§ 2 — Description of the Skeleton. (Plate III.) — Memoir on the Dodo by Richard Owen

§ 2 Description of the Skeleton. (Plate III.)

The bones of the Dodo (Didus ineptus, Linn.) discovered by Mr. Clark, under the above circumstances, which have reached me up to the present date (December 20th, 1865) are the following:⁠—

Name.	Number of bones or parts.
Cranium and lower jaw, in parts	14
Vertebræ and pelvis	30
Ribs	22
Sternum	2
Scapular arch, in parts	7
Humerus, ulna, radius	6
Femora	5
Tibiæ	6
Fibulæ	4
Metatarsals	4
	——
Total number of parts of skeleton of the Dodo	100
	===

The known characters of the skull and metatarsus of the Didus ineptus served to identify those bones as belonging to that species: the agreement in relative size, colour, condition, and locality left no room for hesitation in referring the other bones in the above list to the same species23. They belong, however, to four or five individuals varying somewhat in size. With the bones of the Dodo were the end of the lower jaw of a broad-billed Parrot, two bones (radius) of a small Mammal, and part of the skull of a large Tortoise24.

To the description of the Dodo’s bones I now proceed.

Vertebræ. (Plates III., IV., V., VIII., XI.)

The dorsal vertebræ are chiefly represented, in this series of bones, by three which are anchylosed together by their bodies and neural arches (Pl. V. figs. 1–5): the posterior articular surface of the body of the last of these vertebræ (ib., fig. 4, c) is subquadrate, longer in the vertical than the transverse direction, concave vertically, convex transversely, almost fitting, but being rather too small for, the anterior articular surface of the body of the first of the sacral series (Pl. VII. fig. 1, c). The difference is such as to indicate that only one dorsal vertebra may have intervened; and I conclude that the last of the three coalesced vertebræ is the penultimate dorsal. The anterior articular surface of the foremost of the three (Pl. IV. fig. 1, c) is 11 lines in transverse, and 4 to 5 lines in vertical diameter: it is concave transversely for the middle three-fifths, and convex transversely at the two outer fifths of its extent: it is more or less convex vertically throughout its extent. The bodies of these vertebræ are compressed and wedged-shaped, slightly expanded at their coalesced ends, produced below into subquadrate hypapophyses in the first and second (Pl. V. fig. 1, hy); while this process is restricted to the fore part (ib. hy 3), or may be represented only by a slight anterior production of the lower edge of the wedge, in the third (ib. fig. 5, hy 3).

The hypapophysis of the first of the three expands at its termination (Pl. IV. fig. 1, hy), with the hinder angle bent back to coalesce with the front one of the next hypapophysis, which is somewhat longer, and bent forward with a similar terminal expansion: a full elliptical space is intercepted by this terminal confluence of these hypapophyses (Pl. V. figs. 1 & 5, hy). Each vertebra shows an elliptical articular cavity (ib. figs. 1 & 5, p, p 3) for the head of the rib, near to the anterior articular surface; the long axis of this costal surface is directed from above obliquely downward and forward. The surface of the rib’s tubercle cuts obliquely the lower part of the free end of the diapophysis (Pl. IV. fig. 1, d).

The neural arch circumscribes a canal the anterior outlet of which (ib. fig. 1, n) is oval with the small end downward, 5 lines in vertical, and 3½ in transverse diameter: the sides of the neural canal slightly project inward above the lower third: the posterior outlet (Pl. V. fig. 4, n) is more regularly elliptical in form, and rather narrower in proportion to its vertical diameter. The neurapophysis sends off from the outer and fore part of its base a stout process, which expands and divides into zygapophyses (Pl. IV. fig. 1, z) and diapophyses (ib. d); the articular surface of the former is of a full oval shape, flat, looking obliquely upward and inward; the diapophyses extend outward and a little backward: the articular surface for the tubercle of the rib is transversely elliptical and nearly flat. The hinder part of the neurapophysis expands into the postzygapophyses: these have coalesced with the præzygapophyses in the succeeding vertebra (Pl. V. fig. 2, z), as has happened also between this and the third vertebra. In the last of the three vertebræ the postzygapophyses are entire (ib. z 3), and show very slightly concave, oval articular surfaces, looking obliquely downward and outward (ib. fig. 4, z). The conjugational foramina, continuously surrounded by bone, are a full ellipse, and large, the anterior one (ib. figs. 1 & 5, f) being 5½ lines in vertical diameter; the second (ib. f′) is somewhat less: these foramina are also rather larger in one of the specimens than in the other. The length of the three coalesced dorsals is the same in both, viz. 2 inches 3 lines. The neural spines have run together into a continuous ridge in fig. 1, ns; in fig. 5 the summit is broken off in both, leaving only the anterior angle of the foremost entire; in both this inclines forward; the hinder border of the third vertebra (fig. 1, ns) has the same vertical parallel as the back part of the centrum. The anterior margin of the base of the spine shows a rough surface for the attachment of ligament (Pl. IV. fig. 1, ns). A small foramen behind the base of each of the coalesced zygapophyses (Pl. V. fig. 2, z z) leads to a canal descending to the neural one, and indicates superiorly the limits of the otherwise continuously ossified neural arches.

In the series of detached vertebræ, one (Pl. V. figs. 6 & 7) indicates by its neural spine and hypapophysis a position at the base of the neck. The centrum is barely an inch in length; its anterior surface (ib. fig. 7, c) is narrow vertically, broad transversely; both fore and hind surfaces indicate freedom and extent of flexure. The hypapophysis has a broad, bituberculate base (ib. hy), but is limited in fore and aft extent to the middle third of the under surface of the centrum: its length is shown in fig. 6, hy. The parapophysis (fig. 7, p) is slender, and expands at both attachments, with an indication of a terminal surface. The diapophysis (d) has a larger costal surface: it sends forward a convex ridge midway between the di- and zygapophysis (z). The neural canal (fig. 7, n) has wider and more fully elliptical outlets than the hinder dorsal vertebræ, in relation to the greater extent of motion at the fore part of the series. I conclude that a free pleurapophysis (pl) existed, indicating the present to be the first of the dorsal series, as shown in Pl. III. The neural spine is short, broad, obtusely pointed, with a vertically oblong syndesmotic surface (fig. 7) before and behind. Each postzygapophysis (fig. 6, z′) supports an anapophysial tubercle (a).

A cervical vertebra from a position just in advance of the above has lost the neural spine, but retains the hypapophysis. This process (ib. figs. 8 & 9, hy) is compressed and directed obliquely downward and forward for an extent of 6 lines; the extremity is rounded: the length of the centrum of this vertebra is 1 inch 3 lines; the anterior articular surface is longest transversely, and concave in that direction, convex vertically; the proportions and curvatures are transposed in the posterior surface (fig. 9, c). The parapophysis (ib. p) is continued from the anterior border of the centrum to the middle; it is a depressed plate, confluent with the rib (ib. d). The diapophysis forms a short, obtuse projection above its anchylosis with the rib (ib. pl): this projects backward 7 lines in length, terminating obtusely, and circumscribing a vertebrarterial foramen (ib. v) of a full elliptic shape, 5½ lines in long diameter. The surfaces of the præzygapophyses (z) are larger, and look more upward and less inward, than in the preceding and the dorsal vertebræ: they are very slightly concave. Those of the postzygapophyses (fig. 8, z′), with a downward and slightly outward aspect, are in a similar degree convex. The neural canal, as usual in the cervical series, expands at its outlets, most so posteriorly (fig. 9, n); the middle of the upper surface of the neural arch is impressed by an elliptical, rough, ligamentous surface, which slightly rising in the middle is the sole indication of a neural spine. The upper surface of each postzygapophysis developes a tuberous anapophysis (figs. 8 & 9, a).

The three cervicals that succeed the axis show progressively sinking neural spines, which subside in the six following vertebræ (Pl. III.). The third cervical has also the hypapophysis (Pl. XI. fig. 3, hy).

In all the other cervicals of the present series the hypapophysis is wanting, but each parapophysis developes a plate (Pl. V. figs. 10 & 11, Pl. VIII. fig. 1, p) to form the sides of the hæmal canal through which the carotids ran; and the position of such vertebræ in the cervical series is indicated, respectively, by the degree of convergence of these processes, in none of which, where entire, have they met so as to circumscribe the canal: in some of these vertebræ, however, they are mutilated. They differ chiefly in the position and shape of the anapophyses (fig. 10, a), which advance from above the postzygapophyses (z′), converging towards the middle of the upper surface of the neural arch, being arrested, save in one instance, at the sides of the ligamentous surface occupying the common position of the base of the neural spine.

In the axis vertebra (Pl. V. figs. 12 & 13) the posterior articular surface, concave vertically, and 3 lines in that extent at its middle part, is very convex transversely, being continued upon the sides of the posterior part of the centrum; a thick obtuse hypapophysis (fig. 13, hy) descends below this surface: the anterior or odontoid surface presents the usual form in birds; the odontoid process (ib. x) has a pit at its apex. The prezygapophyses (fig. 12, z), of very small size, project from the outer and fore border of the neural arch, with their articular surface looking outward and slightly upward; a ridge is continued from their back part to the base of the postzygapophyses: the surface (fig. 13, z′) in these, 4½ lines in long diameter, is three times the size of the anterior one; it is concave transversely, and looks downward and a little outward. The anapophyses (ib. fig. 12, a) are large tubercles rising above the articular surfaces. The base of the neural spine, 9 lines in length (ib. ns), is coextensive with the neural arch; the spine rises posteriorly to a height of 6 lines, with a thickness of 2 lines, having a convex upper margin (Pl. III.).

