1904.] FROM MADAGASCAR AND EGYPT. 165 



convex antero-posteriorly, but this projection is not bounded 

 posteriorly by a well-marked groove as in ^Epyornis. The centra 

 of the succeeding (post-acetabular) vei'tebi'te do not run in quite 

 the same straight line as those in front, but slope somewhat up- 

 wards, so that the preacetabular region of the synsacrum makes a 

 slight angle with the post-acetabular, the two being separated by 

 the ventral prominence of the inter-acetabular centra. 



In the thirteenth, fourteenth, and fifteenth vertebrae there are 

 ventral processes, of which the anteiior two are perhaps to be 

 regarded as the true sacral ribs ; the outer ends of these processes 

 unite externally into a common mass which abuts on the ilium 

 immediately behind the acetabulum, and at the same time forms 

 the hinder wall of the cavum acetabulare. Of these processes, 

 that of the thii'teenth vertebra is slender and directed backwards, 

 that of the foui'teenth stouter and running directly outwards, 

 that of the fifteenth much the largest and directed forwards ; the 

 centra of these three vertebrae are flat below. 



The sixteenth vertebra bears a long and broad transverse 

 process, which runs outwards and backwards, and widens out 

 externally, joining the ilium immediately beneath the anti- 

 trochanter ; the upper edges of these tra,nsverse processes are con- 

 tinued upwards into thin cribiform plates which unite above with 

 the neural spine. The next transverse process is more slender 

 but bears a similar plate, and, judging from the pelvis of u.E'pyornis^ 

 these plates were present in all the succeeding postsacrals, and are 

 so arranged that they divide the posterior portion of the pelvis 

 into a series of narrow chambers. In the present specimen all 

 behind the seventeenth vertebra are broken away. 



The dimensions of the pelvis are : — cm. 



Length of synsacrum as far as preserved 26 



Approximate width between outer ends of the 



antitrochanters 13 



Width between ventral edges of acetabula 7"5 



Vertical diameter of acetabulum 4 



Width in front of acetabulum 6 



The Femur (Plate 'V . figs. 2 & 3). — The femur is rather more 

 slender in proportion to its length than in xEpyornis mulleri, but 

 otherwise is not strikingly difierent. The neck is very short and 

 the head is shaped somewhat like the frusti'um of a cone, the upper 

 end of which is represented by the smooth surface wdiich fits into 

 the acetabular foramen. The trochanteric surface is continuous 

 with that of the head, and the trochanter itself is less massive and 

 projects less foi'wards than is the case in uEjyyornis : its outer 

 sui'face is deeply marked by the attachments of muscles (glutei, 

 &c.). On the hinder face of the bone, immediately beneath the 

 edge of the trochanteric surface, there is a large pneumatic foramen 

 {pn.f.) partly closed by a cribriform plate of bone. 



The shaft is slightly compressed owing to the flattening of the 

 anterior and posterior surfaces, and this flattening is especially 



