discs,' the intercentra' or 'basiventral' elements. These are, as has been shown,' originally the portions of each vertebra with which the ribs articulate, from which they are outgrowths. But as the ribs come to articulate with the centra these structures degenerate. In the development of Apteryx T. J. PARKER found a postoccipital and a postatlantal intercentrum, and two in the caudal region, which ossify so as to retain their independence in the adult

skeleton.

Intercentra in the caudal region of the bird's vertebral column are by no means so rare as might be inferred from

In

some published statements upon the matter. They are especially conspicuous among the Limicolæ and the nearly allied auks, and in most water birds. Numenius femoralis, for example, there are three small osseous nodules lying between caudal vertebræ 1-5. Behind these are a series of hypapophyses, which are a continuation of the same series, but much more pronounced and ankylosed to the vertebræ.

They exist also in the duck tribe. In Biziura lobata there are three distinct intercentra in the form of largish nodules. I have found intercentra also in Palamedeæ, Tubinares, Steganopodes, Colymbi, Herodiones, Opisthocomus.

These free intercentra are rare among the Pico-passeres, but in a few of them are present. Thus in Toccus there is a distinct intercentrum lying between the last free caudal

vertebræ.

The hawk tribe have not these bonelets as distinct structures.

In the cuckoos, parrots, Ralli, Otides, Columbæ, and the

This matter of the composition of the vertebra has been recently gone into by GADOW (on the 'Evolution of the Vertebral Column of Amphibia and Amniota,' Phil. Trans. 1896, p. 1), who quotes previous literature.

tinamous I have not seen in the adult skeleton any free intercentra; nor in the Grues, excepting Chunga.

Further details on this matter will be found in the paper cited below.1

Though free intercentra are by no means universal among recent birds, hypophyses of the last caudals are almost so. That these latter are derived from intercentra, and are, therefore, not comparable to the hypapophyses of the cervicals and dorsals, is clear from such cases where the gradual transition between free intercentra and fixed hypapophyses is shown. In reptiles the intercentra are in the tail region constantly in the form of chevron bones, which are V-shaped, articulating with the vertebral column by the free ends of the V. This form of the hypophyses of the caudal region is not so common as a simply bifid condition, but does occasionally occur. I have seen it, for example, in Tubinares, Accipitres, and Cuculi.

The first vertebra of the cervical series is called the atlas (see fig. 66); it is a ring on bone, of which the greater part of the centrum' is formed by the projecting odontoid process (see fig. 67), the rest being formed by a pair of intercentra. In the hornbills the atlas is fused with the following axis vertebra. Generally the atlas has not what the succeeding vertebræ have, a vertebrarterial canal, but this is sometimes present (see under Ribs,' p. 119). The odontoid process sometimes notches the lower part of the atlas, and sometimes perforates it. These two conditions. are illustrated by figs. 66 and 68. It sometimes happens that the neural arch of the atlas is incomplete, e.g. Chunga, Colius, Pandion. As a rule it is perfect.

In the cervical vertebræ the chief facts which appear to be of systematic importance are the relations to each other of the paired processes, to which MIVART has applied the name of catapophyses. These are sometimes inconspicuous processes of the transverse processes on the under side. Very often the last one or two pairs of them closely

BEDDARD, 'Note upon Intercentra,' &c., P. Z. S. 1897, p. 465. 2 In a specimen of Chunga I have found the same fusion.

approach each other in the middle line, as in Psophia. Sometimes a number of these processes unite to form a canal; this occurs in the Steganopodes, and most Herodiones (but not in Scopus umbretta); but the classificatory significance of the fact is marred by the occurrence of a similar canal similarly formed in some Picidæ, and in the case of one vertebra in the parrot, Eclectus polychlorus.

It is sometimes the case that the last of the catapophyses is consolidated into a thick process, which is bifid at the extremity; this process forms a transition to the following hæmapophyses (or hypapophyses). These latter are un

paired median processes, which commence upon the cervical vertebræ, and extend for a variable distance back along the dorsal vertebræ. They are very feeble, and sometimes limited to the cervical region, in the Herodiones. They are most highly developed in Sphenisci, Colymbi, Alcæ, and some Anseres, being in these cases continued to the end of the dorsal series, and even being found upon some of the lumbar vertebræ. In many cases these processes are flattened out at the free end like an inverted T, or are trifid at the same place. This is seen to be due to the gradual shifting in position of a posterior set of catapophyses, which at first are at the sides of, and far from, the hæmapophyses, but get

nearer and nearer, until at length they mount upon the hamapophysis itself and pass to its very end. Details of the formation of the vertebræ will be found in the systematic part of this book.'