The relative size and position of the cervical vertebræ, as coadjusted in the position and degree of flexure of the neck represented in Sir Hans Sloane’s life-size painting of the Dodo, in the British Museum, are given in Plate III. with the varying proportions of the pleurapophyses and other processes.

Ribs. (Plates III. & IV.)

The specimens of ribs include both vertebral and sternal portions; that which appears to be the second or third on the right side (Pl. IV. figs. 7, 7 a) is 4 inches 4 lines in length (following the outer curve), and expands to a breadth of 7 lines at its lower part; the interval between the articular surfaces of the head and tubercle is 6 lines. The appendage (ib. a) has coalesced with the middle of the hind margin of the shaft. The neck is compressed, with a thin upper margin; the lower one is continued with a curve upon a strong internal buttress-like ridge (ib. b), which runs to near the fore part of the flattened body of the rib, where it meets the ridge continued from the tubercle, about 2 inches down the rib: there is a shallow channel between these ridges, contracting to their confluence. The inner surface of the rib is impressed by a deeper and broader channel behind the buttress: the posterior border expands in the form of a triangular plate, with a base of about an inch in extent, due to the complete confluence there of the epipleural process. The anterior border is thicker, and is almost straight. Towards the sternal end the pleurapophysis contracts and thickens, terminating in a rough syndesmotic elliptical surface, 3 lines by 2 (fig. 7, f), for the attachment of the hæmapophysis or sternal rib.

A vertebral rib (ib. fig. 2) which is entire, measures 9 inches in length (following the outer curve). The head and tubercle are at the same distance as in the preceding, but the tubercle is broader. The characters of the body of the rib are very similar; but it is narrower, not attaining a breadth of 5½ lines at its lower end; the narrowing and thickening to the articular surface for the sternal rib is more gradual.

A last vertebral rib is adapted, by the longitudinal extent and partial division of the tubercle, to the vertebra which forms the first of the coalesced series of sacrals; and the body of the rib, instead of preserving the regular outward curve of the antecedent ones, is more suddenly bent soon after it emerges beyond the margin of the ilium; the lamelliform part thence continued is straighter, and, moreover, shows upon its outer surface a flattened facet, indicative of pressure or friction by the movements to and fro of the thigh over a rib in such position. Beyond this surface the rib curves in a way not shown in the other specimens; the distal end has the flat syndesmotic articular surface to which had been attached a hæmapophysis not reaching the sternum. In this last (eighth) free rib there is no epipleural process, nor any definitely marked ligamental surface on the posterior margin indicative of the attachment of such process.

The body of a posterior vertebral rib (Pl. IV. fig. 10) shows a fracture which has been healed, with some irregular ossific deposit on the inner surface. All the ribs have a pneumatic foramen (ib. figs. 2, 7, 8, p) at the fore part of the neck, near the base of the tubercle.

The eight left vertebral ribs (Pl. III.) and the five right ones do not, either of them, constitute a consecutive series, but have come from different individuals, of different sizes, as exemplified in the third rib figured in Plates III. and IV.

The sternal ribs (P. IV. figs. 3 & 12) are characterized by the two facets, nearly or quite meeting at an open angle, into which their sternal end expands (ib. fig. 3, c). One of these ribs, which is entire, shows the single, elliptic syndesmotic surface at the opposite end (ib. b); it is 3½ inches in length, with a greatest breadth of 5 lines, and is straight. Another and longer specimen (ib. 12) shows a moderate degree of curvature. A third specimen is 6 inches in length: the proximal end has a breadth of nearly half an inch (the penultimate rib in Pl. III.).

Five successive sternal ribs are indicated by gradational size and curvature, and a sixth, which does not reach the sternum. Before describing this bone I shall proceed with the account of the sacral vertebræ, and the expanded hæmal arches of such as complete the pelvis.

Pelvis. (Plates III. & VII.)

The pelvis of the Dodo is chiefly remarkable for the flatness and great breadth of the posterior half, corresponding with the characteristic proportions of that part of the body in Pl. I. fig. 2, and in the old woodcuts of the Dutch “Dodaersen”25. It includes sixteen coalesced sacral vertebræ, with which the iliac bones are continuously confluent.

The first sacral shows the transversely extended and concave articular surface of the centrum (Pl. VII. fig. 1, c); the subcircular pit (ib. p) for the head of the rib is behind the middle of the side of the centrum, at its upper part; the inferior surface is ridged lengthwise; and a transverse low but sharp ridge defines the posterior boundary, the depressions in front of which indicate the hindmost origins of the subvertebral muscle (longus colli?). The anterior outlet of the neural canal (ib. n) is subcircular in one specimen, vertically elliptic in others, and 3 lines or less in transverse diameter. From the sides of the neurapophyses stretch out the strong buttresses of bone which blend with the under part of the ilia, giving off from the fore part of their base the præzygapophyses (ib. z), and from the back part of their apex the surface (ib. d), or part of it, for the tubercle of the last moveable rib, the ilium in the latter variety affording the rest of that surface. The fore part of the strong neural spine (ib. ns) is roughened by a syndesmotic surface; it rises to a height of 14 lines, curving forward, and is confluent at its summit with the approximated anterior margins of the ilia. A continuous track of bone, forming a smoothly obtuse longitudinal ridge, represents the summits of the succeeding sacral spines (ib. fig. 2, ns) to the hindmost vertebra of the series, without any trace of their primitive division; but this track rises, posteriorly, above the shallow channel on each side, in which are the foramina (ib. o), indicating most of the constituent vertebræ.

The second sacral vertebra abuts against the ilium by a pleurapophysis (ib. fig. 1, pl 2), as well as a diapophysis (ib. d 2); but the former is a slender, straight filament, or narrow plate of bone, confluent at both ends.

In the next two vertebræ the pleurapophysis (ib. pl 3 & 4) assumes more breadth and robustness, but is short and straight, abutting against the inner surface of the ilium an inch in advance of the acetabulum. The first of these rib-buttresses inclines forward, and is completely confluent with the ilium; the thicker one (ib. pl 4) has retained part of its primitive ligamentous attachment to the ilium: the proportions of both are subject to some variety.

These are succeeded by three or four vertebræ in which the pleurapophysis is not developed, the attachment to the ilia being by diapophyses only (ib. d d), which are short slender lamellæ, directed upward and backward; below and between them are the double orifices for the separate motory and sensory roots of the sacro-spinal nerves. In the next vertebra the pleurapophysis (ib. pl 8) reappears, longer but more slender than in the fourth sacral, extending obliquely backward, and expanding at its extremity to abut against a prominence on the underside of the ilium, opposite the hind part of the acetabulum, with which prominence the rib has completely coalesced by an expanded end. The under part of all these vertebræ is traversed by a sharp median longitudinal ridge, which is more feebly and interruptedly continued to near the end of the sacral series.

Eight vertebræ, abutting by diapophyses only (Pl. VII. d d) against the ilia, succeed the one last described; their coalesced bodies are less than half the breadth of those of the preceding vertebræ: they gradually diminish in depth to the last, without loss of breadth. The diapophyses proceed obliquely outward and backward, are lamelliform, about 9 lines in length, and intercept oblong cavities of the same extent and direction, into which open the orifices (ib. fig. 2, o) noticed on the upper surface of that part of the pelvis. The articular surface of the body of the last sacral is transversely elliptic, 4 lines by 2 lines, and very slightly convex. The outlet of the neural canal, above it, is circular, and about a line in diameter, the whole vertical extent of the last sacral being 5 lines, while that of the first sacral is 2 inches 2 lines.

