quently ('Geol. Soc. Quart. Journ.,' 1883, p. 135) a diagram of the shoulder girdle in this genus was given by that author, which represents the scapula and coracoid as meeting each other on the Elasmosaurian plan; but, unlike Elasmosaurians, the scapula are divided from each other on the visceral aspect by a long triangular interclavicle (named episternum) which shows a mesial notch in front. I have not seen this specimen, which is not assigned to any species, locality, or collection. It would appear to show an intermediate condition between Plesiosaurs and Elasmosaurs, but it is impossible for me at present to affirm this. No specimen is known to me which shows that in Pliosaurus the scapula and coracoid completely enclose the coracoid foramina. The evidence is imperfect, but it leads to the conclusion that the shoulder girdle was Plesiosaurian in plau.

ant-mar

FIG. 4.-Interclavicle, Pliosaurus philarchus. ant-mar, anterior border; c, a lateral surface which may have been a clavicular attachment.

Pliosaurus philarchus, on which the genus Peloneustes has been founded (Cat. Foss. Rept. and Amph.,' Part II), in form of the scapula closely resembles Pliosaurian remains in the British Museum. Their approximating margins are convex, and between those margins Mr. Lydekker has inserted the interclavicle (termed omosternum), with Professor A. H. Green, F.R.S., without finding evidence of this entosternal part of the skeleton. What appear to be scapula of Pliosaurus brachydeirus have the inner and outer borders of the bones sub-parallel, with the anterior extremity but slightly widened. Zittel has interchanged the names to Owen's figures of the shoulder girdles of Pliosaurus and Plesiosaurus. I have not seen the originals of those figures.

which is triangular, flat, very thin, and has perfectly straight sides, which, in their hinder approximating two-thirds, are slighly bevelled. There is no evidence given that the bone occupied the position which has been figured, and I see no reason for believing that it was not placed, as in other Sauropterygians, on the visceral surface of the slightly inclined scapula, where there is a doubtful indication of what may be an imperfectly preserved right clavicle. If the straight lateral border of the interclavicle was in contact with the flat visceral surface of the scapula, the bones would be in harmonious relation. The bevelled margin appears to look inward, and is therefore inferred to have given attachment to a lateral ossification which was still more delicately thin. This condition is shown in the following

figure of the bone.

(iii.) A third modification of the Plesiosaurian type may be* indicated by the specimen in the Leeds Collection in the British Museum numbered 36. It is small, and the bones are not sharply ossified and immature, as Mr. Leeds has always believed. But I have not observed any specimen in his collection which would, with certainty, represent its adult state. The bones of the shoulder girdle are thick, and the scapula and coracoid are formed on the Plesiosaurian type, in that the inner border of the scapula gives no evidence of a median precoracoid prolongation backward to meet the coracoid. There is no indication that the coracoids and scapulæ ever met in the median line, even in the supposed adult condition, since there is no anterior median process to the coracoid; but there is a cartilaginous interval between them in front like that attributed to Pliosaurus. The scapula is a stout triradiate bone with a wide external process, and in form it resembles the bones attributed to Pliosaurus. But the cervical vertebræ have no trace of the Pliosaurian modification, and have the aspect of the vertebræ of Plesiosaurus, except that the articulation for the rib is not divided in the cervical region. Some Plesiosaurs from the Lias have shown the closest possible approximation of those surfaces, but the divided condition of the rib facet did not terminate with the Lias species, since some specimens from the Wealden (which are referred to Cimoliosaurus, 'Brit. Mus. Cat. Foss. Rept.,' Part II, p. 227, No. 2,444, No. 26,000) retain the character in a condition similar to that attributed to Thaumatosaurus carinatus (loc. cit., p. 168, fig. 57). It may be that the imperfect ossification causes the facet of bone to appear single in this Oxford Clay fossil, while its cartilaginous terminations during life may have been divided; but so far as the evidence goes it rather suggests a sub-generic modification of the genus Plesiosaurus as indicated by the scapular arch, distinguished by undivided articular heads to the cervical ribs, if the adult preserved * I am not sure that this immature Plesiosaurian type did not, on attaining maturity, become the Elasmosaurian genus Cryptoclidus.

the characters of the young animal. This inference is supported by the evidence of the clavicular arch, and by the large size of the radius and tibia as compared with the small size of the ulna and fibula. These bones are not in natural association, being free from matrix; but I see no reason to doubt that Mr. Leeds has arranged them in positions which are correct. The characters of the skeleton lead to the conclusion that the species is new, and could not become transformed by growth and perfected ossification into any other known species.

The following are measurements which help to define the species:Lower jaw, 9 inches. Vertebral column, as preserved and arranged, 64 inches. Thirty cervical vertebræ, 23 inches; two pectoral vertebræ supporting ribs on the neural arch and centrum, 1 inch. Twentytwo dorsal vertebræ measure 22 inches; three vertebræ in the sacral region, which support ribs, partly on the neural arch and partly on the centrum, 2 inches. Twenty-two caudal vertebræ measure 15 inches, but the extremity of the tail is not preserved. The height of the dorsal vetebræ and neural arch is about 2 inches. The transverse measurement over the transverse processes of the dorsal vertebræ is 44 inches. The longest dorsal ribs measure about 9 inches. The ilium is 4 inches long. The transverse width over the pelvic articulation is 7 inches. The antero-posterior extent of the pelvis is 9 inches. The pubis measures 4 inches from front to back. The ischium is 3 inches in the same measurement toward the median line. The pelvic foramina were separated from each other by cartilage. The femur is 8 inches long and 4 inches wide. The coracoid is 7 inches long by 44 inches wide; the scapula is 4 inches in length and width. These shoulder girdle bones are exceptionally thick. The transverse width over the two clavicles is 7 inches. The clavicles are thin triangular bones, perfectly ossified, with sharp well-defined margins and no signs of immaturity, probably because they are membrane bones. If they met each other in the

