the other, more definite statements about ancestral positions of the small Oligocene forms can hardly be made. The deciding considerations for authors who placed Palaeogale and Bunaelurus as ancestral to Mustela were the absence of a metaconid on M₁ and the trenchant talonid of that tooth. These characters are found also in Mustela. On the other hand certain structures in the basicranial region of Palaeogale and more especially of Bunaelurus indicate that these genera possibly are not close to the ancestral form of Mustela . . . Martinogale may stand near the ancestral form of Mustela and . . . Pliogale may be ancestral to Martinogale. Pliogale, in turn, may have had an ancestor similar to Miomustela. If this should prove to be the case, Palaeogale and Bunaelurus might be regarded as an independent branch which displays merely a parallelism to Mustela in the loss of the metaconid on M₁ and the development of a trenchant talonid on that tooth. The writer would make it clear that he does not hold such to be the case. The ancestral relation of Martinogale to Mustela is presented merely to show the possibility, and not the special probability, of such an origin for Mustela. Knowledge of the tympanic bullae and other structures of the basicranial region would go far toward answering the question and until these structures are known [in mustelids of the Later Tertiary,] some uncertainty will remain."

At the present writing I can add to the above statement only a few facts. The discovery of better material of Bunaelurus than was available to previous workers led Simpson (1946), correctly I think, to synonymize Bunaelurus with Palaeogale. Simpson figures the cranial foramina in Palaeogale. The differences, between Palaeogale and Mustela, in cranial foramina, possibly are only the result of the elongation of the tympanic bullae. The bullae of the subgenus Mustela are seen to be much elongated posteriorly if comparison is made with the bullae of earlier mustelids. Consequently, it might be concluded that there is nothing in the arrangement of the cranial foramina which would preclude the derivation of Mustela from Palaeogale. However, the anterior situation of the carotid foramen—well forward along the medial margin of the tympanic bulla—is a character typical of other mustelids and the posterior location of this foramen in Palaeogale might indicate that it was not ancestral to Mustela.

SKELETON AND DENTITION

The outstanding features of a weasel's skeleton are its length and slenderness. Whereas the length of the vertebral column measured from the atlas (the first cervical vertebra) to the last sacral vertebra is 175 per cent of the length of the hind leg (as measured from the head of the femur to the tip of the longest claw), the corresponding percentage is only 116 in the raccoon. Stated in another way, the vertebral column and the hind leg are of approximately equal length in a raccoon, but in a weasel the vertebral column is one and three-fourths times as long as the hind leg.

VERTEBRAE

The vertebral column consists of 7 cervicals, and ordinarily 14 thoracics, 6 lumbars, 3 sacrals and, depending on the species, 11 to 23 caudals. For the three species of which skeletons were examined, variations from the normal number of vertebrae are noted in the following table:

Table 1
Data on vertebrae in three species of the subgenus Mustela
(Numerals in parentheses indicate number of specimens)

Variation according to the species is evident in the number of caudal vertebrae, but in the other categories of vertebrae no consistent difference in number according to species was found in the material examined. Apparently there is also some geographic variation in the number of caudal vertebrae within a species. For example, the one skeleton seen of Mustela rixosa eskimo (no. 219036, U. S. Nat. Mus., from St. Michaels, Alaska) has only 11 caudal vertebrae, whereas in the 11 Mustela rixosa rixosa from Roseau County, Minnesota, the usual number is 15 with extremes of 14 and 16. Similarly specimens of Mustela frenata from Idaho and California almost always have 1 or 2 more caudal vertebrae than do individuals of the shorter-tailed subspecies of the same species from eastern Kansas.

Of the vertebrae, only the cervicals, of which there are 7, were found to be constant in number. In M. erminea, two of the seven individuals in which the anticlinal vertebra was the 12th (instead of the 11th) had 15 instead of the customary 14 thoracic vertebrae. In M. frenata, seven of the twenty-seven individuals in which the anticlinal vertebra was the 12th (instead of the 11th) had 15 instead of 14 thoracic vertebrae. The one M. erminea with a pseudosacral vertebra had only two instead of the customary 3 sacral vertebrae but the same individual had 15 thoracic vertebrae.