The ilium is divided, as usual, into two parts by the ridge on its upper or outer surface (ib. fig. 2, r), extending obliquely backward to behind the acetabulum—the anterior division being narrower and concave, the posterior broader and convex but in a minor degree. The anterior (slightly thickened) border of the ilium is curved with the convexity forward, extending 8 or 9 lines in advance of the fore part of the neural spine of the first sacral vertebra. The ilia almost meet above that of the second and third sacrals, with which they coalesce, and then diverge to the oblique boundary ridge, which is thence continued, in some with an angular bend, more directly outward. At this angle the bone is so confluent with the sacrum that the orifices leading to the ileoneural canals26 are almost or quite obliterated. These canals are, here (ib. i ï), the longitudinally extended cavities intercepted between the fore parts of the ilia and the continuous coalesced sacral spines and diapophyses, widening to their anterior outlets. The extent of that part of the ilium in advance of the acetabulum is 3 inches 8 lines; the breadth at its middle part is 2 inches. As the ilium approaches the acetabulum it increases in thickness, and is grooved at the outer margin by a vessel which leaves impressions of its ramifications upon the upper concave surface of the bone (ib. fig. 2, 62). The acetabulum (ib. a a) is circular, 11 lines in the diameter of its outlet, 9 or 10 lines in that of its inner circumference, being widely open, as usual in birds, towards the cavity of the pelvis; the trochanterian surface (ib. t t) above the acetabulum is elliptic, with the long axis lengthwise, 9 lines by 6 in its diameter, with its upper border sharp and produced; the anterior border (ib. b) of the acetabulum is slightly produced; the position of this articular cavity is about midway between the fore and hind ends of the pelvis. The oblique external ridge of the ilium terminates in the outer margin of the broader part of the bone (ib. r′), 7 lines above the sharp and prominent margin of the trochanterian surface (ib. t). The ilia have diverged from each other for the extent of an inch and a half behind the beginning of the boundary line (ib. r), which interval is occupied exteriorly by lateral ossification from the neural spines to the diapophyses of that part of the sacrum: the mesial borders of the ilia (ib. fig. 2, ₆2′) slightly converge to the fifteenth sacral vertebra, where they are separated by an interspace of 1 inch, and then again diverge to the last sacral; they coalesce with the diapophyses (ib. fig. 2, d d). The inner or under surface of the ilium is thickened into a kind of buttress (ib. fig. 1, e), terminating behind the ischiadic foramen. The breadth of the iliac bones and intervening sacrals, 1 inch behind the acetabulum, is 5 inches; at the back part of the pelvis it is 4 inches. The outer border of the posterior part of the ilium (ib. fig. 2, g) projects as an obtuse ridge above the ischiadic foramen and the succeeding expanded and confluent part of the ischium (ib. ₆3), which is vertically concave externally: the ilium, ischium, and pubis (ib. fig. 1, ₆4) have completely coalesced around the acetabulum. The pubis, which in this part is 7 lines thick, contracts as it becomes free to a diameter of 4 lines; it is smooth and convex below, and has been broken off near the acetabulum on both sides; the fracture shows its pneumatic structure. The ischium, as it recedes from the acetabulum, contracts to a trihedral column, with a vertical diameter of 4 lines; it is concave outwardly, convex inwardly, and suddenly expands below, about an inch from the acetabulum, to form part of the posterior boundary of the obturator foramen (ib. fig. 1, f), which is 9 lines in length, and is situated one half in advance of, and the other half beneath, the ischiadic foramen (ib. m). This latter is oval, with the large end forwards, 1 inch 3 lines by 10 lines in its principal diameters. Behind this foramen the ischium is confluent with the ilium for an extent of 2 inches, or perhaps rather more, as the posterior margin of the pelvis is not entire in any of my specimens. The inner surface of the ischium forms a low, obtuse longitudinal ridge towards the pelvic cavity, losing thickness as it recedes from the acetabulum. The chief pneumatic foramina in the pelvis are on the inner surface, above the acetabulum, behind the trochanterian articulation, and behind the iliac confluence of the last sacral pleurapophyses,—also at the hinder part of the ilium, on each side of the transverse buttress (ib. e) near its posterior junction with the ischium. The prærenal fossa (between pl 4 & pl 8, fig. 1) is deep and subdivided by the diapophysial plates: the postrenal fossa is wide and shallow.

Sternum. (Plates III., IV., VI., XI.)

Of this instructive and determinative bone there are two specimens, the one most entire (Pls. III., IV. fig. 4, & VI.) measuring in a straight line, from the costal process to the hind border, 7 inches. The extreme breadth between the lateral processes (Pl. IV. h) is 4½ inches; from this diameter the bone contracts anteriorly to a breadth of 3½ inches at the costal processes (ib. d), and posteriorly it contracts more rapidly to an obtuse, horizontally flattened apex (Pl. VI. fig. 3). The anterior border of the sternum (Pl. IV. fig. 4) is widely and rather deeply emarginate at the middle (e), less deeply so on each side: the breadth of the mid notch (b e b) is 1 inch 9 lines, that of each side notch (b d) is 1 inch 2 lines. The sternum is deeply hollowed above (Pl. XI. fig. 4), correspondingly convex beneath (ib.); the keel (s) is low and thick, commencing by a pair of broad obtuse ridges (Pls. IV. fig. 4, & VI. fig. 1, r r) from the mesial ends of the outer walls of the coracoid grooves (ib. b′), which gradually rise from the surface of the bone as they extend backward, converging to form the beginning of the keel about 2 inches from the anterior emargination (e): the keel gains a depth of ¾ of an inch at the middle of the sternum, then gradually sinks to the level of the bone, as it extends backward, at 1½ inch from the hind end (Pl. VI. fig. 3), a little increasing in thickness as it subsides: its free border describes a pretty regular convex curve (Pl. III.); it is thick, flat, partially canaliculate: the sides of the base of the keel expand, to be continued gradually into the body of the sternum (Pl. XI. fig. 4). Behind the costal surface (Pl. VI. c), on each side, extends a lamelliform process (Pls. III. & VI. h), ½ an inch in breadth, upward and a little outward, slightly expanding to its free termination, which, however, is not entire in either specimen: the longitudinal extent of this characteristic process, where it is best preserved, is 1 inch; it is conjecturally restored in Plate III.; it answers to the ectolateral process (h) of the gallinaceous sternum (Pls. III. & XII. fig. 3): there is no trace of an entolateral process (ib. i). The thin margin of the Dodo’s breast-bone, behind the ectolateral process (Pls. III. & VI. h), is entire and uninterrupted to the obtuse apex, and the body of the sternum is imperforate: the notch (f) behind the process (h) represents the ectolateral notch of the gallinaceous sternum (Pl. XII. figs. 1 & 3, f). The costal border (Pl. VI. fig. 2, c) is 1 inch 9 lines in extent, and 6 lines across its broadest part; it shows articular surfaces for five sternal ribs, of which the four posterior (2–5) are bilobed, the anterior one (c 1) simple, and limited to the outer half of the border; the second sternum shows some variety in this respect: the deep interspaces, in both, are perforated by pneumatic foramina. The costal process (d)27 in advance of these surfaces expands, as it rises upward and a little outward and forward, to the extent of nearly an inch; the hinder and outer side is impressed by a concavity, continued from the costal border; the inner side is smooth and convex: it is not quite entire on either side. The coracoid grooves (Pl. IV. fig. 4, b b′) are small in proportion to the sternum, and are divided from each other by an interspace of about an inch; the outer wall of the groove (b′), 9 lines in extent, is moderately produced and convex; it appears to be a continuation of one of the initial ridges (r) of the keel: the inner wall of the groove (b) is deeper, and is formed by the obtuse angle of the anterior border of the sternum, between the medial and lateral emarginations. External to each coracoid groove is a large elliptical pneumatic foramen (p) or depression. There is no episternal process. On the convex outer surface of the body of the sternum the “pectoral” ridge (Pl. VI. fig, 1, k)28 is feebly indicated, extending from the outer end of the coracoid groove backward and inward to near the posterior third of the keel. The concave surface of the sternum (ib. fig. 2) shows a number of small pneumatic foramina, chiefly along the middle line to near the posterior third. Behind the costal border the substance of the sternum gradually increases in thickness from the sharp lateral margins to the middle, above the base of the keel, and shows there a fine pneumocancellous texture (Pl. XI. fig. 4).

Scapular Arch. (Plates III. & VIII.)

This consists of the scapula (Pl. VIII. figs. 6, 7, 8 & 9, 51), coracoid (ib. figs. 4 & 5, 52), and clavicle (ib. 58), the latter ending in a point and here tied by ligament to its fellow, to form a furculum. I have received the elements of this arch in three conditions:⁠—one in which the bones, though of full size, are separate; a second, in which the scapula and coracoid are confluent, but the clavicle distinct; a third, in which the three bones are confluent at the ends converging to the humeral articulation. The scapula (ib. figs. 6, 7, 8 & 9, 51), 3 inches 7 or 8 lines in length, has the usual sabre-shaped body, slightly expanding and decurved at its free extremity, the breadth of which is 7 lines: it terminates obtusely: varieties of shape are shown in figures 6 & 8. The outer surface of the bone, at the two posterior thirds of its extent, is slightly concave and marked by muscular attachments; the inner surface of that part is smooth and slightly convex: the bone increases in breadth, with some diminution of thickness, towards the articular end, and is remarkable for sending off from the lower border, at 7 or 8 lines from that end, a short process (ib. 51); between this process and the articulation the breadth of the bone is little more than 3 lines; the breadth of the articular end is 9 lines. Nearly one-half of it is occupied by the almost flat, subcircular humeral surface (fig. 8, a), with a diameter of 4½ lines, and directed upward, outward, and a little forward. From this is continued an oblong, much narrower coracoidal surface, beyond which the acromial process (fig. 6, c) extends forward, curving toward the coracoid, and terminating obtusely.

The coracoid (ib. figs. 4, 5, 8 & 9, 52), averaging a length of 3 inches 7 lines, expands to a breadth of 1 inch 3 lines at its sternal end (52), of which the articular surface (e) occupies an inch; the non-articular part forms the outer angle (m), and extends in advance of the pneumatic foramen (Pl. IV. fig. 4, p) at that part of the breast-bone: the outer border which extends from this free angle to the body of the bone, into which it subsides, at one-third of the extent of the bone, is sharp; the inner border is obtuse to near the inner angle (Pl. VIII. figs. 4 & 5, n). The outer surface of the expanded sternal end is smooth and convex; the inner surface is flatter and more irregular, perforated by pneumatic foramina; the diameter of the subcylindrical part of the shaft is 4 lines: the extremes of difference in the distal expansion of the coracoid are shown in figs. 4 & 8, 52, Pl. VIII. A muscular ridge and rough surface (ib. fig. 9, r) mark the back part below the middle of the shaft. The bone then expands to its upper articular end, which is obliquely truncate from within outward: it shows, first, the oblong surface for the scapula, which is extended upon the inner prominence of that end; next, the larger and full oval surface for the humerus (h), from which the thick, obtuse, inner continuation of the scapular end projects inward, forward, with a slightly upward curve, and shows the narrow oblong surface for the articulation and ultimate confluence of the clavicle (58). The coracoid unites with the scapula at an angle of 100°.

