodonata of the Upper Carboniferous were admittedly the ancestors of our present-day Odonata, although an unfortunate gap in the Trias hides the exact line of descent from our view. The Protodonata were large insects with well-developed sub- equal wings. They were undoubtedly Dragonflies in facies and habits. l] INTRODUCTION 7 though lacking those high specializations of the wing-venation which mark off the Odonata of to-day. Now we can also trace back the Plectoptera to the Jurassic, where we find forms differing little from those of the present day except in the possession of less reduced hind-wings. The fossil record undoubtedly points to a gradual reduction taking place in the hind-wings of the Plec- toptera, from an ancestral form in which fore and hind- wings were sub-equal. A study of Permian fossils also bears this out. Coming at last to the Carboniferous period, we find evidence of the existence of a group of moderately delicate insects with four almost equal wings. These, under the name Protephemeroidea, are generally admitted to have been the ancestors of our May-flies. It will thus be seen that, in the Upper Carboniferous, the two Orders Plec- toptera and Odonata were already differentiated, but far more closely related than they are to-day. It seems therefore not un- reasonable to recognize a very archaic connection between the two, of a far more definite character than we can point to in the case of the Perlaria. We need, however, only remember the numerous and overwhelming points of difference between the two Orders to realize how very slight the relationship really is. We have only to recall the differences in the structure of the head and thorax, the atrophy of the mouth-parts in the May-flies, the very characteristic arrangement of interpolated sectors in their wing- venation, together with numerous points of difference in the internal anatomy, to feel the force of this. We have not mentioned the sub-imago of the Plectoptera, since a somewhat analogous process is evident in the case of the pronymph of the Odonata. But for the fact that the Planipennia have a complete meta- morphosis, we might be tempted to see a considerable resemblance between the very ancient Protascalaphine genus Stilhopteryx and the Odonata. That Stilhopteryx is remarkably like a Dragonfly in its flight and appearance, nobody who has seen this insect alive will deny. Moreover, the fact that it is admittedly the most archaic of all the Ascalaphidae makes it impossible to accept this resemblance as one purely due to convergence. The more highly specialized Ant-lions depart widely from the type shewn by Stilbopteryx, and do not resemble Dragonflies in the least. It seems allowable to postulate an exceedingly remote ancestor 8 INTRODUCTION [CH. I common to these two groups — an ancestor with an incomplete metamorphosis, and only such structure of wing-\enation as underlies both types. From that ancestor, the Odonata broke away by the adoption of the aquatic larval habit. The consequent change in the source and direction of the oxygen-supply to the developing wings may have been the determining factor for those wonderful specializations in wing-venation in which the Odonata are unique. On the other hand, a more direct line of descent would lead to Stilhopteryx by the intercalation of a pupal stage, and high specialization in the habits and form of the larva. Palaeontology cannot help us here, since any connection of the kind supposed must have existed before the Protodonata became differentiated out. To conclude this discussion, we find in the Odonata a singularly isolated group, marked by very high specializations of structure, superimposed upon an exceedingly archaic foundation. No near relatives of these extraordinary insects exist to-day, nor have any such existed since Palaeozoic times. We may, however, most certainly point to the Plectoptera as being the group which has the closest affinity with them, slight though it be. In a serial arrangement of the Orders of the Insecta, the Odonata are best placed next to, and just after, the Plectoptera. CHAPTER II THE IMAGO. EXTERNAL FEATURES AND SKELETON In this chapter we shall deal with the general external features of the Dragonfly, the exoskeleton of the head, thorax, and abdomen, its internal prolongations or endoskeleton, and all the appendages except the wings, which are treated separately in chap. iii. The Head (fig. 1). The head of a Dragonfly, like that of other insects, shews no clear traces of metameric segmentation. Embryology, however, tells