340 'My. C. W. Andrews on the Development of 



beyond the underlying thin portion of the scapula, so as to be 

 visible on tiic ventral surface. Probably, however, in a later 

 stage the scapula grew forward and inward, coin])letely 

 shutting in the clavicles and extending the symphysis of the 

 scapukt to their extreme anterior end. This appears to have 

 taken place in the shoulder-girdle of Coli/nibosaurus figured 

 by Seoley * and in that of Elasmosaurus figured by Copef, 

 in both of which genera, however, the clavicles are unknown. 

 The general form of the coracoids at this stage is shown in 

 fig. 1. These bones are extremely thick and massive in the 

 inter-glenoidal region, but further back become very thin 

 except at the lateral borders, which are thick and rounded 

 and are produced postero-laterally into short stout prolonga- 

 tions with truncated ends. 



Successively younger stages may now be com^pared with 

 the mature or nearly mature condition above described. 



Tiie first of these is represented by a left scapula, which, 

 allowing for more imperfect ossification, closely resembles the 

 last, and no doubt belonged to an individual of the same 

 species. In this specimen the anterior ramus was already 

 well develo|)ed and had grown back in the middle line for 

 some distance towards the median prolongation of the coracoid, 

 but had not yet quite reached it, as can be seen from the fact 

 that it terminates posteriorly in a sharp edge. The syra- 

 physial surface is niaiked with pits and vascular impressions, 

 but probably was still separated from the scapula of the 

 opposite sidi'. The thin anterior portion of the scapula is 

 much smaller than in tlie last stage, and does not extend so 

 far forwards and inwards, so that the scapulai were separated 

 in front by a much more widely open V-shaped interval, and 

 a much larger part of the clavicles would be visible from the 

 ventral surface. The dorsal ramus is shorter and its anterior 

 border was incompletely ossified. The clavicle is unfortu- 

 nately wanting, but the impression of its posterior edge shows 

 that it occupied exactly the same position in relation to the 

 scapula as in the last specimen. 



The next stage is represented by a shoulder-girdle (no. 05) 

 in Mr. Leeds's collection (fig. 3 A)X ; the left scapula only is 

 wanting, but the bones are somewhat crushed. This crushing- 

 does not, however, affect the anterior ramus of the scapula, 



* Quart. Joiiru. Gool. Soc. veil. xxx. (L'^74) p. 447. 

 t Traas. Amer. Phil. Soc. vol. xiv. (1870) p. 51, tio;. ~. 

 X The clavicle should be placed a little further inwards than is shown 

 in this tijjure. 