The clavicle (ib. figs. 4 & 5, 58), at its scapular end, is slightly expanded, compressed, with an obtuse recurved termination articulating with the above-named surface of the coracoid, and in one instance coalescing therewith, and by extended ossification with the “acromion scapulæ” (ib. figs. 8 & 9). As the clavicle descends it curves slightly and contracts to a point. The angle at which the pair meet is shown in figs. 4 & 5.

Bones of the Wing. (Pls. III. & VIII. Figs. 12–17.)

Of the humerus the series contains two specimens, both measuring 4 inches 3 lines in length, one right, and the other left (Pl. VIII. figs. 12–14), but differing slightly in their proportions and in colour—one being of the olive-brown tint with which most of the bones are stained, the other black. The articular head (ib. a) is an elongate oval convexity, with the larger end toward the radial side, prominent toward the back and rather flattened toward the front of the bone, which there swells out beyond the base of the articular surface. The radial tubercle is small, and descends from the radial end of the head for about 5 lines; the pectoral process (ib. b) is triangular, obtuse, short, and bent, or directed toward the front side of the bone: the ulnar tuberosity (ib. c) is more produced in that direction; it is oblong, obtuse, with its base impressed by a large pit both above (fig. 12, h) and below—the lower one (ib. g) being the deepest, and perforated by a pneumatic foramen; the convex, broad, ulnar border of this tuberosity has two slightly produced processes, an upper or posterior (ib. fig. 12, c) and a lower and internal (ib. g), which is the smallest. The breadth of the proximal end of the humerus, across the tuberosities, is 1 inch 5 lines, beyond them the bone contracts to a smooth subcylindrical shaft, showing at the back part of the proximal third a longitudinal ridge (fig. 12, r), half an inch in length; it gradually expands at the distal third to a breadth of 10 lines, where the articulations offer the usual avian characteristics of the elbow-joint. The head of the humerus is occupied by a fine cancellous structure: into the large vacuity below this, crossed in the section figured (Pl. XI. fig. 5) by a transverse slender bar of bone, the small pneumatic foramina at the bottom of the wide and deep fossa for the axillary air-cell open. The part of the hollow proximal end giving off the pectoral and other processes for the attachment of muscles is strengthened by similar abutments. The pneumatic cavity of the main part of the shaft of the humerus is simple, with a compact wall thicker than at the ends of the humerus, but not exceeding that which is characteristic of the long air-bones in birds. The portion of the distal end chiefly serving for muscular attachments and the antibrachial articulation are also cancellous.

The radius (Pls. III. & XII. fig. 15) is a straight and slender bone, 3 inches 1 line in length, and 2 lines in chief diameter of the shaft. The proximal articular surface is subcircular, 3 lines in diameter, moderately concave; the distal end expands to the same extent, but is compressed, as usual.

The ulna (Pls. III. & VIII. figs. 16 & 17) is 3 inches 1 line in length, of the usual ornithic character, with a well-defined, narrow, elliptic, rough muscular depression, 8 lines in length (fig. 16, c), extending upon the shaft from below the anterior or palmar angle of the proximal articular surface. This bone has no pneumatic foramen; the orifice for the medullary artery is above the middle of the same palmar surface, the canal inclining distad. The shaft of the bone is nearly straight; the back or anconal surface, which is slightly convex, shows feeble impressions of the attaching ligaments of the alar plumes, which are represented in all the figures of the entire or living bird. A second ulna is 3 inches 3 lines in length.

There was no carpal or pinion bone in the collection of remains submitted to me: this part of the wing is conjecturally restored in dotted outline in Plate XV.

Bones of the Leg. (Pls. III., IX., X. & XI.)

Of the five femora in the above defined series of remains of the Dodo, two measure 6 inches 3 lines in length; one (Pl. IX.) is 6 inches 4½ lines; the shortest is a little under 6 inches, with proportionate differences in the diameter of the shaft. All of them show a small pneumatic foramen (Pl. IX. figs. 1 & 2, p) on the inner side of the anterior ridge of the great trochanter (ib. c), and on the same transverse line with the head of the bone. This part shows an oblong depression (ib. figs. 2 & 3, a) for the “ligamentum teres” at the upper and back part. The articular surface on the same aspect of the neck (ib. fig. 3, b), adapted to the trochanterian prominence of the pelvis (Pl. VII. t), is well-defined. The trochanter (Pl. IX. fig. 1, c) rises, ridge-like, above the level of the head, and is continued from behind the middle of the articular surface on the neck, forward, with a convex outline upon the fore and outer part of the shaft, where it gradually subsides; a narrow intermuscular ridge (ib. fig. 1, r), inclining to the middle of the fore part of the shaft, is continued from the trochanterian one. The small trochanter (ib. fig. 3, d) is a small subcircular tuberosity, in some specimens a ridge, 3 to 4 lines in length, on the inner side of the shaft, about an inch below the head. The muscular impressions on the fore part of the bone are well defined. A minute medullary canal (ib. fig. 3, m) perforates the middle of the back part of the shaft; the popliteal fossa (ib. fig. 3, o) shows a few small pneumatic orifices; a triangular rough flat surface divides the fossa from the outer condyle. Above the fibular depression (ib. fig. 3, g) there is a well-defined, slightly raised, rough surface (ib. k) for the head of the ectogastrocnemius muscle. The ridge (ib. n) extending to the back part of the inner condyle is not sharp; the rotular groove (ib. fig. 1, p) is deep and moderately wide, with the inner boundary, formed by the narrow anterior part of the inner condyle (ib. fig. 5, e′), most produced. The breadth of this end of the longer femora is 1 inch 9 lines; the character of the distal articular surface is shown in Pl. IX. fig. 5.

The head, neck, and great trochanter (Pl. XI. fig. 6) are occupied by a pneumatic cancellous structure, with a thin compact wall on the upper part and sides: this begins to gain thickness at the under part of the neck and at the lower and back part of the trochanter, the compact wall acquiring a thickness of a line at the beginning of the shaft, where the cancellous structure is confined to the outer side of the pneumatic cavity; this structure gives way to a few delicate filaments of bone crossing the cavity of the major part of the shaft, and is not resumed until the bone expands to form the distal condyles (ib. fig. 7).

The five tibiæ of Didus in the same collection range in length from 8 inches 8 lines to 9 inches. The procnemial ridge (Pl. X. figs. 1, 2, 4, p) is a triangular plate, with the base longest and the apex rounded off: it inclines outwardly, and does not extend much more than half an inch from the level of the proximal end of the bone: the length of its base rather exceeds an inch: on its inner side a triangular muscular surface is well defined by an irregular inferior line or ridge (ib. fig. 2, n). The ectocnemial process (ib. figs. 1, 3, 4, e) is thicker, shorter, and terminates roughly and obtusely. There is a low, narrow ridge (ib. fig. 2, g), about half an inch in length, on the inner side of the proximal end of the shaft, beginning about 9 lines below the articular surface at that end. The fibular ridge (ib. figs. 1 & 3, h), beginning 1 inch 8 lines from the proximal end, extends about 2 inches down the outer side of the shaft. The epicnemial ridge (ib. figs. 1 & 4, k) is obtuse, and but little produced above the upper articular surfaces or condyles (t d) of the tibia: the breadth of that end of the bone, in the longest specimen, is 2 inches 3 lines. The tendinal canal at the fore part of the distal end is bridged by bone (ib. fig. 1, l), and is situated on the inner half of that aspect of the shaft; the lower opening is subcircular and close to the anterior end of the inner lower condyle (ib. a), which is more produced forward than the outer one (ib. b). Their hind ends project very little beyond the level of that aspect of the shaft of the tibia. An intermuscular ridge (ib. fig. 1, r) strengthens into a tuberosity (r′) at the inner side of the tendinal groove.

The cancellous structure in the tibia is limited to an extent of about half an inch below the proximal articular surfaces (Pl. XI. fig. 8), and to about an inch and a half from the distal end of the line (ib. fig. 9): the shaft is occupied by a large air-cavity, with a compact wall of half a line in thickness at the upper third, gradually increasing to about a line at the lower fourth, until the cancellous structure is reestablished; the transverse direction of a plate of this structure indicates the extent of the original distal epiphysis of the tibia (fig. 8).