us that it consists of a number of segments, most probably six, which can be classified according to their position with regard to the mouth, and the nature of the nerve-centres supplying their appendages. The following table shews us at a glance the usually accepted theory of head-segmentation : No. of Segment Position Appendages, etc. Central Nervous System c , o 'First Second Third Preoral Compound Eves^, Lateral Ocellii Antennae, Median Ocellus^ Intercalary segment, with no appendages Protocerebrum \ Deutocerebrum Tritocerebrum Brain, or r Supra-oesophagea Ganglion •5 ^ - OS r Fourth Fifth Sixth Postoral Mandibles First Maxillae Second Maxillae (= La- bium) First Pair of Ven Ganglia Second Pair of \ tral Ganglia Third Pair of \ tral Ganglia trals Vn- ''cn- Sub- . oesophageal Ganglion Generally agreed upon as not being true appendages. 10 THE IMAGO [CH. It seems fairly certain that, originally, only the first of these segments was preoral. Thus it corresponded with the prostomium of Annulate Worms, and is not strictly to be considered as an appendage-hearing segment. The mouth itself has moved from an original ventral position, to take up a place at the extreme anterior end of the head. Hence the original ventral surfaces of segments vt ant Fig. 1. Exoskeleton of head, thorax, and first two abdominal segments, a. Austro- phlebia costalis Tillyard, 9> dorsal view. b. The same, ventral view. c. The same, lateral view. d. Head of the same, front view. e. Head of Lestes, dorsal view. f. The same, lateral view. g. Metathorax of Orthetrum, ventral view. A analis; A' secondary analis; ac anteclypeus; air ante-alar ridge; awi antenna; ap costal process; /Is urosternite; .4< urotergite; a.T axillary; C costa ; Cu cubitus ; ex coxa ; dr dorsal carina ; e eye ; epic epicranium ; epm epimerum; ejis episternum; fr frons; /rj its superior, fr,^ its anterior portion; Fm; fore-win^ ; fygena: /(S humeral suture ; i/w hind-wing; ips inira- episternum ; U lateral lobe of labium ; Im labium ; Ir labrum ; Zsj first, Zsg second lateral suture ; 71/ media; mi membranule; mcZ mandible; me median eye-line; ml median lobe of labium; ml mentum; A'' notum; occ occiput; ocl lateral, ocm median ocellus ; orb orbit ; ot occipital triangle ; P prothorax ; pc postclypeus ; PI pleurum ; PN postnotuni ; pol postocidar lobe ; PS post- sternum ; B radius ; S sternum ; Sc subcosta ; Sp spiracle ; t temple ; v vertex ; vt vertical tubercle, (a-d and Gx2; e, f x 5J.) Original. it] external features and skeleton 11 1-3, bearing the clypeus and labrum, have become pushed dorsally upwards into their present positions. The Compound Eyes (see chap. vii). In the Dragonfiies, as is the case also in a somewhat less degree with their Vertebrate analogues the Birds, the shape of the head, and the sizes and relationships of its parts, have become profoundly modified in correlation with the extraordinary development of the sense of sight. The orbits have become expanded to an enormous diameter, in order to accommodate the huge compound eyes (e), which, in many forms, occupy by far the largest part of the whole head. In the Aeschninae and Lihellulidae this development culminates in the meeting of the two eyes mid-dorsally on the head in a long line of contact, the median eye-line (a, me), the true natvire of which is explained on p. 146. The stages leading up to this result are preserved for us in a long line of existing Anisopterid forms. Of these, the Gomphinae and Petalurinae lie nearest to the primitive type. Their eyes are rounded, not excessively large, and separated by a width of epicranium less than the dorsal diameter of the eye. The eyes of Meganeura (fig. 156) and other early fossil forms seem to have been very similar to this. In the Chlorogomphinae the eyes become somewhat more transversely elongated, and take on a more oval shape. They just fail to touch one another mid- dorsally. In most of the Cordulegastrinae a further slight advance is seen, the eyes just meeting in a point. In the Petaliini there is another slight advance, the eyes meeting for a very short distance. In the Brachytronini the eye-line becomes of moderate length; while in most of the Aeschnini and Lihellulidae it is so long as to obliterate the greater part of the epicranium, and approaches in length the diameter of the eye itself. In the Zygoptera (e, f) the eye has remained rounded and comparatively small. The tendency has been towards improving the power of sight by a progressive movement of the eyes laterally outwards. Thus the highest forms in this suborder have the head transversely elongated to form a kind of cross-stalk carrying the button-like eyes at its two ends. The distance between the eyes then becomes much greater than the dorsal diameter of a single eye. In the older forms (Calopterygidae) the transverse elongation of the head is not so noticeable. 