Ribs.2 The ribs of birds vary greatly in number. There are as a rule three series of ribs to be distinguished. The last cervical vertebræ, more or fewer of them, are furnished with short ribs which do not reach the sternum. Behind these are, again, a variable number of true ribs, which do reach and articulate with the sternum. These true ribs consist of the vertebral portion, which articulates with the vertebra, and of a sternal portion, which is articulated with the vertebral half of the rib above and with the sternum below; it is bent at an angle with the vertebral portion. Attached to, originally separate from, and sometimes permanently separate from, the vertebral half of the rib is the uncinate process, of which there are a variable number. These processes are absent in Archæopteryx and in the Palamedeæ only. Behind the true ribs, which articulate with the sternum, are a variable number, in all degrees of

The relationship of the so-called catapophyses to the unpaired hæmapophyses varies, and suggests—what has been advanced on other grounds-an occasional excalation of vertebræ. Without wishing to commit myself to a belief in the actual dropping out of a vertebra from the middle of the series, I may mention some of the facts which may be regarded as pointing in this direction. In the grebe Echmophorus the catapophyses form on certain vertebræ a complete ventral canal for the carotids. The summit of the arch thus formed gradually acquires a median dorsal process. This increases, and the catapophyses finally end in the obliteration of the canal which they sur round, and a solid arch is formed; the hypapophysis of the succeeding vertebra is single and no longer retains traces of its evolution from a ring of bone surmounted by a process. In other cases the catapophyses suddenly end and the hypapophyses begin without such intermediate stages. An intermediate stage is seen in certain types where the catapophyses end suddenly, but the first hypapophysis is double, either formed of two clearly fused pieces or with merely a bifid spine. These latter cases suggest the dropping out of one or more vertebræ, effecting the transition between the paired catapophyses and the unpaired hypapophyses.

* In all birds except Archeopteryx the ribs are two-headed with a capitulum and tuberculum.

They have been often said to be absent in Dinornis, but they are not. W. BEHRENS, Untersuchungen über den Processus uncinatus der Vögel und Crocodile, Inaug. Diss., Göttingen, 1880.

degeneration, of floating lumbar ribs. Morphologically equivalent with ribs are the processes firmly ankylosed to the cervical vertebræ, which form a canal for the vertebral artery; these are, as a rule, absent from the atlas, but are present on that vertebra in the Anseres, Opisthocomus,' Triponax Feddeni, Dromaus2 (fig. 66). Archæopteryx is alone in possessing the abdominal ribs of the crocodiles and other reptiles.

The Shoulder Girdle.3-The shoulder girdle of birds consists of at any rate three separate elements-the scapula above; the coracoid, articulating with the sternum; and the clavicles, generally united into a U-shaped piece. Of these the first two are preformed in cartilage, the last in membrane.

The scapula is a thinnish sword-shaped bone which is attached by muscles to the ribs and to the vertebræ, and lies in a direction, as a rule, nearly parallel to the long axis of the body. The scapula does not show great variability of form among birds; the most considerable variation is to be seen in the penguins, where the bone is, comparatively speaking, of enormous width. A free suprascapula has been noted by PARKER in Opisthocomus. The coracoids articulate on the one hand with the scapula, and on the other with the sternum, where they are received into grooves on its anterior margin. There are some variations in the way in which these grooves are arranged in some birds the two coracoids at their insertion are not in contact at all; in others they are in contact; and finally they may overlap, as in reptiles.

The coracoid has in many birds a procoracoid process, which is believed to be the equivalent of the procoracoid of reptiles. This is especially prominent in the ostrich, but is present in a large number of other birds, though more reduced in extent. But its large or small size is so capricious

Stated by PARKER to be absent.

* MIVART figures a canal on one side of the atlas of the ostrich. A. SABATIER, Comparaison des Ceintures et des Membres Antérieurs et Postérieurs dans la Série des Vertèbres, Montpellier, 1880; PARKER, quoted below. See also LÜHDER in J. f. O. 1871, p. 321.