The fibula (Pl. X. figs. 6–8) presents the usual ornithic characters of the bone: it varies from 4 inches 4 lines to 4 inches 6 lines in length, with a greatest proximal breadth of 8 lines. No adequate gain would result from a detailed description or comparison of this bone; and the rest of the bones of the foot have received every requisite attention in this way in the excellent work on the Dodo and its kindred, already quoted. A longitudinal section of the metatarsus, taken in the direction from side to side (Pl. XI. fig. 10), shows the loose cancellous texture of the common epiphysis of the three long metatarsals, and the remnant of their contiguous coalesced walls reduced to a thin lamella of bone. As the moiety of the bone figured is the posterior one (of the left metatarsus), the usual oblique position of the middle metatarsal (iii), with its proximal end nearer the back part and its distal end nearer the fore part of the coalesced series, produces a corresponding direction of the section, with narrowing and termination of the exposed part of the medullary canal about one-third from the distal end of that metatarsal. The medullary canal of the outer metatarsal (iv) is wider and descends lower before the breaking up of the inner surface into decussating lamellæ or filaments, than that of the inner metatarsal (ii): the peripheral compact wall of the inner is twice the thickness of that of the outer metatarsal. I may remark that the more posterior position of the middle metatarsal at its proximal end, from which and the corresponding part of the common epiphysis the calcaneal process is developed, is related to the greater share taken by the middle toe in the act of walking and scratching. I will only remark that of the four metatarsals of as many Dodos in the present series, one exceeds by a line the length of that figured in plate xi. op. cit., and one falls short thereof to the same trifling amount.

Skull. (Plates III. & XI. Fig. 1.)

Of the skull of the Dodo, the series of bones transmitted to me include the cranial part with the detached upper mandibular bone (more or less mutilated) of two mature birds, and the lower mandible of three individuals. In the latter the dentary elements (Pl. XI. fig. 1, 32), confluent at the “gonys,” are distinct from the hinder halves of the rami formed by the confluent, or perhaps connate, articular, surangular and angular elements (ib. 31): if the “splenial” were ever distinct, it has coalesced with the dentary, where its upper boundary is indicated by a linear groove or series of small foramina.

In size, shape, and all other characters of these important evidences of the specific nature of the remains from the Mahébourg morass29, they agree with those of Didus ineptus detailed in the ‘Proceedings of the Zoological Society’ for January 11th, 1848 (part xvi. pp. 2–8), and in the work entitled “The Dodo and its Kindred,” pp. 76–96.

The occipital condyle (ib. 1) presents the same hemispheroid or reniform shape, with the median vertical notch or depression above. The upper margin of the foramen magnum is broad, as it were excised, with the sides slightly prominent. The superoccipital foramen is present in both specimens, as in the one originally described (Proc. Zool. Soc. part xvi. p. 2). This foramen also exists in Owls and Parrots, but not in all Pigeons; the Didunculus (Pl. III. fig. 2) shows no trace of it; I have also failed to find it in the skull of a Crown-pigeon (Goura coronata). The superoccipital ridge is defined by the subsidence of the surface beneath it being continued directly from the upper, almost flat, smooth surface of the cranium: the middle part of the ridge is more produced than the angles. In the great breadth of the occipital surface compared with its depth, in its flatness from side to side, and its aspect backward and a little upward, Didus most resembles Dinornis. The basioccipital curves downward, and unites with the basisphenoid in developing the pair of larger tuberosities (Pl. XI. fig. 1, 5), which terminate about ½ an inch below the occipital condyle. There is nothing of this structure in the Columbine cranium. In one of my Dodo’s skulls there is a pair of small tubercles between the larger basioccipital ones; these are not developed in the other cranium. The basisphenoid is subquadrate, and flattish below, impressed by a shallow median longitudinal channel.

The hypoglossal nerve escapes by two small foramina on each side of the base of the condyle; external to these is the vagal foramen; still more external is the depression (ib. a) perforated below by the entocarotid, glossopharyngeal, and sympathetic, above by the tympanic vein. The entocarotid canal opens into the hind part of the sella or pituitary fossa: the vagal canal begins within the skull, above the hypoglossal foramina. The paroccipital carries the posterior surface of the skull downward and outward to a much greater degree than in any Dove, but to a less degree than in Dinornis. The Eustachian tubes impress the outer and fore part of the basisphenoid.

The temporal fossæ (Pl. III.), in the present specimens, show the same contraction in proportion to their depth by which the original skull of the Dodo, compared with that of the Dinornis, ‘Proc. Zool. Soc.’ (1848, p. 3), differed from the larger extinct wingless bird. In the approximation of the postorbital process to the mastoid, Didunculus shows a closer resemblance to Didus than does Goura, in which the temporal fossa, besides being narrow, is shallow. The temporal muscle appears to spread its origin above the fossa upon the sides of the cranium, forward half an inch in advance of the postfrontal process, and backward to the outer angle of the superoccipital ridge.

The parietal region is broad, flat, and short, as in Dinornis, not convex as in Doves; it is also impressed at its middle part by a shallow transverse groove, continued outward and forward of less depth and definition, so as to mark off the convex interorbital part of the swollen frontals.

The outer side of the mastoid is convex, smooth, overhanging the tympanic cavity, and sending off a short process, the base of which is defined in one cranium by a transverse ridge in front of the anterior articular cup for the tympanic bone. A similar process is developed in Didunculus, not in Goura, where it is barely indicated.

The presphenoid is compressed, but thickened and rounded below, where the palatines and pterygoids at their junction with each other abut against it: the pterygoid sends off a short process from the middle of its hinder border; but this is not met by a corresponding “pterygoid process” of the basisphenoid as in Didunculus.

The frontals are broad and convex, rising abruptly (as in Didunculus) above the coalesced cranial ends of the nasals and premaxillary (Pl. III.); in Didus the breadth greatly exceeds the length of the interorbital frontal convexity, as compared with Didunculus, and the convexity reigns in the transverse as well as the antero-posterior direction; in Didunculus, however, it is less concave transversely than in Goura. In the breadth or thickness of the interorbital septum Didus resembles Apteryx and Palapteryx and shows the same pneumatic cancellous structure. The posterior olfactory chambers are partially divided, as in Dinornis, by an upper median septum; each compartment, which is 7 lines across and an inch in length, is perforated posteriorly by an olfactory foramen more than a line in diameter, from which grooved impressions of ramifications of the nerve diverge upon the hind and upper wall of the chamber: external to the olfactory foramen is a longer one for the passage of a vein into the fore and inner part of the orbit.

The cranial ends of the nasals and nasal process of the premaxillary (Pl. XI. fig. 1, 22) are flat, depressed, thin plates; the latter at its junction with the frontal is 6 lines broad, partially divided by a median groove above and a ridge below, and by short linear fissures from the nasals: the forward extension of these bones is feebly indicated by linear grooves terminating at the outer margins of the nasal branch of the premaxillary, about 4 inches from its vertical end. The proportion of the base of the upper mandible attached to the frontal contributed by the nasals is the same as that indicated in the ‘Proc. Zool. Soc.’ l. c. The nasal branch of the premaxillary presents a full elliptical transverse section where it quits the maxillary processes, losing both depth and breadth as it recedes to join the nasals; here it retains its breadth, viz. 6 lines, but continues to be thinned off vertically to the plate above named joining the frontal. The under surface of the narrower part of the stem is angular, the upper one being gently convex.

“Where the nasal and maxillary processes diverge, there is a deep groove externally, terminating in a canal directed forwards into the rostral part or body of the premaxillary”30. This part is subdecurved, pointed, roughened by irregular vascular perforations and grooves, with a sharp alveolar border, which describes a sigmoid curve lengthwise, and with a deeper concavity of the palatal surface than in Dinornis or Didunculus. Moreover the concavity is partially divided lengthwise by a median ridge. The palatal surfaces of the maxillary processes and maxillaries are narrow and very convex transversely, intercepting a long narrow palato-nasal fissure. The outer side of the maxillary process is deep vertically and slightly concave lengthwise—a structure not known in Didunculus or any Dove, and related, like most other deviations from the Columbine cranial characteristics, to the provision of unwonted strength of beak in the Dodo. The maxillary branches of the premaxillary have completely coalesced with the maxillaries, as these have with the palatines; and the halves of the upper mandible here swell out laterally and more so vertically, the maxillaries rising to combine with the outer divisions of the nasals, and sending back a short process from their lower and lateral part to join the malar. The inner surface of the maxillary process (Pl. XI. fig. 1, 22*) is smooth and slightly convex vertically; both upper and lower borders are obtuse and thick.

The palatines arch outward from their posterior attachments, are broad and smooth mesially; the margin here is angular, with a slightly produced obtuse apex, divided by a channel on the under surface of the palatine from the outer convex border; the upper and outer ridge extends forward to the maxillary; the inner one subsides before reaching that bone. “The palatines form the posterior boundaries of the naso-palatine aperture, and approximate each other at both ends, but more closely posteriorly, yet here without meeting; whilst in Didunculus they coalesce before receiving the abutment of the pterygoids.

“The tympanic bone is subquadrate, with the four angles produced, and the upper and hinder are bifurcate, forming the double condyle for the mastoid articulation”31. There is a larger pneumatic foramen, communicating with the tympanic cavity, between the articulating cavities for these condyles.

The brain is singularly small in the present species of Didus: and if it be viewed as an index of intelligence of the bird, the latter may well be termed ineptus. The length of the cranial cavity (Pl. XI. fig. 1, v c) is 1 inch 8 lines, its extreme breadth 1 inch 6 lines, its greatest height 1 inch (and this is at the cerebellar fossa). The most remarkable feature in the cranial structure of Didus is the disproportionate size of the brain-case to the important part of the neural axis it contained and protected: some approximation to this condition is made by Dinornis32, the Owls, and a few large Cockatoos, e. g. Microglossum aterrimum; but it is fully paralleled only by the Elephant among air-breathing vertebrates, as may be seen by comparing the section Pl. XI. fig. 1 with the figures of a similar section quoted below33.

