I VOLUTION \KY SUMMARY 27 



tion as in the type of Styracosanrus. Very rarely, as in Chasmosaurus kaiseni and Cera/ops montanus, 

 arc those horns of even moderate length. The subsequent forms are all well endowed with brow 

 horns although these are not always equally long, as in Trkeratops brewcornus and Triceratops 

 alticornis by way of contrast; nevertheless, there is no evidence of retrogression. 



There is a tendency in Belly River genera, but notably in the short-crested Centrosaurus and 

 Styracosanrus, toward the development of peculiar horn-like excrescences around the periphery of 

 the parietal. The former genus may be ancestral to such later Ceratopsia as Triceratops, the latter 

 I think died without issue. At all events, these excrescences disappear with the Belly River-Two 

 Medicine forms, for all of the Lance types are conservative in the ornamentation of the crest, if 

 such it was. If, as I think, these growths were due to a sort of acromegalic response to the chemical 

 environment, the altering of conditions might lead to a more normal racial health and hence their 

 disappearance. With the long-squamosal types, the height of this ornamentation occurs in Anchi- 

 ceratops of the Edmonton, lessens but is still apparent in the Fruitland Pentaceratops, but in Toro- 

 saurus of the Lance the crest is bereft of all appendages. Thus, in two details, in the length of the 

 nasal horn, but not its position, and in the ornamentation of the posterior part of the crest, the long- 

 squamosal phylum lags behind that of the short-crested types. 



The closure of the parietal fenestrae is foreshadowed in the Belly River types and complete 

 in the Lance in the short-crested phyla, while in the long-crested types the fenestrae never disappear. 

 Anchiceratops, although with comparatively long squamosals, has relatively small fenestrae. It may 

 be a precocious side line culminating in the Edmonton, although it is undoubtedly related to the 

 Chasmosaurus-Pentaceratops-Torosaurus line. 



The long crests seem to have been less closely invested in the horny covering than were the 

 short crests, if one may judge by the vascular impressions. Although, if this means that the crest in 

 the former was not free of the dorsal body musculature, the creatures must have had a singularly 

 immobile head. It would seem, on the contrary, that when the fenestrae persisted there was a 

 thicker, more fleshy covering over the crest both above and below, except perhaps at the periphery 

 in Chasmosaurus and Pentaceratops. Anchiceratops, with its tendency to close the fenestrae, has deep 

 vascular impressions in spite of its phyletic relationship. The variable extent of vascular impressions 

 on the under side in the closely invested crests seems to indicate more accurately the extent to which 

 the crest is free from the body. 



As an outcome of the present study, I have attempted to present graphically, in Figure 3, cer- 

 tain conclusions in regard to the phyletic sequence of the Ceratopsidae. They are in no sense final, 

 and the figure is to be regarded merely as a trial sheet which is imperfect, due perhaps to misinter- 

 pretation, but more, I hope, to the incompleteness of the record. 



The two main phyletic lines, which may be of sub-family rank, are distinguished principally by 

 the form of the crest, specifically in the length of the squamosal and its reaction upon the crest, and by 

 either the persistence or closure of the parietal fenestrae. 



The long-crested line, with persistent fenestrae, begins with the Belly River Chasmosaurus, of 

 which Chasmosaurus belli is the most generalized form, while Chasmosaurus kaiseni is progressive 

 in the development of its brow horns, although it may be an actual contemporary with belli of the 

 rudimentary horns. In the Edmonton, this phylum is represented by Anchiceratops; but this genus, 

 while showing certain characters comparable to those of Pentaceratops of the homotaxial Fruitland 

 and Kirtland formations, seems to depart from the main evolutionary line, especially in the high 

 degree of vascular impressions on both surfaces of the crest, and in the tendency toward the diminu- 

 tion of the fenestrae. According to available evidence, it seems to represent an abortive side line 

 which left no descendants. This is not true of Pentaceratops, however, for the typical species stern- 

 hergii could be the lineal descendant of Chasmosaurus kaiseni with little alteration, other than of size 

 and the more abundant vascularity of the squamosal. It is far removed in space from its predeces- 

 sors and descendants, but otherwise fits into its place in the sequence admirably. Arrhinoceratops, 

 also of the Edmonton, presents certain difficulties. It is clearly a long-squamosal form with vascular 

 impressions on either side of the crest and rather small fenestrae. It may prove to be ancestral to 

 Triceratops of the short-crested phylum and the resemblance is rather close, especially between 