12 THE IMAGO [CH. With the above facts borne in mind, we may now profitably study the conformation of the head itself. The various pieces or sclerites of the head do not appear to be closely connected with the original segmentation. With the exception, perhaps, of the labrum, clypevis and gula, they cannot be assigned to definite segments. Thus they are chiefly named from their position in relation to the head and mouth. The Epicranium (ep^c). This is by far the largest part of the head-skeleton. If we compare the head to a hollow rectangular box with a lid on, the epicranium represents not only the lid but the upper portions of the front, back, and sides of the box as well. It is bounded laterally on either side by the eyes, which are set in two large hollows with thickened rims, the orbits (a, b, orb). Postero-laterally, each orbit becomes somewhat swollen at a place where the eye itself is slightly concave when seen in profile. This is the temple (c, t). Posteriorly, the epicranium ends in a sharp shelf, the occiput (e, occ), which varies greatly in size and shape, according to the extent of the eyes. It is broadest in the Zygop- tera, where it usually has a wavy outline, and may be divided into a median portion, the occiput proper, and two lateral postocular lobes (e, pol), corresponding with the more prominent lobes of the same name in the larva. These serve for the attachment of the strong muscles of the mandibles and maxillae. In the Gomphinae the occiput is an almost straight shelf, often hairy, and modified in the female, to assist in copulation, by the presence of hooks or tubercles -which help to hold the complicated anal appendages of the male in position. In the Aeschninae and Libellulidae, the occiput becomes reduced to a small occipital triangle (a, ot). This carries, in some forms, a prominent horn or tubercle. A little in front of the middle of the dorsal surface of the epicranium is a triangular area called the vertex (e, v), marked by the pres- ence of the three ocelli, one of which is median and anterior (ocm), the other two lateral, smaller and posterior (ocl). Originally flat, as in most Zygoptera, we find that in Anisoptera it tends to become raised up, in proportion to the restriction in the area of the epicranium caused by the increase in the size of the eyes. In the Gomphinae the vertex may be a small raised plateau, or a strong transverse ridge with a posterior sunken portion, or a ll] EXTERNAL FEATURES AND SKELETON 13 prominent ridge with projecting lateral tubercles. In all these cases, the effect is to force the lateral ocelli forward almost in line with the median one. In the Aeschninae and LihelluUdae. the vertex, like the occiput, becomes excessively reduced, forming a small vertical tubercle or vesicle (a, vt) round which the ocelli are grouped close together. Just in front of the antennae (p. 15) there runs a transverse groove, the frontal furrow or suture. This separates the vertex from the projecting anterior portion of the epicranium, known as the frons (e, f, fr). In the Zygoptera, the frons is a sloping shelf, usually rather narrow, and generally somewhat convex. Below it, on either side, lies a further small piece of the epicranium called the gena or cheek (g), bordering the eye. In the Anisoptera, and in the genus Ceriagrion amongst Zygoptera, the frons becomes raised up into a more or less prominent shelf, having a posterior or superior horizontal portion (/r^) and an anterior or vertical portion (/rg), forming a definite forehead. In some Aeschninae (e.g. Austrophlebia, a, c, /r^) the frons projects in the form of an anvil, with the superior portion slanting upwards to a median point. The Clypeus. This is a separate piece inserted below the frons. It forms the chief part of the face of the insect. In all Dragonfiies it is divided into an upper postclypeus or nasus (pa) and a lower and smaller anteclypeus or rhinarium (ac). The former embraces and often overhangs the latter (d). In many Zygoptera the post- clypeus forms a kind of shelf comparable with that of the frons in Anisoptera (f, pc). In Rhinocypha (Plate I, figs. 4, 5) the clypeus projects forwards as a kind of snout. The Labrum. Below the clypeus, and bounding the mouth from