Not only was the brain of very small proportional size in the present large extinct bird, but the division of the cranial cavity appropriate to the cerebrum proper is less in proportion to that for the cerebellum and optic lobes, at least in vertical and longitudinal diameters, than in any other known bird.

In the Elephant the thickness of the pneumatic diploë between the fore part of the cerebral cavity and that of the outer cranial wall equals the longitudinal diameter of the cavity containing the cerebral hemispheres: in Didus it exceeds that diameter. The thickness of the pneumatic diploë above the cerebral cavity equals the vertical diameter of that cavity in Didus: the diploë gradually decreases in thickness as it approaches the foramen magnum. The disposition of the osseous lamellæ forming the cells or cavities of the diploë is very different in the Elephant and Dodo: they extend for the most part vertically between the outer and inner tables of the skull in the proboscidian mammal, leaving long and narrow interspaces; in the heavy ground-bird they form a congeries of small subequal and subspherical air-cells, and this structure obtains in the basal and lateral walls as well as in the superior or “roofing” wall of the cranial cavity. The extent of this cancellous structure at the sides of the cranial cavity may be known by the ratio of the breadth of that cavity to the breadth of the cranium, which is 3 inches and 8 lines at the broadest part of the brain, viz. the prosencephalon. It would seem, at first sight, as if the poorly developed brain of the Dodo had needed, on some account, unusual protection; but the true explanation rests on the size, weight, and power of the bill, and the concomitant necessity for adequate extent of attachment of the facial to the cranial part of the skull, and of the muscles from the trunk destined to sustain and wield the long and heavy-beaked head. The cerebrum of the Dodo does not greatly, and by no means proportionally, exceed the size of that part of the brain in the Crown-pigeons (Goura). If the great Ground-dove of the Mauritius gradually gained bulk in the long course of successive generations in that uninhabited thickly-wooded island, and, exempt from the attacks of any enemy, with food enough scattered over the ground, ceased to exert the wings to raise the heavy trunk, then, on Lamarck’s principle, the disused members would atrophy, while the hind limbs, through the increased exercise by habitual motion on land, with increasing weight to support, would hypertrophy.

In the long course of generations subject to this slow rate of change, there would be nothing in the contemporaneous condition of the Mauritian fauna to alarm or in any way to put the Dodo to its wits; being, like other Pigeons, monogamous, the excitement, even, of a seasonal or prenuptial combat, might, as in them, be wanting: we may well suppose the bird to go on feeding and breeding in a lazy, stupid fashion, without call or stimulus to any growth of cerebrum proportionate to the gradually accruing increment of the bulk of the body. Whatever part of the brain was concerned in regulating or controlling muscular actions, might, indeed, be expected to show some concurrent rate of increase with the growing mass of the voluntary contractile fibres; and the size of the cerebellar division (Pl. XI. fig. 1, n o) of the cranial cavity accords with the generally accepted physiology of the superincumbent mass of the epencephalon. The lateral depression at the fore and under part of the side of the postcerebral division of the cranial cavity indicates that the optic lobes, like the eyes, remained almost stationary during the progressive acquisition of the bulk that distinguishes the Dodo from the largest existing Doves.

The proportions of Didus, Pezophaps, Casuarius, Rhea, Dromaius, Struthio, Aptornis, Cnemiornis, Palapteryx, Æpyornis, Dinornis, &c. among terrestrial birds, of Notornis among the lake-haunting Coots, and of Aptenodytes and Alca impennis among seabirds, point to the disuse of wings in flight as the main condition of increase of size in species of birds—the next condition being absence of lethal enemies during the years requisite for such course and rate of growth.

Let foes arise from whom a power of flight is the main condition of escape, and the wingless giants of the feathered class soon succumb. Among the genera above-cited, Aptornis, Cnemiornis, Æpyornis, Palapteryx, Dinornis, Didus, and Pezophaps, with the largest of the Auks, have thus passed away, while Notornis and Apteryx are on the verge of extinction through the rapid increase of population in the small island to which they are restricted. In sparsely peopled continents, such as Africa, South America, and Australia, brevipennate giants may still range the deserts, pampas, and unfrequented wilds. The ascertained recent advent of Man in New Zealand, New Britain, Ceram, Banda, Salwattie, Mauritius, Rodriguez, significantly points to the conditions under which have come to pass, in lapse of time, so strange an anomaly as a bird with the specially modified instruments of flight reduced below the power of exerting that mode of locomotion, yet, as a bird, retaining the conditions of the respiratory and tegumentary systems of the volant class, of which it has become a degenerate member. With the cessation of the chief of those conditions, viz. the absence of enemies, such birds necessarily perish.

Refraining, however, from further indulgence in an easy and seductive vein of speculation, I would recall attention to the notable protuberance in the cranial cavity of the Dodo (Pl. XI. fig. 1, o) developed towards the upper part of the vertical tentorium, contracting at its lower part into the ridge dividing the prosencephalic from the mesencephalic chamber. In the latter are the orifices for the issue of the trigeminal nerve, the larger and posterior (ib. tr) giving passage to the third and second divisions, and answering to the combined foramen ovale and rotundum of mammals, and the smaller and anterior foramen dismissing the first or orbital division of the fifth nerve. At the upper part of the mesencephalic fossa the narrow groove for the lateral venous sinus impresses and defines the back part of the tentorial protuberance, above which it bifurcates, the lower branch bounding or defining the wall of the superior semicircular canal and the upper part of the primitive acoustic capsule. Below this arch is an oblong cerebellar fossa (ib. n) which appears to have received veins from the cranial diploë. Beneath this fossa, and just behind the mesencephalic chamber, is the multiperforate internal auditory depression. Next behind this is the outlet for the vagal nerve and entojugular vein. Below this are the small precondyloid foramina. There is a falcial ridge, low and thick, indicating the division of the prosencephalic chamber into lateral compartments for hemispheres; and this ridge shows a narrow groove as for a small longitudinal sinus. A transverse linear groove abruptly defines the fore part of the ridge.

The vertically expanded anterior part of the premaxillary (ib. fig. 1, 22) has a large pneumatic cavity communicating by a reticulate wall with the cells of a cancellous structure, larger than those of the cranial diploë. The maxillary branch of the premaxillary (ib. 22*) consists of a light open-work air-diploë, with a very thin outer case of bone. The short symphysis mandibulæ shows a small cavity, surrounded by more minutely cancellous structure and thicker compact walls, especially at the upper and hinder parts.

Although some characters have been too much insisted on (e. g. the “superoccipital foramen”) as exemplifying the affinity of the Dodo, the more essential characters of the skull relate to its true Columbine character, while the deviations from that part of the skeleton of volant Doves are explicable in the adaptive developments needed for the wielding of long, powerful, massive mandibles, serving most probably to enable the bird to subsist on some proportion of animal diet, in addition to such vegetable food as it might gain from the ground. Such indiscriminate feeding doubtless rendered its flesh less palatable than that of the winged Pigeons of the Mauritius to the Dutch navigators of the sixteenth and seventeenth centuries.

But the affinities of Didus will be more fully and decisively brought out in the comparison of the, in this respect, more instructive and light-giving parts of the skeleton.

§ 3. Comparison of the Skeleton

The dorsal region of the vertebral column shows, in some birds, a confluence of certain vertebræ: I have observed four to be so welded together by both centrums and neural spines in Phœnicopterus, viz. the second to the fifth dorsal inclusive, leaving the sixth free, which articulates with the first costigerous sacral vertebra. In Platalea three dorsals coalesce in advance of the antepenultimate free vertebra. In the smaller diurnal birds of prey five dorsal vertebræ are usually confluent, leaving one free vertebra for the lateral movements of the trunk between such dorsal “sacrum” and the pelvic one. In Vultures, Plovers, Bustards, Cranes, Psophia, Cariama, Palamedea, Auks, Penguins, and in all flightless land-birds save the Dodo, no such anchylosis takes place. The Columbidæ are the species in which the dorsal vertebræ, homologous and the same in number with those of Didus, undergo the process of confluence into one mass of bone: they are the three which immediately precede the last (moveable) dorsal vertebra; and of these the two anterior develope, in Goura and Didunculus, hypapophyses closely corresponding in shape and proportion with those in the Dodo.

The chief difference which Didus offers in the present region of the vertebral column from that of Columbidæ is in the greater number of the vertebræ or segments which are typically completed by bony hæmapophyses articulating with pleurapophyses and directly with their mass of coalesced and expanded hæmal spines constituting the sternum. Of these typical thoracic segments there were five in Didus (Pl. III.); Didunculus (ib.) shows four; Goura three. In both existing genera these segments are succeeded by a single one, anchylosed to the fore part of the sacrum, but with the pleurapophysis long and moveable, with its hæmapophysis terminating in a point before reaching the sternum, and extensively connected with the antecedent hæmapophysis or sternal rib: in both genera two dorsal vertebræ in advance of the typically complete one have moveable pleurapophyses terminating freely in a point, with no hæmapophyses other than the costal processes of the sternum may represent. In Goura, which has six pairs of moveable or thoracic ribs, the second pair belong to the first of the three anchylosed dorsal vertebræ: in Didunculus, which has seven pairs of thoracic ribs, the second pair belongs to the free dorsal immediately in advance of the anchylosed mass. Supposing Didus to have had one pair of ribs behind, and two pairs in front of those that directly articulate with the sternum, as the vertebra Pl. V. fig. 7 indicates, it would have had eight pairs of thoracic ribs; and I think this excess of one pair beyond the formula in Didunculus to be very probable in the large-bodied, small-winged, extinct Ground-dove.

