The family Heteromyidae was defined by Wood (1935:81) essentially as follows: Cheek teeth brachydont to hypsodont and even rootless; usually six cusps per molar, three on each loph; enamel rarely divided into two plates, never reduced to one; skull light, thin and papery; mastoids inflated, mastoidal breadth often greatest, never appreciably less than zygomatic breadth; interorbital space wider than rostrum; palate nearly horizontal and little if any below level of zygomata; nasals extended beyond incisors; zygomata slender, with greatly reduced malar, almost, or quite, abutting against tympanic; frontals and parietals broad, with latter reaching, or nearly reaching, orbits; frontal trapezoidal; parietal quadrate, to pentagonal or triangular; interparietal primitively large, secondarily reduced; squamosal mostly or entirely confined to orbit; tympanic vesicular and inflated, in some forms highly inflated; mastoids inflated and bullous, reaching top of skull, and forming part of occipital surface; occipitals contracted and limited in area on occiput, but extending onto dorsum of skull; coronoid processes small, inclined caudad and lying below level of condyle; jaw small and weak with large, everted angle; tail as long as, or longer than, head and body; claws of manus elongate, fossorial, but forelimb slender; pelage usually coarse and frequently spinose; ears and eyes large; body murine in form; locomotion in many forms saltatorial.

This characterization of the family includes all of the members of the subfamilies Perognathinae, Heteromyinae and Dipodomyinae as well as the genus Microdipodops which I am disinclined to place with any of the three subfamilies. Apparently it is more closely related to the subfamily Perognathinae.

The subfamily Dipodomyinae, which contains the genera Dipodomys, Prodipodomys and Cupidinimus, might be characterized after Coues' (1875) original description of the subfamily as follows: Cheek teeth progressively hypsodont, in Dipodomys becoming ever-growing; enamel progressively interrupted, eventually reduced to anterior and posterior plates; upper and lower third molars reduced in size; tooth pattern rapidly destroyed, leaving only an enamel oval; upper incisors smooth (some fossils) or grooved (living forms); progressive expansion of the auditory bullae and increase in saltatorial ability; pterygoid fossa double; calcaneal-navicular or even calcaneal-cuneiform articulation; tail tufted.

Owing to the fact that so little paleontological material is known and because even that is fragile and not easily accessible for study, knowledge of the fossil representatives has been drawn primarily from the literature, especially from Wood's (1935) account.

Heteromyids are known from the Chadron formation, of early Oligocene age, in which a single tooth was found. In the Orellan stage of the mid-Oligocene where the genus Heliscomys occurs, it is notably generalized, in comparison with other members of the family, but it may not be ancestral at all. The lower premolar is tricuspidate and the first and second molars are quadritubercular with a broad cingulum. The teeth are bunodont and brachydont, with the cusps not uniting to form lophs. Wood (1935:78) shows Mookomys formicorum (from the Arikeean) as the next heteromyid in the evolutionary sequence and postulates that this species arose from Heliscomys gregoryi. Mookomys is judged by Wood to be the common ancestral form of the perognathines and the dipodomyines.

Cupidinimus, the genus next in line, is characterized by smooth upper incisors; lower molars with incipient H-pattern; cheek teeth progressively hypsodont and lophate (but always rooted); and calcaneal-navicular articulation.

The time range of this genus is from the late Miocene (Niobrara River, Local Fauna) of Nebraska to the medial Pliocene, Thousand Creek (Hemphillian) of Nevada.

Hibbard (1937:462) described Dipodomys kansensis from the Ogallala formation (Hemphillian age) of Kansas. He redescribed his species, and made it the type of the new genus Prodipodomys (Hibbard, 1939:458), differentiating it from Dipodomys on the basis of the three-rooted p4, double-rooted m1 and m2 and the single rooted m3. It is shown to be closely allied to Dipodomys by the form and position of a large foramen posterior and labial to m3, and by the development of the masseteric ridge.

