The transverse processes in presacrals I to III and in v are broken and lost. In those vertebrae in which they are preserved, these processes have been too much distorted to admit of their direction being determined with certainty. In the posterior members of the series they were evidently much reduced. In iv a single transverse process is preserved which is little more conspicuous than the tubercular facet. In vi and vii these processes rapidly increase in length. They arise from four roots very much as do those of the posterior dorsals in Apatosaurus. They are broad at the base antero-posteriorly, rugose on the anterior margin, and recurved at the distal end.

The capitular facets for the rib attachment are borne high above the centrum throughout the present series. In presacrals I and ii there is little evidence as to their position, but in III a well developed facet is preserved. From this point forward the facets are borne by the lateral surface of the prezygapophyses as well as the anterior margin of the transverse process, and are supported by a buttress descending to the anterior margin of the neural arch. The zygapophyses, together with the hyposphene-hypantrum articulation constitute one of the distinguishing features of the vertebrae of this genus Throughout the dorsal series, so far as known, the zygapophyses are narrow and placed close to the median line. There is no evidence of dichotomy in the anterior dorsal region as observed in Apatosaurus, Camarasaurus, Morosaurus, and Diplodocus The weakness in the zygapophyses is compensated by the strength of the hyposphene-hypantrum articulation. The articular surfaces of the former are directed horizontally. The mesial surface of the prezygapophysis curves rapidly downward and is continuous with the vertical articular surface of the hyposphene. Similarly the articular surface of the postzygapophysis is continuous with that of the hypantrum. The two form a firmly interlocking joint capable of resisting lateral strain. The hypantrum is also slightly expanded inferiorly so as to prevent a downward displacement of the prezygapophyses.

The prezygapophyses are supported by a single pair of buttresses, which also form the anterior margin of the neural arch, and by the anterior plate of the transverse process. The postzygapophyses are supported from above by a pair of stout buttresses which descend from the posterolateral margin of the spine; laterally they are supported by the posterior root of the transverse process and inferiorly by a pair of buttresses arising from the posterior margin of the neural arch and attaching to the anterior surface of the hypantrum. In the sixth and seventh presacrals this support is strengthened by passing to the inferior surface of the hypantrum.

The neural spines in this genus are single and median so far as known, and there is no reason to doubt that they continue so throughout the dorsal series. Unlike those of any other known members of the Opisthocoelia, the spines are short in the first presacrals, and become more and more elongate as far as the middle thoracic region. They are made up of the usual median plate expanded at the crest into a rugose knob and roughened on the anterior and posterior margins for the attachment of interspinous ligaments. They are flanked at the anterior and posterior margins respectively by a pair of greater and lesser lateral plates. The posterior plates arise from the superior root of the prezygapophyses, pass upward along the lateral border of the median plate near the posterior margin, and form a stout lateral support to the crest. The anterior lateral plates arise similarly from the base of the prezygapophyses and strengthen the anterior border of the median plate. In presacral I the spine is low, massive, and strongly reinforced by buttresses arising from the postzygapophyses. In ii it is noticeably higher and less firmly braced. In iv and v the base of the spine becomes broadened anterio-posteriorly by the diverging roots of the lateral plates. A lateral vacuity appears between the posterior and inferior roots of the transverse process and the base of the postero-lateral spinous plate. In vi and vii the spine becomes more elongate and somewhat more slender.



The sacrum of Brachiosaurus is at once distinguished from that. of the other large opisthocoelian dinosaurs by its unusual breadth across the sacral ribs as compared with its length and the height of its component vertebrae. (Plate LXXIII.) The specimen under consideration has been damaged somewhat by compression, but more seriously by 'later exposure to weathering at the surface. As a result, the neural arches of sacrals ii-iv inclusive were too badly damaged to be restored; and with them the centra above the lateral cavities. With these exceptions their structure can be determined with certainty.

The sacrum is made up of five ilium-supporting vertebrae instead of four, as originally described,* all of which are firmly coössified by their centra and the distal ends of their sacral ribs. Sacrals I to IV are coalesced by their zygapophyses and the base of their diapophyses, and sacrals ii and iii by their spines. As in the typical opisthocoelian sacrum** the ribs of sacrals II, Iii, and Iv enter into the composition of the sacricostal yoke and may be regarded as compromising the' primary sacrum. The dorso-sacral is highly specialized, but its support of the acetabulum is a secondary function. The caudo-sacral is functional as an ilium-supporting vertebra, though its caudal affinities are still clearly marked.

