The genus Brachiosaurus was recently described by the writer form the humerus, femur, coracoid, and such parts of the sacrum and vertebral centra as could be seen before the specimen had been remove from the matrix. During the past winter the sacral and presacral vertebrae, which had been badly damaged by weathering, have been reconstructed with great care and patience by Messrs. J. B. Abbott and C. T. Kline. As these parts have on after another been worked out, the unusual character of this animal, which was first indicated by the extraordinary proportions of the humerus, has become more and more evident. So different is its structure form theta of the members of the Opisthocoelia that the writer feels justified in placing it in a new family. The Brachiosauridae is, therefore, proposed as a family group, to include this genus together with the smaller and more primitive form recently described by Hatcher under the name Haplocanthosaurus.

The family characters so far observed are: Humerus as long as femur; neural spines of vertebrae simple; dorsal vertebrae more than ten. Other characters equally distinctive will doubtless develop as these animals become better known. The following key will aid in determining members of this group:

Opisthocoelian dinosaurs with fore leg longer than hind; vertebral spines simple throughout; number of dorsal vertebrae more than ten; Family Brachiosauridae.

(a) Size medium, dorsal vertebrae fourteen; centra not elongate; neural arch unusually elevated, diapophyses directed obliquely upward and outward, hyposphene-hypantrum articulation moderately developed: Genus Haplocanthosaurus.

(b) Size large; neural arches not unusually elevated, spines increasing in length form sacrum to mid-dorsal region, hyposphene-hypantrum articulation unusually developed, center of dorsal vertebrae elongate: Genus Brachiosaurus.

The type specimen of Brachiosaurus as now prepared for exhibition, consists of the sacrum, seven presacral and two anterior caudal vertebra, the right humerus, coracoid, ilium, and femur, and a number of ribs more or less complete. It was collected by the Museum paleontological expedition of 1900 from the Grand River valley of western Colorado. (Pate Lxxi.) When found, the vertebral column was lying with spines downward and the vertebra were but little displaced from their normal relations. The ventral surface of the sacrum, the ilium, and the anterior caudal vertebrae were exposed, and had suffered more or less from weathering.

At the seventh presacral vertebra the thin clay stratum in. which the specimen was imbedded "pinched out," and was replaced by a massive layer of sandstone with coarse sand and pebbles at the base. This, together with the uniform displacement of the ribs, humerus, and coracoid to the left and the presence of an isolated ilium of Diplodocus, which had the appearance of having been drifted up against the broken vertebral series, indicated that the anterior portion of the skeleton had been carried away by the invasion of a water-current after the specimen had been partially covered with sediments.

The humerus and coracoid were displaced some ten feet to the left, and when found the distal end of the former was exposed at the surface, broken and displaced. When the fragments had been gathered up and fitted to the portion still in the matrix the bone measured almost seven feet in length. This length so much exceeded that of any humerus previously known, that the writer at the time believed it to be a crushed and distorted femur. Had it not been for the unusual size of the ribs found associated with it, the specimen would have been discarded as an Apatosaur, too poorly preserved to be of value.

The conclusion that the bone in question was a distorted femur was given additional weight by the discovery soon after of a well. preserved femur of almost identical length, associated with the ilium and sacrum. But later, when the two leg bones were removed from the matrix and carefully compared in the laboratory, the identity of the humerus was at once established by the structure of the head as well as by the clearly defined deltoid area from which the deltoid crest had been broken away and lost. From figures 3 and 4, Plate LXXIV, the characteristic structure of the opisthocoelian humerus will at once be recognized.



The seven dorsal vertebrae preserved in the specimen were found in a series, and but little displaced from their relative positions. The center and transverse processes are considerably distorted by the compression to which they have been subjected. These distortions have been corrected in the specimen as far as was practicable but no effort has been made to further correct them in the drawings. (Plate LXXII.) The vertebrae are distinguished by the lightly constructed and elongate centrum with its large lateral cavity, and by the single neural spines, short in the posterior members of the series, but becoming more and more elongate anteriorly. Equally distinctive is the unusual development of the hyposphene-hypantrum articulation. The whole structure suggests lightness and flexibility attained with an evident sacrifice of that strength which is everywhere apparent in the unwieldy Apatosaurus.

The number of vertebrae in the dorsal series cannot, of course, be determined from this specimen. Reasoning from certain similarities between this genus and Haplocanthosaurus, in which the number has been determined as fourteen,* we may expect a more numerous series than characterizes Apatosaurus and Diplodocus. As this number must for the present remain conjectural, however, the vertebra will be referred to as presacral and numbered from the sacrum forward.

The first presacral vertebra may be distinguished, as in all opisthocoelians so far as the writer has observed, by the massive postzygapophyses which overhang the posterior end of the centrum, by the low stout spine and the short centrum+ with small lateral cavities or pleurantra. From this point forward, both the center and spines rapidly elongate, the diapophyses become more and more expanded, but, contrary to all that might be expected the zygapophyses become reduced almost to insignificance.

* Memoirs of the Carnegie Museum, vol.2, No. 1. + In the preliminary description of this genus the writer, estimating from the specimens sill in the matrix, characterized the posterior dorsal center as longer than wide. Closer examination shows that this is not true of the last dorsal.

The centrum in the dorsal vertebrae is opisthocoelous in type, but less pronouncedly so in the anterior members of the series than is common in the mid-dorsal region of most forms. In the first presacral the centrum is similar in length to that of Apatosaurus, but in the preceding vertebrae it rapidly increases in length. The pleurantrum is of moderate size; the anterior end of the centrum is truncate, with a slight convexity above the middle: the posterior end, now badly distorted, was doubtless uniformly concave. The second centrum takes on a more typical opisthocoelous outline. From this point forward the center increase slightly in length, the pleuracoele enlarges, and the general structure displays even greater lightness. A cross-section shows an internal arrangement very similar to that of Apatosaurus (Brontosaurus) as figured by Marsh.

However, the lateral cavities are larger, and the peripheral walls thinner than have been observed by the writer in that genus. The median septum is so fragile as to be lost entirely in some of the vertebrae.

The neural arch is unusually slight in this genus. Its posterior surface is narrow and rounded; anteriorly it is produced into two laterally directed ridges which descend from the base of the prezygapophyses to the anterior rim of the centrum. Inclosed by these ridges and by the buttresses supporting the prezygapophyses is a deep fossa, into the lower margin of which the neural foramen opens.