398 HARRIS HAWTHORNE WILDER ON 
are very variable in their mutual relations. Occasionally all three will be distinct and of 
about the same size, but usually the ventral spine is much smaller than the other two. 
There may also be varying degrees of bifurcation of the genuine transverse process, or it 
may present one or more small processes in addition to the usual three. 
Vertebra bearing the first haemal arch. — This is normally the 23d in the series, but 
it may be the 22d or even frequently the 24th. After the sacrum, the neural plates rap- 
idly narrow, and, by the time the vertebra in question is reached, this part has become 
about as narrow as the base of the neural spine, so that the latter appears in direct con- 
tinuation with it, the posterior zygapophyses appearing as lateral processes borne upon 
the sides of the spine. This increase of the apparent length of the neural spine renders 
it strikingly similar to the haemal spine, and thus is seen the first suggestion of that 
dorso-ventral bilaterality which becomes so marked in the more posterior caudal vertebrae. 
The transverse process is thin and poorly developed, and the occasional bifurcation of its 
free end gives.an indication of its division into tubercular and capitular portions as seen 
in the vertebrae immediately posterior to this. The haemal arch, which is a new element 
and thus gives distinction to this and the following vertebrae, begins as a pair of low, thin 
ridges springing from the ventral side of the centrum and uniting across the mid-ventral 
line near the posterior limit of the vertebra in such a way as to frame in a haemal 
foramen just ventral to the centrum, which, with the foramina of the succeeding verte- 
brae, forms the haemal canal. The ventral laminae are perforated as usual by the two 
ventral foramina, and, as they are nearly covered by the two ridges that form the roots 
of the haemal arch, a pair of foramina occurs also in these latter opposite the usual 
ones. In most instances the first haemal arch appears, as it were, suddenly and perfectly 
developed, and the vertebra immediately preceding the one that bears it usually shows no 
suggestion of such a structure. Occasionally, however, a rudiment of the foot or root of 
the arch appears upon one or both sides of the previous vertebra in the form of a thin 
plate lying parallel to the ventral lamina, and extended across from centrum to outer end 
of the transverse process, and, in one instance observed, a slender haemal process appeared 
upon the left side, which projected back over the following vertebra and plainly repre- 
sented one side of a rudimentary haemal arch. There was no trace of this upon the 
other side. 
The caudal vertebrae posterior to the above, with especial reference to the terminal 
ones. —The most conspicuous characteristic of the caudal vertebrae posterior to the above 
is the presence of nearly equal neural and haemal arches running out into long spines 
with their free ends pointed obliquely backwards. The reduction of the transverse 
process, which begins just posterior to the sacrum, progresses rapidly after passing the 
first haemal arch. The process first becomes divided into its two portions, the dorsal one 
