MORPHOLOGICAL REVISION OF THE SUBORDER. 



119 



45. — Cross-section through 

 third dorsal of D. gigas. 

 No. 1002 University o( 

 Chicago. X 1. Show- 

 ing sudden contraction o( 

 sides of centrum below 

 notochordal canal. 



The vertebrae reckoned as the ninth dorsal, probably the fifteenth or sixteenth of 



the series, is similar to the posterior dorsals of Dimetrodon incisivus. The keel is still 



narrow and sharp, but is less deep than in the anterior dorsal, being not over one-fifth 



the vertical height of the centrum. The centrum is longer than high or wide and 



the anterior and posterior faces have become rounded. 



The spine of this vertebra is one of the first that is nearly 

 perfectly preserved. In comparison with the spines of the 

 other species the spines of Dimetiodon gigas do not exhibit 

 the swelling shortly above the base, and the point is marked 

 by a slight rugosity of coarse striations only. Just above the 

 rugosities the spines assume a quadrilateral or circular section, 

 which they maintain to the tip, not showing the peculiar flat- 

 tening of specimen No. 112 University of Chicago (fig. 14). 

 The diapophyses have become shortened and stand out from 

 the vertebra more than in the anterior portion of the series. 

 A rib attached to this vertebra has a rather short tuberculum 

 and long capitulum corresponding to the shortened diapophysis 

 of the vertebra (fig. 47). 



The lumbar vertebrcE : The posterior ones of the separate 

 vertebrae show the capitular face on the edge of the centrum 

 characteristic of the lumbars. They are crushed too badly to 

 admit of accurate description, but all have the general form and proportions of the same 

 vertebrae in Dimetrodon incisivus. The spines of the anterior ones are elongate and 

 have much the same shape as the dorsals, but the posterior ones are shorter and with 

 the base narrower and elongate antero-posteriorly. Near the middle the spine becomes 

 more quadrilateral. The spine rises from the posterior portion of the neural arch. 



The sacrum (plate 17, figs. 3 and 4) is represented by the first two vertebrEe which 

 are closely coossified and have the sacral ribs complete. The neural arches are badly 

 rotted and the spines are nearly gone, but enough remains to show that they are short 

 and slender compared with those of the dorsals and anterior lumbars. This sacrum 

 differs from that of Dimetrodon dollovianiis^ No. 1 14 University of Chicago, especially in 

 the form of the ribs, and resembles more closely that of Diinetrodon incisivus, No. 4008 

 Am. Mus. The centra are abruptly more elongate than the posterior lumbars and are 

 firmly attached to each other, but not coossified. The sides are concave, but there is 

 no keel on the median line of the lower side. The firm union of the two excludes 

 the intercentrum from between them, but it is present as a very thin and scale-like 

 element, coossified with the posterior of the two vertebrce, and in part underlying the 

 point of union of the two. The intercentrum between the posterior lumbar and the 

 first sacral is firmly united with the first sacral ; it is thin and small and projects for- 

 ward under the line of union of the two. The zygapophyses connecting the sacral 

 vertebrte are much reduced in size and are very closely united. 



The sacral ribs are supported on strong, short transverse processes rising from 

 the neural arch and centra and coossified with them. The sides of the arches are 

 swollen and expanded laterally in the position of the transverse processes to support 

 the ribs. The rib of the first vertebra is attached to the anterior end of the vertebra 

 such a manner that the anterior half of the proximal end extends beyond the 



m 



anterior end of the centrum as an articular process which embraced the posterior edge of 

 the last lumbar. The ribs of the two sides form a pair of projections on the anterior 