*Am. Jour. Sci.. 4th Ser.. vol. z5, p.303. **This publication. Geo. Ser.. vol. ii. No.4. p. 180. The centrum of sacral I is somewhat shorter than that of the last dorsal, and is irregularly convex on the anterior end. The pleurantral foramina are almost closed by the expanded sacral ribs. The centrum of sacral II is similar in length to that of I, but is laterally expanded to meet the unusual development of its sacral rib. There is no evidence of a pleurantrum. The centra of iii and iv are less expanded latterly, and bear traces of the pleurantral foramen. That of v bears no trace of the lateral cavities. Its posterior end may have been convex above and concave below, as is the typical structure, although this feature cannot be determined with certainty.

The sacral rib in 1 is developed into a broad, winglike appendage, whose primary elements can scarcely be traced. It arises from the upper half of the centrum anterior to the much-reduced pleurantrum and from the lateral surface of the neural arch. The capitular portion passes backward, joins the rib of sacral ii, and articulates with the mesial surface of the greater peduncle. The tubercular element is fused with the distal end of the diapophyses so that it is impossible to determine where the one ends and the other begins. It is probable, however, that the tubercular portion of the rib is represented by the rugose end only. It attaches to the crest of the ilium above and in front of the base of the greater peduncle.

The second pair of sacral ribs are by far the strongest of the series. The proximal ends are greatly expanded and attach to the whole lateral surface of centrum II as well as to the posterior half of centrum' i. The shaft is constricted at the middle, and the distal end is expanded to enter into the composition of the sacricostal yoke. The third pair of ribs are much reduced in size. They arise from the anterior half of centrum iii in common with the posterior margin of the second pair. The shaft is slight, but the distal end is expanded to form the middle section of the yoke. The fourth pair arise from the anterior two-thirds of the centrum. From the superior margins of ribs ii, iii, and iv broad plates arise to connect with the corresponding diapophyses. The fifth pair of sacral ribs arise from the mid-lateral surface of the centrum, and passing diagonally forward, unite with the distal ends of the fourth pair in supporting the lesser peduncle of the ilium. These bear traces of the primary caudal structure although they are modified as ilium-supporting elements.

The diapophyses, like the sacral ribs, are remarkable for their unusual length and slenderness. The first pair arise from the lateral surface of the spine and the stout prezygapophyses. Their distal ends unite with the sacral ribs in forming the great lateral plates. The second and third pairs arise in common from the fused spines of sacrals II and Ill. Distally they diverge to unite with the iliac crest. The fourth pair arise independently; they are directed diagonally backward and attach to the border of the ilium above the lesser peduncle. Midway between the base of the diapophyses and the attachment with the ilium there is a marked rugosity which suggests that the diapophysis may terminate at that point and the distal portion be made up of the tubercular portion of the rib. This could be determined only from the study of a very young specimen.

The sacral spines are remarkably short and blunt. In structure they are made up of a thickened median plate expanded at the crest to form a blunt knob, and flanked in the anterior three by a slight pair of lateral spinous plates. The first is free, but stands close to the second. In structure it resembles the posterior dorsal, though its lateral plates have almost disappeared. The second and third spines are so firmly united as to show no trace of the line of fusion. Their lateral plates arise from the middle of their respective surfaces. The fourth and fifth take on the caudal type of structure, which consists of a median plate expanded into a simple knob at the crest. The fourth stands quite close to the third; the fifth is isolated. All have marked rugosities on the anterior and posterior margins for the interspinous ligaments.



Two anterior caudal vertebrae were found in close apposition with the posterior end of the sacrum. The first has been so distorted in a diagonal direction that the structure of its centrum and zygapophyses can scarcely be determined. The second (Figs. I and 2, Plate LXXV) has suffered less from distortion, and is otherwise fairly well preserved. The centra are in general amphicoelous The anterior end of the first is concave in its lower half, but is so badly distorted that it does not show whether or not it had the characteristic convexity usually found in the upper half of the first centrum. The posterior end is slightly concave, a feature which the crushing has tended to lessen. In the second caudal the anterior end is uniformly concave, the posterior end slightly more SQ. There is no trace of the lateral cavities found in almost all the opisthocoelians. It cannot be determined whether or not these vertebrae were chevron-bearing.

The neural arches are massive, broad on the anterior surface, but drawn to a rounded angle about the posterior opening of the neural foramen. Laterally there are traces of a slight ridge leading downward from the prezygapophyses to the base of the transverse process.

The zygapophyses are slight for a vertebra of such dimensions; the anterior pair has been pushed upward by compressure. The articular facets normally face inward and slightly upward. The posterior pair in the second caudal bears distinct evidence of a hyposphene.

The neural spines are characteristically simple. They are laterally compressed at the base and expanded at the crest into a roughened knob. The anterior and posterior margins bear the strong rugosities for interspinous ligamental attachments which persist throughout the-vertebral series.

The caudal ribs are equally simple. The capitular element is represented by a stout process arising from the lateral surface of the centrum above the middle. It is vertically compressed and apparently expanded into a rounded end. No trace of the tubercular element is to be seen.