above, lies a strong median piece with a more or less rounded lower border. This is the labrum or upper lip (Ir). It is narrower than the clypeus, and is bounded on either side by the strong pear-shaped bases of the mandibles {md). In the Anisoptera it continues the plane of the face vertically downwards; in the Zygoptera it projects obliquely forwards and downwards. To- gether with the epipharynx, which it carries on its inner surface, the labrum forms an organ of taste (chap. vii). The Mouth. When at rest, the mouth is completely closed by 14 THE IMAGO [CH. the labium above, and by its appendages below and at the sides. These appendages are the mandibles (md), the first maxillae (mxj) and the second maxillae or labium (Im) ; the first two placed laterally, the last ventrally. Hence the mouth is capable of great extension, the gape being as wide as the face itself. The labrum and labium act as upper and lower lips respectively, while the mandibles and first maxillae constitute two pairs of laterally- working jaws. The Gula. The only remaining portion of the head-skeleton proper is a small membranous piece called the gula. Referring again to our simile of the box, this would occupy the lower part of the back of the box, below the occiput. In the higher Anisoptera the large orbits unite in a median line which runs downwards behind the head. Below this line, or, in the other forms, below the lower portion of the occiput, the head joins the neck. If, now, the head be removed, the rounded opening of the jjosterior fora- men will be seen, through which the cavities of the head and thorax are in communication with one another. The small gula closes this foramen from below. Although so insignificant in size, it is of interest in being generally considered to represent the ventral plate or sternum of the sixth or last head-segment. It forms a base of attachment for the labium. The Tentorium. We have already seen that the frons is really the most anterior portion of the epicranium. The anterior edge of the epicranium, bordering the clypeus, is turned inwards to form an internal chitinous shelf, separating off an upper frontal cavity from a lower clyjjeal cavity. Again, the tw^o ends of the frontal furrow also send a pair of ingrowths into the head-cavity. These run dowuAvards and backwards close together, and meet the above-mentioned shelf at its posterior end. They then run on a little further to end up on either side of the posterior foramen. These two rods are in their turn connected by a short transverse rod which divides the posterior foramen into upper and lower halves. The lower passage carries the nerve-cord, the upper the oesophagus, salivary ducts and tracheae. Thus we see that the head has a definite endoskeleton, formed of four separate pieces or endosfernites. The whole structure is called the tentorium. The passage between the parts of the tentorium is called the tentorial II] EXTERNAL FEATURES AND SKELETON 15 foramen. It slants upwards and backwards. Hence the oesopha- gus, in running forward towards the mouth, passes downwards along this foramen. The brain lies above the oesophagus, between the supporting endosternites of the tentorium. From it the two circum-oesophageal commissures pass downwards, encircling the oesophagus, to join the sub -oesophageal ganglion, which lies below the tentorial foramen. The Appendages of the The Antennae. These are placed epicranium (fig. 1, ant; fig. 2). Each has its base inserted into a small pit or hollow, the torulus. The antenna itself is always short, and is formed of from three to seven joints. The latter is the usual number. In the Gomphinae the number of joints does not exceed four. The first or basal joint is called the scape; it is short and thick. The second joint or pedicel is longer, and only slightly thickened. The remaining joints or distalia are very slender. The com- jDarative lengths of the joints vary in different genera. In Calopteryx and Synlestes, the pedicel becomes hyper- trophied, so as to equal or even exceed in length all the other joints put together. Fig. 2 shews a number of different types of Odonate antennae. The sensory function of these organs is very poorly developed in all Dragonflies (see chap, vii). The Mandibles (fig. 3). These are very strong, thick, unjointed appendages, hinged to the head-capsule by a true ginglymus articulation. Viewed extern- ally they are somewhat pear-shaped, the pointed distal end representing the Fig. 2. Various foi'ms of an- tennae. A. Acschna brevi- styla Ramb., imago. B. Austrogomjihus