FIG. 5.-Clavicles of a young individual Plesiosaurus durobricensis.

median line it can only have been by squamous approximation. Thus arranged they would be inclined to each other. As preserved, each clavicle is about 4 inches wide; and on its inner border measures 2 inches from front to back, and at the external angle the corresponding measurement is of an inch. The anterior border is straight; the inner border is sinuous and unsymmetrical on the opposite sides; the posterior border is 3 inches long and concave, with the concavity broken on the inner third by a sharp prominence which separates a slight inner concavity from the longer external concavity.

The external extremities of the bones are truncated and striated. The only specimen which distantly approximates to this in the large size of the radius as compared with the small ulua is an Elasmosaurian indicated in the Leeds Collection by the number 31. In that also there is no trace of an interclavicle, but the shoulder girdle is not perfectly preserved, and its clavicles are of dissimilar form. If the scapulae in mature individuals of this species united in the median line and extended back to the coracoids, then the fossil would be Elasmosaurian, and possibly a species of Cryptoclidus.

III. THE CLAVICULAR ARCH IN THE ELASMOSAURIDE.

§ 1. The Nature and Limits of the Family.

When the Elasmosauridae was defined in 1874 its clavicular arch was unknown and supposed to be wanting, and the family was based upon the circumstance that the bones named scapula met each other in the median line, and were prolonged backward to unite with the median processes of the coracoids in Elasmosaurus and Colymbosaurus. I owe a knowledge of the clavicular arch in this family to A. N. Leeds, Esq., of Eyebury, who for twenty years has collected the fossil Vertebrata from the Oxford Clay near Peterborough. In this family the cervical vertebræ have the ribs attached by undivided articular heads. The carpal and tarsal bones are polygonal and well ossified. The genera on which the family is based are Elasmosaurus, Colymbosaurus, and Murænosaurus (Geol. Soc. Quart. Journ.,' 1874, p. 436), none of which appear in the 'British Museum Catalogue of Fossil Reptiles,' Part II. The only genera in that enumeration which could be so referred are Polyptychodon and Cimoliosaurus. Excepting Polyptychodon only, all English as well as all American Elasmosaurians have been referred to the latter genus in the Catalogue referred to. Hence, as it will be presently shown that the Elasmosauridae develop remarkable modifications of the clavicular arch, which may be regarded as of generic importance, it is convenient to determine as far as possible the synonymy of the genera comprised in the family.

The genus Cimoliosaurus figured by the late Dr. Leidy in 1865 ('Smithsonian Contributions to Knowledge') rests upon thirteen centrums of vertebrae without arches or processes, noticeable chiefly for their transverse width. Fourteen other vertebræ from the Greensand of New Jersey are described; but there is no evidence of any other part of the skeleton. Leidy expressed doubt whether his genus Discosaurus might not prove to be founded on vertebræ of Cimoliosaurus. This view was adopted by Professor Cope ('Amer. Phil. Soc. Trans.,' vol. 14), but that identification only contributed a knowledge of the carpal and metacarpal bones. Hence the characters by which the is defined in the British Museum Catalogue' genus are not drawn from Leidy's type.

The generic characters of Cimoliosaurus which may be obtained from Leidy's figures are: Articular face of the centrum flat or flattened, short from front to back, transversely extended in the cervical region. The neural arch is small, with compressed lamellar neurapophyses, which appear to be anchylosed to the centrum. The facet for the cervical rib is single, at first compressed from above downward, afterwards becoming ovate; the facets are on short pedicles. The chevron articulations impress both the anterior and posterior margins of the short centrums in the middle of the caudal series. The carpals are transversely oblong. The metacarpals and phalanges are compressed from above downward.

The name Brimosaurus (Leidy, 'Philadelphia Acad. Nat. Sci. Proc.,' 1854, Pl. 2, p. 72), was proposed for Plesiosaurian vertebræ which have the ventral surface flat instead of concave, as in Cimoliosaurus; but, as the genus is not mentioned in that author's 'Cretaceous Reptiles of the United States,' 1865, it may be regarded as probably abandoned and included in Cimoliosaurus.

Dr. Leidy also proposed a genus Oligosimus ('Philad. Acad. Nat. Sci. Proc., 1872, p. 39). It is unfigured and based upon an early caudal vertebra. It has the neural arch anchylosed to the centrum. A groove defines the limit of the articular face of the centrum. The chevron facets only impress the posterior border of the centrum. Its measurements are: length, 1 inch; width, 2-3 inches; depth, 19 inch. These characters seem insufficient at present to distinguish the type as a genus.

Professor E. D. Cope has decribed five other genera which he regards as distinct from Cimoliosaurus; they are named Elasmosaurus, Polycotylus, Orophosaurus, Uronautes, and Piptomerus.

Polycotylus from Cretaceous Limestone, near Fort Wallace, Kansas ('Amer. Phil. Soc. Trans.,' vol. 14, Part 1, p. 35, Pl. 1, 1870), is founded upon dorsal and caudal vertebræ. It is characterised by the very short dorsal vertebral centra, which are deeply biconcave. The tibia is broader than long. The neural arch is anchylosed to the

VOL. LI.

L