Skulls of adult males of spadix from Elk River, Minnesota, as compared with those of longicauda from Alberta, are larger in every part measured. Relative to the basilar length these skulls of spadix are broader across the mastoid region, narrower across the zygomata, deeper through the plane of the postorbital processes, shallower through the braincase and have relatively shorter tympanic bullae. Whereas the tympanic bullae of longicauda are, on the average, approximately as long as the rostrum (orbitonasal length), in spadix the rostrum is longer than the bulla. Viewed posteriorly, the braincase of spadix is seen to be much shallower and wider than that of longicauda. Indeed, the depth of the braincase, measured at the anterior end of the basioccipital, amounts to only 56 per cent of the mastoid breadth in spadix as against 61 per cent in longicauda. The longer, waistlike, postorbital constriction, relatively smaller braincase, and especially the relatively narrower zygomatic expanse in spadix imparts to its skull a more slender appearance than has the skull of longicauda. These differences are not shown by the skulls of females. To be sure, spadix, in most of its cranial measurements, averages slightly larger, has a relatively shallower braincase and is relatively deeper through the postorbital processes, but these differences are so slight that inclusion of one more specimen, of slightly different proportions, in the average might cause the average measurements to read as they do in longicauda.
Compared with noveboracensis, from Massachusetts, adult skulls of spadix, taking sex into account, are larger in every part measured and are relatively as well as actually wider and deeper throughout. Also, in spadix: Sagittal and lambdoidal crests higher, especially in females; anterior margin of tympanic bulla projecting up sharply from squamosal; occiput more flattened in posterior view; tooth-rows relatively and actually longer but orbitonasal length relatively shorter; postorbital processes more robust; zygomatic arches widely bowed outward rather than evenly rounded; canines larger; squamosal less swollen ventrally, especially in females. Between noveboracensis and spadix, the differential cranial characters are greater in number and degree between females than between males. Comparison of the skull with that of M. f. primulina is made in discussion of that subspecies.
Remarks.—Edgar A. Mearns in 1889 and the early nineties took several specimens of this weasel and it was principally on these that Bangs in 1896 (p. 8) based his description. The best material, however, is that from Elk River, Minnesota, collected in later years by Bernard Bailey, and supplemented by one specimen taken in 1885 by Vernon Bailey and another by his sister Anna Bailey in 1891 at the same place.
Mustela frenata spadix has just one structural feature of a "unique" kind which serves to differentiate it from the geographically adjoining subspecies. This feature is large size. The other diagnostic characters ascribed to spadix are of an intermediate sort—intermediate as between two extremes, one found to the westward in longicauda and the other to the eastward in noveboracensis. For example, the dark-colored upper parts are merely darker than in longicauda and merely lighter than in noveboracensis. The color is not "different"; it is only "intermediate." Furthermore, each of the characters ascribed to spadix, including large size itself, undergoes change from one part of its geographic range to another; the characters are not constant over a wide area. Indeed, excepting the large size which remains relatively uniform over the northern two-thirds of the range, no two localities have been found from which the specimens can be said really to agree in characters.
By way of illustration, the coloration of the upper parts may be cited. Near the range of noveboracensis the average coloration of individuals from one locality is only a little lighter than in noveboracensis. Farther westward the average coloration is a little lighter and farther westward yet, toward the range of the extremely light colored longicauda, the average coloration is lighter still. Although all these animals are darker than longicauda and lighter than noveboracensis, those from the three places do not agree among themselves. Because of the lack of more than one character of a "unique" kind and because of the inconstancy, geographically, of other characters, and for that matter, lack of constancy geographically in combination of characters, the writer regards spadix as a barely recognizable subspecies.
Examination of the specimens of spadix shows that the individual variation in a single species is greater in a region of intergradation than it is some distance inside the borders of the geographic range of a well-marked subspecies. This is illustrated by three specimens of M. f. spadix in fresh summer pelage from the single locality, Elk River, Minnesota. In these, the color of the upper parts varies from a little darker than Cinnamon Brown to Vandyke Brown. At any one locality well within the range of longicauda, or noveboracensis, there is nowhere nearly so much variation in color, even in much larger series of specimens.
Study of the specimens here assigned to spadix reveals that some features regarded as of diagnostic value for one or the other of the two races, longicauda and noveboracensis, behave differently. For example, the dark coloration of the upper parts, which is characteristic of noveboracensis, manifests itself far westward within the range of spadix whereas the wider extent of the light-colored underparts, which is characteristic of longicauda, and the Olive Ocher, rather than Pale Orange Yellow, color of these underparts, are seen in varying degree all the way across the range of spadix. Thus, these animals are colored above like noveboracensis and below like longicauda, but not vice versa. In these animals, then, the longicauda type of underparts is dominant, in one sense of the word, over the noveboracensis type of underparts, and the noveboracensis type of upper parts is dominant over the longicauda type of upper parts. Each of these features is subject to actual intergradation and does not always behave as a "unit character," that is to say, one which is either present or absent. However, the noveboracensis type of upper parts is carried much farther west before being diluted than is the noveboracensis type of underparts. Indeed, within the range of noveboracensis itself, the broad extent of the longicauda type of underparts is manifest. This is, of course, near the western margin of the range of noveboracensis.