ochraceus Selys, imago, c. Tlie same, larva. d. Hemigomphus heteroclitus Selys, larva. E. Neosticta canescens Till- yard, at metamorphosis. F. Synlestes iveyersi Selys, imago. G. The same, larva, rfi to fZj distalia ; ped pedi- cel; sc scape. (All figures X n.) Original. stalk of the pear. The 16 THE IMAGO [CH. basal portion is shiny, slightly convex, and often brightly coloured, forming a kind oifree or mandibular cheek, in front of the true gena. Fig. 3. Mandible of Aeschna hrevistyla Ramb. a. External view, view, t incisors; m molar. ( x 1.3.) Original. B. Internal The teeth of the mandible are exceedingly strong They lie from the distal end to near the base. They may be divided into two groups : (a) a distal group of three incisors (t), and (6) a large basal molar (m). The incisors are sharply pointed and slightly curved. The largest incisor forms the apex of the mandible. The molar has four cusjos (originally, no doubt, separate teeth), set on a broad raised base in the shape of a Z. The cusps are placed one at each end and one at each angle of the Z. The structure of the mandibles is very constant throughout the Order. Biramous mandibles occur in a few archaic larvae. The First Maxillae (fig. 4). These are jointed appendages, slenderer, longer, and much less powerful than the mandibles. They lie between the latter and the lateral lobes of the labium, Fig. 4. First maxilla of Atschna hrevistyki Ramb. ( x 15). c cardo; il inner lobe; p palp; st stipes. Original. Il] EXTERNAL FEATURES AND SKELETON 17 and are barely visible when the mouth is closed. The basal joint or cardo (c) is somewhat twisted. The distal joint or stipes (st) is longer and wider. It carries on its distal end two freely movable processes or lohes. Of these, the otiter lobe or palp (p) is curved, hairy, and only moderately pointed. It is obviously an organ of touch. The inner lobe {il) represents the lacinia and galea of more typical mandibulate insects, fused together. It has a rather flattened base, from which project six large, sharply pointed and somewhat curved teeth, occupying the outer two-thirds. The inner third forms a small lobe or tubercle clothed with long sensory hairs. The inner lobe is probably both prehensile and tactile in function. Mastication is performed almost entirely b"" the mandibles. Fig. 5. Labium of Aeschna brevistyla Raiub. ( x 10). eh end-hook; II lateral lobe; m mentum; mh movable hook; ml median lobe; sm sub-mentum; fsq squame. Original. The Labium (figs. 5, 6). In all insects this organ is formed by the fusion of the second maxillae. It is attached basally to the gula, and represents the paired appendages of the sixth or last head-segment. In the Odonata, both larvae and imagines, it is of a highly specialized form. The question of the homologies of its parts has given rise to much controversy. The conflicting views are dealt with in the account of the larval mask on p. 80. In the imago, a considerable variation of form is seen in the labia of different groups. In all cases, however, we can recognize the same principal parts. There is a broad short base with a projecting middle portion, at the sides of which are articulated T D.-F. 2 IS THE IMAGO [CH, two lateral pieces or lobes. The broad base is considered to be formed by the fusion of the sub-mentuni {sm) and mentum (m). Fused with the mentum distally is the projecting median lobe (ml), probably representing the ligula of other mandibulate insects. The form of this lobe varies greatly. In the Zygoptera and the older Anisoptera it is very prominent and definitely bifid (fig. 6 a). In the Aeschninae (fig. 5) the median cleft has disappeared, but Fig. 6. A. Labium of Synlestes weyersi Selys ( x 25). b. Labium of Diplncodes haematodes Burm. ( x 15). cl cleft of median lobe; other refeiences as in fig. 5. Original. a slight median groove or line marks the place of its disappearance. In addition, one or two hard black tubercles may project on either side of this median groove. In the Lihellulidae (fig. 6 b) the median lobe is reduced to a small triangular piece, over which the huge lateral lobes fold so as to meet one another in a long median line. On either side of the mentum, the base is widened by the development of two side-pieces or squames (sq) which support the n] EXTERNAL, FEATURES ANIJ SKELETON 19 lateral lobes {II). The changes in form of these lobes are correlated with those of the median lobe. Normally, each lateral lobe has its outer border more or less strongly convex, its inner border either slightly concave, waved