The ancestral type must be sought in some such a form as _Henricosbornia_, where the upper molars are brachydont, have the four primary cusps distinct, and the connecting crests of small size, and a cingulum moderately developed on the front and rear sides. Progress is in the line of enlarging the crests, so that, in the later forms, the two external cusps are united to make a wall; and the anterior external and the anterior internal cusps are united into the large anterior lobe; while the posterior external and the posterior internal cusps unite to make the posterior lobe. These may remain relatively simple as in _Rhynchippidae_[14]; or with this simple arrangement of the cusps, the cingulum may be developed into a platform around the anterior, inner, and posterior sides of the molars, as in the _Isotemnidae_; or, with relatively simple molars, the incisors may be specialized into caniniform-like teeth as in the _Leontinidae_; or secondary processes (or cristae) may develop from the wall, making the complicated teeth characteristic of _Nesodontidae_. [Footnote 14: I have abandoned the family term _Notohippidae_, as the genus used as a basis is very little known, and the forms Ameghino assigns to the family, to my mind, mostly belong with the _Nesodontidae_.] For convenience in discussing the modifications of the toxodont tooth, I have, throughout, used the nomenclature illustrated in fig. 52, taking one of the most complicated to show the ultimate development. In the upper tooth there is, first, the external wall, from which springs the anterior lobe, always the larger lobe, and composed of the protocone and paracone of Osborn. In the rear is a smaller narrower posterior lobe, composed of the hypocone, the metacone, and the metaconule of Osborn. Between these is a large basin, which may be subdivided by two cristae into secondary bays, referred to as bays 1, 2 and 3, while the cristae are in the same way referred to as cristae 1 and 2. In some genera, the cristae are entirely wanting, in others incipient. When fully developed, they are most marked in young individuals and, as the tooth is worn, appear progressively shorter. Behind the posterior lobe, there is a variable bay, number 4 which is bounded behind by crista 3, which is apparently a development of the posterior cingulum. This last crista and bay may or may not be present. [Illustration: Fig. 52. _A_, upper molars of Coresodon; _B_, lower molars, somewhat worn—natural size.]