The next youngest heteromyid fossils which have been described are of the genus Prodipodomys? from Arizona. Gidley (1922:123) described Dipodomys minor from the Benson (Blancan) which Gazin (1942:486) refers to the genus Prodipodomys?. Wood (1935:156) described Dipodomys gidleyi from the Curtis (Pleistocene). Both of these species are primitive as regards dentition; that is to say, the enamel ring of the tooth is complete and lacks any sign of a break. The limb bones of D. gidleyi show lesser saltatorial ability, and therefore appear to be more primitive, than those of any living Dipodomys.

Several heteromyids which have not been assigned to any genus are known. Wilson (1939:36-37) recognized some from the Avawatz (Clarendonian) and the Ricardo (Clarendonian). Another, possibly of the genus Diprionomys?, from the Barstow (Barstovian) was described by Wood (1935:197) as follows: "The general shape of the tooth as figured strongly suggests either one of the most advanced species of Dipodomys or else a Geomyid.... It is much more advanced than are any known contemporary heteromyids, and compares fairly well with such late Tertiary and Pleistocene geomyids as have been described. It certainly is not referable to any known heteromyid genus other than Dipodomys, and should probably be called a Geomyid." Wilson (loc. cit.) refers to these specimens as Dipodomyine (?) n. gen. and sp. If these specimens referred to by Wood and Wilson are true heteromyids then a change in the phylogenetic scheme proposed by Wood (1935) would be necessary. Wilson (loc. cit.) says, referring to the Avawatz specimen, "The cheek teeth are very hypsodont but are apparently not persistent in growth,... Wide enamel breaks are present in M/1 dividing the enamel into anterior and posterior bands. The enamel of P/4 is complete in the present stage of wear, but an examination of the tooth indicates that breaks would develop with additional attrition at the buccal and lingual margins of the metalophid, and at the buccal border of the protolophid. The incisor is of the slender heteromyid type."

Wood (1935:118) in referring to the ancestry of Cupidinimus with regard to the grooving of the incisors says: "The philosophy of evolution which would prohibit its derivation from Mookomys, because of the grooved incisors in the latter genus, would require a separate line leading back at least to the Lower Miocene."

In view of the above statements, it is conceivable that additional material will be found carrying the dipodomyine line back into the early Miocene. Perhaps the line involving Mookomys and Cupidinimus which was regarded by Wood as the line of descent, is merely an aberrant side branch that parallels in its structures the main line of evolution of the dipodomyines (Figure 1).

As Wilson (1939:37) says: "Indeed it is hard to recognize such a form as Cupidinimus nebraskensis as directly ancestral to Dipodomys in view of the occurrence of the much more advanced Avawatz specimen in deposits that are at most only slightly later than those in which the former is found. The kangaroo rats were apparently much farther along in their development by lower Pliocene time than heretofore supposed."

Wood (1935:78) suggested that Dipodomys gidleyi gave rise to Dipodomys spectabilis and Dipodomys ordii, and Dipodomys minor gave rise to Dipodomys compactus. However, my own study indicates that Dipodomys compactus is conspecific with Dipodomys ordii and should stand as Dipodomys ordii compactus. Consequently a different phyletic arrangement than that proposed by Wood (loc. cit.) is required. Since D. compactus is more closely allied to Prodipodomys? minor than D. ordii is to D. gidleyi, it is possible that P.? minor gave rise to D. ordii and that D. spectabilis is the end product of the phyletic trend of D. gidleyi (Figure 1).

The trend of phyletic development in the dipodomyines has been toward the saltatorial habit. To acquire this habit from a scampering ancestor, certain morphological modifications were necessary. Among these modifications were a lengthening of the tail, a lengthening of the hind legs, the development of a calcaneal-navicular-ectocuneiform contact instead of a calcaneal-navicular contact for additional strength in leaping, a shortening of the forelimb, an increase in size and inflation of the mastoid and tympanic portions of the skull with a consequent reduction in size of the interparietal region and the fusion of certain of the cervical vertebrae. Late Miocene (Cupidinimus) and Pliocene (Avawatz specimen and Prodipodomys) forms had acquired certain of these morphological modifications that are present in the modern genus Dipodomys.