FAMILY, BUNOMOSTODONTIDyE 331 



they constitute a distinct subfamily, which may be named the Kotoros- 

 trinfe,^ or short- jawed mastodons of the South American continent. 

 Tliere is a very great variety of form among the ten species whicli have 

 been described, including, in addition to those mentioned above, M. 

 holivianus Philippi, M. chilensis Philippi, M. platensis i^meg., M. rectus 

 Ameg., M. argentinus Ameg., and M. superhus Ameg. The two beautiful 

 skulls in the Stockholm Museum, referred to M. andium by Nordenskjold, 

 display a spiral twist of the tusks — a feature quite unique among Pro- 

 boscideans. 



Family, Mastodontid^e 



As pointed out by Matsumoto, the lophodont Palceomastodon of North 

 Africa is readily distinguished by the absence of intermediate cones and 

 trefoils, which block the valleys in the contemporary genus Phiomia; its 

 grinding teeth are |)urely crested or lophodont. It is also distinguished 

 by its hilophodont superior intermediate molars, whereas those of Phio- 

 mia are fully trilophodont. The lower intermediate molars of Palceo- 

 mastodon have only two complete crests and one small lobe representing 

 a third crest instead of the complete third crest of Phiomia. Thus they 

 have not attained the trilophodont stage characteristic of the genus Mas- 

 todon, but may lead into that stage. Although the skull is not fully 

 known, it is relatively broader and shorter than that of Phiomia. The 

 relative rarity of this animal in both the American Museum and British 

 Museum collections is also in keeping with the theory that it was a forest 

 dweller and that its remains, like those of all the Mastodontinae, escaped 

 fossilization until the true mastodons appear in enormous abundance in 

 the Pleistocene forest formations of eastern North America. 



It now appears probable that the Mastodontidffi sprang from the genus 

 Palceomastodon of the Oligocene Fayum deposits of North Africa, while 

 the Bunomastodontidaa sprang from the genus Phiomia of the same de- 

 posits. From materials in the American Museum hitherto undescribed, 

 Matsumoto has positively separated these two genera, the species of which 

 have been more or less confused ever since the original descriptions of 

 Andrews of Palwomastodon in 1901 and of Phiomia in 1902. Pliiomia 

 is certainly ancestral to the Bunomastodontida^ only, in fact, barely sepa- 

 rable generically from Trilopliodon angustidens, the typical longirostrine. 



The name Palceomastodon, which has figured in all the literature from 

 the date of its establishment to tlie present author's preceding paper 

 (1921. 514), therefore applies only to the lophodont type of P. headnelli, 



^ From the Latin of Virgil, "notus," a southwest wind : Greelc, "Notogwa," the South 

 American region. 



XXIII — Bull. Geol. Soc. A.m., Vol. ?,2, 1920 