As far as the series of Dodo’s neck-vertebræ under my observation exhibit such characters, the proportion of those with neural spines, or with hypapophyses, or both, is the same as in the Columbidæ. In this family, as in most birds, the greater part of the series want both processes. The cervical parapophyses, descending to form the sides of the carotid canal, do not meet, coalesce, and circumscribe it in any cervical vertebra of Goura or Didunculus; and not any of the vertebræ of Didus, which I have yet received, shows such circumscription of the hæmal canal. The majority of the cervicals in Didus (those, viz., that lack both neural spines and hypapophyses) are broader and more massive in proportion to their length than in the winged Doves. The third cervical in Didus has both the above processes, as in Columbidæ: the characters of the axis vertebra in the same family are closely repeated in that of the Dodo. In the Raptores the axis vertebra is shorter in proportion to its length, and a greater proportion of the cervical vertebræ at both ends of the series have both neural spines and hypapophyses.

The ribs of the Dodo are as broad, in proportion to their length, as in Doves, but are relatively longer in proportion to the dorsal region, encompassing a more capacious thoracic-abdominal cavity. The ribs of the Vulture are more expanded than in Didus, especially where they afford the extensive attachment to the epipleurals. But I shall not dwell further on the comparative characters of this part of the skeleton, as more decisive ones of the affinity of Didus are afforded by other parts.

In comparing the sternum of the Dodo with that of Doves of flight, the first well-marked difference is in the adaptive development of the keel in the last (Pl. III. fig. 2, Didunculus), and in the provision for the concomitantly broader coracoids, the grooves for which meet and run into each other across the fore part of the bone in existing Columbidæ (Pl. XII. fig. 2, b); consequently the inner or upper wall of the confluent grooves forms a median prominence (ib. e) at the front margin of the sternum, contrasting with the wide notch at that part of the bone in the Dodo (Pl. IV. fig. 4). The next difference, as compared with Goura and most Pigeons, is the absence of the entolateral processes (Pl. XII. fig. 3, i) in the Dodo’s sternum: but Didunculus singularly exemplifies its nearer affinity to Didus by a like absence of those processes; only the sternal margins behind the ectolateral processes (ib. fig. 1, h), instead of converging with a slight convexity to an obtuse apex, as in Pl. VI., describe a concavity, through an expansion of the posterior truncate end of the breast-bone. The sternum of Didunculus may be said to show one pair of posterior notches (Pl. XII. fig. 1, f), that of other Pigeons two pairs (ib. fig. 3, f f′); but the sternum of Didus, which is relatively broader, shows no other trace of the anterior notch (Pl. VI. f) than is afforded by the rounded angle at which the ectolateral process (h) rises from the bone. Although the costal margin is relatively shorter in Doves of flight than in the Dodo, again an intermediate condition is manifested by Didunculus as compared with Goura, in which latter Dove there are articular surfaces for three sternal ribs (Pl. XII. fig. 3, o 1, 2, 3), whilst in Didunculus there are four (ib. fig. 1, c). Didunculus also exhibits, more strongly than Goura, the obtuse ridges (ib. fig. 2, r) converging like buttresses from the outer wall of the coracoid groove to the fore part of the keel, where they subside. In Didunculus there is a pneumatic foramen exterior to the coracoid groove, corresponding with p, fig. 4, Pl. IV., which I do not find in the sternum of Goura; but in the Crown-pigeons the pneumatic foramina along the middle line of the upper surface of the sternum are conspicuous; they are confined to the fore part of that surface in Didunculus (Pl. XII. fig. 1).

In the direction of the ectolateral processes Goura (ib. fig. 3, h) is intermediate between Didunculus and Didus. The pectoral ridge on the outer surface of the sternum, continued backward from the outer end of the coracoid groove, is adaptively better marked in Pigeons of flight than in the Dodo; and the pair of ridges are more nearly parallel in their backward course, not so convergent as in Didus. In Goura the subcostal ridge is better marked than in Didunculus. In no Dove of flight is the body of the sternum so broad and hollow as in Didus (Pl. XI. fig. 4); in this respect the Vulture more nearly resembles the Dodo, as it does also in the more convex anterior contour of the keel: but the vulturine sternum does not lose breadth as it extends backward; it is a square-shaped shield in birds of prey, shorter in proportion to its breadth, with a greater extent of costal process and margin, and with the ectolateral processes, when they exist, extending backward as far as the hinder border of the bone. In the thorough quest of resemblances to the Dodo’s sternum which I have made through the class of Birds, I came upon an unexpected superficial likeness to it in the sternum of a Night-jar (Podargus humeralis). The ectolateral processes (Pl. XII. fig. 4, h) rise behind the moderately extended costal borders, c; and beyond them the body of the sternum converges to an obtuse end, with a contour similar to that in Didus. Moreover the coracoid grooves are divided from each other by a free concave border, less deep and extensive, indeed, than in Didus, but as free from any trace of episternal projection. The ectolateral processes, however, are extended backward to beyond the sternal body; and this part usually shows a pair of small entolateral notches, f′, of which one was present on one side in the specimen figured.

Through the reduction of the coracoids in all flightless birds, there is an interval between their sternal articulations: this is long and concave in the Dodo, but is longest and most deeply concave in Apteryx; it is long but almost straight in Rhea; in Casuarius and Dromaius it is narrow but deeply notched; in Struthio it developes a short episternal process. In no Grallatorial sternum with both ecto- and ento-lateral processes (as e.⁠g. Otis, Œdicnemus, Charadrius) do the former project, as in Didus and the Rasores, immediately behind the costal margin, but they are continued, parallel with the keel, from the outer and posterior angle of the sternum, distant from the costal margin. In old Plovers the entolateral process joins the contiguous angle of the sternal body, and converts the inner notch into a foramen.

In the breast-bone of the Dodo we plainly discern the Columbine modification of the Gallinaceous type, simplified in the minor development of those parts relating adaptively to the power of flight, and expanded and excavated for the support of the larger gizzard with its heavier grindstones34.

In comparing the pelvis of Didunculus and Goura (Pl. XII. fig. 5) with that of Didus (Pl. VII. fig. 1), the correspondences are:⁠—in the general shape, proportions and disposition of the ilia; in the articulation therewith of the last pair of moveable ribs, and of the short straight confluent pleurapophyses of the three succeeding sacral vertebræ; then follow, as in Didus, three vertebræ without pleurapophyses, these reappearing in the next two with their extremities converging to abut against a prominence of the inner surface of the ilium in the same relative position. The difference here is in the two equal and more slender rib-buttresses, in place of the single stronger one, which is the more common structure in Didus; but in Goura I have noted an instance in which it agreed with the Didunculus on the left side, and with Didus on the right, in the last-specified character. In the Crown-pigeons, also, there is an indication of the transverse ridge marking off the under part of the centrum of the first sacral from the rest, and those that follow are less expanded than in the Dodlets; moreover in Didunculus they show a median canal instead of a ridge, while the ridge is feebly indicated here and there and there is no canal in Goura. In neither Didunculus nor Goura do the sacral centrums behind the last rib-abutments diminish in breadth so suddenly as in Didus: in both the winged Pigeons the hinder part of the pelvic cavity is relatively deeper and narrower than in Didus; in both, also, the upper and anterior concave tracks of the ilia are deeper; and in Didunculus the mesial borders do not attain the neural crest, but leave a pair of open longitudinal canals at that part of the pelvis; in Goura those margins reach the neural crest, but do not overtop it at any part. In Goura the acetabula are more in advance of a median position than in Didunculus, Columba magnifica, or Didus. Although the ischiadic foramina are completed by terminal confluence of the ilium and ischium in Dromaius and Casuarius, yet the length of those foramina (which are unclosed) in Struthio and Apteryx, concomitant with the greater relative length of the pelvis, shows the difference of Didus from the cursorial Brevipennates in this part of the skeleton. The ischia of the winged Pigeons resemble those of the Dodo; but the inner longitudinal ridge is more strongly marked in Didunculus: in the Goura it is less developed than in Didus; the bone is longer also in proportion to its breadth, and the ischiadic foramen is longer and narrower: the proportions of that in Didunculus are more like those in Didus. In Didunculus the pubis coalesces with the ischium behind the small obturator foramen, but leaves a second or posterior elongate ischio-pubic vacuity. The greatest amount of resemblances with the pelvis of the Dodo is found in that of different members of the Dove-tribe.

In comparing the pelvis of the Dodo with that of the Vulture (Pl. XII. fig. 6), we find in the latter that the first two confluent sacral vertebræ; supporting moveable ribs are succeeded by several with short abutting ribs, the extent of this part of the sacrum being nearly one-half of the whole, instead of one-fourth as in Didus and the Doves. The reappearance of rib-abutments after four ribless sacrals is in the posterior third of the sacrum, and they are continued to the end of that bone from the last four vertebræ of the series, constituting a very marked difference, both as to number and the character of the vertebræ in the sacral part of the pelvis.