or nearly straight. The inner border ends distally in a sharp end-hook {eh) which is fixed. Just external to this hook is a small articulated piece, called the movable hook {mh), much resembling a small palpal joint, curved, hairy, and only moderately pointed. The whole lobe is more or less hairy. The end-hook varies in shape. It is most sharply pointed in the Zygoptera (fig. 6 a), and in the older Aeschninae. In Austrophlebia it is strongly bifid. In Cordulegaster it has a set of supplementary teeth. In the higher Aeschninae it tends to become reduced in size. The movable hook also tends to become reduced in size along with the end-hook. In the Libellulidae (fig. 6 b) the end-hook has disappeared, while the movable hook has lost its articulation, and is only represented either by a small spine or a thickened chitinous spot. The lateral lobes themselves are greatly enlarged, with both inner and outer borders convex. As already described, they meet together above the reduced median lobe. Thus the closed mouth of the Libellulidae shews the very characteristic appearance of having a large T-shaped slit, the stem of the T being formed by the line of meeting of these two lateral lobes. The labium carries internally, projecting from its base, a pro- minent tongue-like organ projecting into the mouth. This is the hypopharynx or lingua. This is dealt with in chap. vii. The salivary ducts open close to its base. The Neck, or Microthorax (fig. 7). In the Dragonfly the neck is exceedingly small and slender, pivoting the head almost, as it were, on a point, and allowing of turning movements to some degree in every direction. Although the neck is firmly fixed to the prothorax, yet it is morphologically part of the head itself. The plates of the neck, or cervical sclerites (1-4 mi), together with the gula, represent the skeleton of the last head-segment. The neck is supported by four pairs of these plates. Two of these (1 mi, 2 mi) are rather large and irregular 2—2 20 THE IMAGO [CH. in shape. Posteriorly they are closely connected with the sides of the prothorax. Two others (3 mi, 4 mi) are very small, and lie freely in the neck membrane [157]. The orifice of the neck, which is the posterior continuation of the posterior foramen of the head, is so minute that it is difficult to believe that the ventral nerve-cord, oesophagus, salivary ducts and tracheal trunks could possibly all pass through it. Yet such is actually the case. It is interesting to note that, if the head of a Dragonfly be removed, either accidentally when striking with the net, or purposely with fine scissors, the body will continue to live for as much as two days longer, while the wings vibrate vigorously whenever the thorax is touched. The movements of the head on the neck are brought about by a number of pairs of small muscles arising from the prothorax and anterior part of the mesothorax. They are placed dorsally, ventrally, and laterally, so that the head can turn freely in all directions. The Thorax (figs. 1, 7). The thorax of the Dragonfly is very highly specialized, in accordance with the requirements of the very powerful wings and the altered functions of the legs. On the one hand, the prothorax, though remaining a distinct segment, becomes greatly reduced in size. On the other, the meso- and metathorax are large and powerful, and become fused together into one solid mass, which may conveniently be termed the synthorax. In general, three regions can be distinguished in each of the three thoracic segments. These are the tergum or true dorsal region, the sternum or true ventral region, and the pleura or lateral portions. By the addition of the prefix pro-, meso-, or meta-, we indicate to which segment the part in question belongs. The same prefixes may be attached to the name of each particular sclerite. The Prothorax (fig. 7). This is a small segment supporting the first pair of legs, and the neck or microthorax. It varies greatly in shape. As a general rule, it shews the more normal proportion of parts in the most archaic groups, while it becomes highly reduced or otherwise altered in the most advanced forms. nj EXTERNAL FEATURES AND SKELETON 21 In all Dragonfiies the tergum is occupied by a strong sclerite, the 2>'''onotum (N), divided into anterior, median and posterior transverse lobes. The median lobe is generally the largest, and is divided into two by a median longitudinal depression. The posterior lobe may send out a slender strip descending behind the epimerum on either side, and known as the postejyimeral strip. The ijropleurum is complex in the older forms, such as the Gom- phinae, where it consists of a large epimerum {epm) behind, and a narrow elongated episternum (eps) in front. This latter is con- stricted into an upper and a lower portion; the lower piece has been named the proinfraepisteryium. In the Lihellulidae (a) both parts are reduced, and the episternum is undivided. The pro- sternum (Sj) in the older forms is a single ventral plate with expanded Fig. 7. Skeleton of neck and prothorax. A. Libdlula auripennis Burm. B. Lestes uncatus Kirby. cx^ procoxa; epm proepimerum; eps proepisternum ; 1-4 mi cervical sclerites; N pronotum; ps presternum; s^ prosternum. After Snodgrass. antero-lateral wings, which fuse with the proinfraepisternum. In Lihellulidae these expansions become separated off as distinct lateral pieces, the presierna (a, ps). Internally, the prosternum carries a bifurcated piece, the f urea, forming the endoskeleton of the prothorax. Two slender rods connect the prosternum with the mesosternum, and serve to hold the prothorax in position. They also restrict its mobility upon the synthorax to small dimensions. In the Afjrionidae many interesting modifications of structure are to be found in the pronotum. This may take the form of ornamentation — as in the males of the Oriental genus Amphi- cnemis, where the posterior lobe projects backwards and upwards in a huge spine — or it may be concerned with the process of copulation. In such a case, the prothorax of the female is 22 THE IMAGO [CH. modified to receive the reduced and highly specialized anal appendages of the male, in such a way that a firm interlocking of parts may be secured (see p. 37). The Synthorax (fig. 1). The remarkable form of the synthorax in Dragonflies is correlated with special conditions of the legs and wings. The former have moved forwards towards the mouth, carrying the sterna with them. The wings, however, have moved backwards away from the head, carrying the terga with them. The reduction in size of both terga and sterna is more than counter- balanced by an enormous development of the pleura. To under- stand the significance of these changes fully, the student should trace the development of the thorax in the larva, from the time that the wing-buds appear (p. 74). The well-known obliquity or "skewness" of the Dragonfly thorax has been ingeniously measured by Needham and Anthony [107]. If the wings be spread out into a single plane, as if for flight, that plane practically represents the notal plane of the synthorax, and its obliquity to the long axis of the body may be measured by the slant of a line drawn through the tw^o axillaries (c, ax2, ax^) or other suitable points. This angle of obliquity is much smaller in the Anisoptera than in the Zygoptera, the approxi- mate ranges being from 19° to 38° in the former, and from 35° to 62° in the latter. Thus we see that, in this character, the Anisoptera have remained in much the more primitive condition. The ability of the Zygoptera to fold their wings back above the abdomen is of course correlated with the increase in the angle of obliquity. The obliquity of the synthorax is brought about chiefly by the great development of the mesepisterna {eps^, which grow fonvards and upwards in front of the tergum, so as to meet in a long middle line, forming the dorsal carina or ridge (dr). Thus they form, in their new position, not only the humeral portion of the thorax, but also all the dorsal part in front of the ante-alar ridge (air). The true terga are pushed backwards and lie between the wing- bases. Further, there is a correlated growth of the metepimera (epm^) doivmvards and backwards, which pushes the sterna and the legs forwards, so that the latter come into position close behind the mouth, and are used for holding the prey during flight. In most Dragonflies (g) the metepimera actually meet and fuse n] EXTERNAL FEATURES AND SKELETON 23 together just behind the metasternum (S^). A median gap left between the line of union and the base of the abdomen is filled in by a thinner plate or plates, the post-sternum (ps). In the older Aeschninae, the fusion is not accomplished, and we have the condition shewn in b. The Terga. The mesotergum is the reduced area lying behind the ante-alar ridge (air) and between the bases of the fore-wings. It bears two prominent plates, often brightly coloured. The upper of these, or tnesonotum proper (a, iVg), consists of a median shield-shaped scutum and a pair of smaller lateral scutella. The lower plate, known as the post not imi