With regard to the iliac bones, the anterior concave track occupies two-thirds of the extent of the bone in Vultur, not one-half as in Didus and most Doves; the breadth of the posterior parts of the ilia with the intervening sacrum in the Vulture is relatively less than in the winged Doves, and differs in a greater degree from that characteristic part in the sacrum of Didus. In Ciconia the antacetabular part of the pelvis is relatively longer, and the iliac bones are more expanded anteriorly. In Platalea the proportions are more nearly those in Didus. In Otis the ilia touch the fore part of the sacro-spinal ridge, but leave both posterior and anterior apertures of the ilio-neural canals widely open. In Œdicnemus and Charadrius they are grooves, the ilia not reaching the sacral spines. The external concavity of the ilium is longer, narrower, and deeper, in most waders, than in Didus. In Eudyptes and Aptenodytes the ilia are more expanded anteriorly, but the whole pelvis is narrower and longer than in Didus. The Gar-fowl (Alca impennis)35, Uria, Podiceps, and Colymbus, all show still longer and narrower proportions of the pelvis.

In the Doves of flight the proportions and relative position of the three compartments of the cranial cavity differ from those in the Dodo. Both the pros- and mes-encephalic ones are proportionally larger than the epencephalic; and the mesencephalic compartment lies more directly below the prosencephalic one. A very thin stratum of finely cellular diploë divides the two tables of the skull along the medial line of the upper surface: it is thicker between the orbits. The falcial ridge at the inner surface of the prosencephalic roof resembles that in Didus. The tentorial ridge bifurcates halfway down, the front portion dividing, almost horizontally, the pros- from the mesencephalic compartment, the hinder and more obtuse ridge dividing, almost vertically, the mes- from the epencephalic compartment. The angle of bifurcation is slightly produced and obtuse, but represents very feebly the tentorial tuberosity (Pl. XI. fig. 1, o) in the Dodo: from it, in Goura, is continued backward the arch of bone formed by the superior semicircular canal, above which is the groove for the venous sinus, as in Didus. The internal auditory fossa is less deep than in Didus: above it is a similarly vertically oblong cerebellar pit. The nerve-foramina correspond with those in Didus: the entocarotid canal opens into a rather deeper sella in Columba palumbus.

On comparing the cranial cavity, as exposed by a vertical longitudinal section in the Dodo (Pl. XI. fig. 1), with that of a Dinornis similarly exposed36, the first difference is the smaller proportional depth of the diploë in the larger wingless bird, which is not greater over the prosencephalic than over the epencephalic compartment; next may be noticed the larger relative size of the former compartment, indicating the larger cerebrum of the Dinornis, then the absence of the tentorial tuberosity, the sharper and more produced superior part of the tentorial ridge arching transversely between the cerebrum and cerebellum, the smaller internal auditory fossa, and the deeper sella: the mesencephalic compartment, or cavity for the optic lobe, is less in proportion to the prosencephalic compartment than in Didus; it holds, however, a similar relative position: finally, the cerebellar pit, above the internal auditory fossa, is wanting in the Dinornis.

The Dodo agrees with the Doves in possessing a slender furculum, forming an acute angle: it resembles Columba galeata, more especially, in the halves of that bone being united by ligament below, and forming separate styles or “clavicles.”

The humerus of the Goura closely repeats most of the characters described in that of the Dodo; but its length is proportionally greater, being 3 inches 9 lines, nearly equal to that of the sternum or pelvis, whereas the humerus of the Dodo is little more than half the length of either sternum or pelvis. The processes for the attachment of the muscles are, nevertheless, fully as strongly developed in Didus (Pl. VIII. figs. 12 & 14) as in the volant Doves (Pl. XII. figs. 8 & 9, Goura); that, indeed, which is a ridge (r) on the back part of the shaft in Didus, is a mere rough surface in Goura, and does not show in Didunculus. The pneumatic fossa, which varies in depth in the two humeri of the Dodo, is in both relatively larger and shallower than in Goura. The pectoral process is thinner, but relatively rather more produced, in Didunculus. The humerus in Œdicnemus, Otis, and Charadrius has a more longitudinally extended, thinner, and more produced pectoral ridge than in Didus and the Columbidæ; there is a more marked ectocondyloid tuberosity, which in Charadrius becomes a pointed process.

There is nothing to be gained by giving the details of the more striking differences which the humerus presents in Penguins, Auks, and birds of prey, as compared with that bone in the Dodo; but a few words may be recorded of the comparison of the humerus of the Dodo with that of the flightless bird of New Zealand so nearly approaching to it in size, which bird is described in the 5th volume of the ‘Transactions’ of the Society under the name of Cnemiornis (p. 395, pl. 66. figs. 7–10). In that extinct species, although the humerus is 5½ inches in length, the parts indicative of the forces by which it was worked are comparatively feebly developed. The ulnar tuberosity is narrower, thicker, more obtuse, and its base has neither the upper nor lower excavation; it rises above the articular head, which is less prominent and narrower than in Didus; the pectoral ridge is shorter and situated lower down upon the shaft, not on the same level with the radial tuberosity as it is in Didus; the distal articulation is of the same size as in Didus, but neither the radial nor the ulnar convexity is so prominent or well-defined.

The ulna of the Dodo is shorter absolutely, and much more so proportionally, than in the Goura and most other volant Doves. In these it exceeds the humerus by about one-fourth its own length; in Didunculus (Pl. III.) it is a little longer than the humerus; in the Dodo (ib.) it is shorter than the humerus. The length of the ulna in Goura coronata is 4 inches 6 lines; it is more bent than in the Dodo; the quill-tubercles, seven or eight in number, are more prominent; nevertheless the rough depression for the insertion of the chief flexor is less deep and less defined. The plumed winglet of the Dodo would seem, therefore, to have been frequently and forcibly moved.

In comparing the femur of the Dodo with that of the largest Dove, the bone appears gigantic. The length of the femur in Goura coronata (Pl. XII. fig. 11) is but 3 inches 3 lines, and it is more slender in proportion to its length than in the Dodo; it, however, repeats the few characteristics, if they may be so termed, of the Dodo’s femur. It has the pneumatic foramen in the same position, perhaps proportionally larger; it has the same large oblong surface for the ligament at the head of the bone; the great trochanter has the same form and disposition, but is not quite so much produced anteriorly; there is a slight depression instead of a ridge for the trochanter minor; the fore part of the inner condyle is relatively thicker and less produced. The femur in Otis and Œdicnemus has a thicker and shorter trochanter major, & more narrow and shallow rotular channel; it is shorter in comparison with the tibia, and more especially with the metatarsus, than in Didus and the Doves.

The femur of Aptornis otidiformis37 is of the same size as that of the Dodo; but it has no pneumatic foramen, the head is more hemispheroid and inclined forward, the ligamentous pit is deeper and more circular, the supracervical articular surface is not defined from that of the head, there is a wider and deeper depression at the fore part of the proximal end of the femur, and a more prominent tuberosity on the back part; the ridge continued from the back part of the shaft to that of the inner condyle is more produced and sharper in Aptornis, the fore part of the same condyle is less produced.

The femur in Cnemiornis38 and Dinornis39 is much thicker, in proportion to its length, than in either Aptornis or Didus. In Pezophaps the great trochanterian ridge rises higher above the neck, and the shaft has a more uniform thickness, with the inner contour less concave, than in Didus.

The characters which have been noted at the proximal and distal ends of the tibia of Didus are repeated in those of the tibia of the Goura. The difference in size is more marked than in the femur; the length of the tibia of Goura coronata is 4 inches 7 lines, and its shaft is more slender, in proportion to its length (Pl. XII. fig. 13), than in Didus (Pl. X.). The tendency to a trihedral form of the shaft is less marked in Goura; the anterior prominences of the distal condyles are thicker in proportion to the intervening fossa.

In the Vulture the fibular ridge is more parallel with the long axis of the shaft than in Didus; the tendinal canal is less cylindrical, has an oblique course from the middle of the anterior surface towards the inner condyle; the fore parts of both distal condyles are less produced and less convex; the distal end is narrower from before backwards in proportion to its breadth; both extremities of the bone are less expanded in proportion to the shaft than in the Dodo.

In the great Plover (Œdicnemus crepitans) the tibia, as in other Grallæ, is longer in proportion to its thickness than in Didus; the epicnemial process rises higher above and projects further in front of the condylar surfaces before it divides into the pro- and ectocnemial plates; and these are relatively more produced. The fibular ridge is shorter in proportion to the length of the tibia, is more prominent, and more parallel with the axis of the shaft. The distal condyles project further backward than in Didus. The tibia in Charadrius, Otis, Tantalus, Grus, Ciconia, Mycteria, Porphyrio, opposes similar or equivalent differences to those in Œdicnemus, against the affinity of Didus to any of those Grallæ.

In the comparison of the tibia of this extinct flightless bird with that of the Cnemiornis, the wonderful development of the plates and processes at the proximal end of the bones in the New Zealand bird is strikingly manifested. In Cnemiornis the fibular ridge runs in a line with the shaft, and does not incline from above obliquely forward as in Didus and the Doves; the ridge on the outer side of the distal fourth of the bone is stronger and sharper in Cnemiornis; the tendinal canal is transversely elliptical, medial in position, with a slight inward inclination; the intercondyloid fossa is much wider in Cnemiornis. The differences, indeed, in all the characters of the tibia, as compared with Didus, in the Vultures, Plovers, Penguins, and terrestrial flightless birds tend to render more instructive and convincing the resemblances which Pigeons present in the same characters to the extinct Mauritian bird.

← Back to Memoir on the Dodo