or pseudonotum (PN^), is a short wide piece, likewise divided into three parts, a median post- scutum and two lateral postscutella. The metatergum lies between the bases of the hind-wings. It carries a large upper plate, the metanotum proper {N^), bearing two convex bosses, and a short, wide, raised postnotum (PN^) n the form of an obtuse triangular ridge. The Pleura. The suture between the meso- and metapleurum is partly obliterated, except in the Calopterygidae, where it runs from between the meso- and metacoxa upwards and backwards to end below and behind the base of the fore-wing. In the other families only the lower half, or less, can be distinguished, running from the coxal region to a point just in front of or above the metastigma. This dividing suture is called the first lateral suture (c, Zsj). The mesopleurum, lying in front of it, is divided into two pieces by the very distinct humeral suture (c, hs). These pieces are an anterior mesepisternum {eps^} and a posterior mesepimerum (epm^). From the former a small portion is divided off above the mesocoxa {CX2) and is called the mesinfraepisternum^ {^s^. The metapleurum, lying behind the first lateral suture, is like- wise divided into two pieces by the very distinct second lateral suture (c, ls2)- Of these, the anterior piece or metepisternum (eps^) is rather narrow. It carries the posterior spiracle or metastigma (Sp^). A short cross-suture at the level of the spiracle separates off from it a small lower portion, the metinfraepisternum^ (^joSg), ^ The arrangement of the thoracic apodemes or endoskeleton lends support to the view held by some students of Odonata that these pieces are really a part of the sternum. 24 THE IMAGO [CH. just above the metacoxa (cx^). The posterior piece or metepi- merum (epm^) is broad and subtriangular. The Wing-Bases. These are outgrowths from the thoracic integument between tergum and pleurum. They are flexible, and merge into the edge of the notum. Each wing-base bears two hard swellings, the anterior and posterior colli, the former lying at the base of the costa, the latter at the base of the fused radius + media, with a weaker extension to the cubitus. Each callus is supported by a stiff process from the pleurum. Of these, the anterior or costal process (c, ap^, ap^) closely supports the costal or anterior callus, while the larger posterior process or axillary {ax^, ax^ is freely articulated with the posterior callus. The axillary forms the chief fulcrum for the action of the wing- muscles (p. 210). It will be seen that the wing of the Dragonfly agrees with that of the May-fly, but dift'ers from those of all other Insects, in having only the one true axillary, which corresponds with the second axillary of other insect- wings [157]. The Sterna. These are very much reduced (b). The meso- sternum (S^) is a small plate lying between the two mesocoxae; the metasternum {S^) is a similar plate between the metacoxae. The post-sternum has been already mentioned. The Thoracic Spiracles. There are two pairs of spiracles on the thorax, the anterior pair being the larger. The anterior spiracles or ynesostigmata (c, Sp^j are transversely elongated slits, lying between the sides of the posterior lobe of the prothorax and the mesepisternum, a little above the mesinfraepisternum, in such a position as to be more or less completely hidden from view. These spiracles are developed on special plates, the mesostigmatic laminae, which extend so as to fuse with one another across the mid-dorsal line. Thus they form a single spiracular dorsum which fuses behind with the steep vertical front of the mesepisterna. The posterior spiracles or metastigmata (Sp^) are smaller, broadly oval, and very distinct. They lie in the metepisterna, at the junction of the cross-suture with the first lateral suture. The Synthoracic Endoskeleton (fig. 8). Along all the pleural and sternal sutures the thoracic integument is folded inwards to form an internal ridge or apodeme. The pleural apodemes of the syn- thorax number five pairs, named, from in front backwards, the II] EXTERNAL FEATURES AND SKELETON 25 precostal (pc), the mesopleural (pl^), the interpleural (ip), the meta- pleural (pl^), and the posterior efnmeral (ep). The mesopleural lies cxm Fig. 8 A. Internal ventral view of skeleton of a Dragonfly's thorax, to shew endoskeleton. Adapted from Berlese. b. T.S. through metathoracic neural canal in Jsosticta sim/phx Martin ( x 15). Original, ac acetabulum ; cvt cuticle; cx^ metacoxa; cxm coxal muscles; hy hypoderm; ?ia stemai apodcmis, >