Evidence in support of this hypothesis is given below.
Along the Pacific Coast from British Columbia southward, M. erminea (see fig. [25] on page [95]) is brown in winter. This is an area where snow rarely falls and the temperature in winter ordinarily is above freezing. In the remaining part of the American range of this species the temperature in winter is below freezing much of the time and snow remains throughout the winter or for long periods. In this colder part of the animal's range, only white coats occur in winter. M. frenata likewise has a white coat in winter in the part of its geographic range where snow and freezing temperatures prevail throughout most of the winter and a brown coat in warmer, snowless areas to the southward and along the Pacific Coast. The third species, M. rixosa, exhibits a corresponding correlation between coat color and climate. On the Asiatic continent, several species, including M. erminea, provide parallel correlations and nowhere are there any exceptions for the subgenus Mustela. These data are an important part of the material on which we have based the induction that the underlying cause of seasonal change in color is a need for protective coloration.
As regards molt, most naturalists who have written upon the subject regard it as responsible for the change from the white winter coat to the brown summer coat. However, the change from brown summer coat to white winter coat has been thought by several writers to be effected by change in coloration of the individual hairs. Among those holding this opinion there may be cited Bell (1874:197) in reference to Mustela erminea, and Coues (1877:123) in reference to American specimens to which he applied the same name. More lately Hadwen (1929) has taken this same view, and Gunn (1932) also discusses the possibility of the hairs changing color. Bachman (1839:228-232), Macgillivary (1843?:158), Audubon and Bachman (1851 (vol. 2):62), Schwalbe (1893:538), Pearson et al. (1913:447), Miller (1930, 1931A), Hamilton (1933:300) and Rothschild (1942), among others, have been inclined to the opinion, or positively affirm, that the color change in autumn is the result of a molt. The papers cited above contain, in turn, references to many other printed accounts dealing with this question.
To my mind, it has not so far been demonstrated that the change in color of weasels in autumn is accomplished without a molt. Also so far as I am aware, no explanation has been given of how the pigment may disappear from the hair of weasels. Metchnikoff's (1901:156) idea that the senile whitening of the hair in man is accomplished by phagocytes which remove the pigment granules would hardly seem to explain the relatively sudden and complete autumnal change occurring in weasels. Anyhow, Danforth (1925:108), and some other students have thought that the action of these phagocytes was at most a factor of slight importance in the whitening of hair. Whatever be the complete answer to the question of how the weasel changes color in autumn, at least one specimen of long-tailed weasel, which is in process of color change in autumn, presents clear evidence of molt of the overhairs. This specimen of M. f. longicauda is no. 188408, U. S. Nat. Mus., taken on November 12, 1897, at Rapid City, South Dakota. Other specimens of M. erminea which were taken in autumn similarly show molt to be in progress. For these and other reasons, I am inclined to the opinion that the autumnal change in color, like the one in spring, is effected by molt. During the period of the autumnal color change, Noback (1935:27) had a captive M. f. noveboracensis and, each morning, found clumps of brown hair on the floor of its cage; this was strong indication that molt was responsible for the color change in this instance.
However, I freely admit that the evidence does not prove that the change from brown to white can be accomplished only by molt; in the present state of knowledge it would be unscientific to deny that the change were possible of accomplishment by other means. Also, it is true that the fifteen specimens before me of Mustela frenata, subspecies included, in process of change from brown to white, with the exception of the one from Rapid City, South Dakota, if taken individually, do not, in macroscopic examination, show definite molt lines or other absolutely convincing evidence of molt. However, these same specimens, insofar as examined microscopically, do show overhairs all white, or overhairs pigmented throughout. The lighter color of the proximal parts of the overhairs in itself should not be accepted as evidence of color change, for in the fresh summer pelage, the same condition exists. Also, careful macroscopic examination suffices to show that in the transitional pelage of autumn, the brown overhairs generally are longer than the intermixed white overhairs.
Whether the underfur behaves in exactly the same way as the overhair, I have not myself definitely ascertained, but I assume that the underfur is molted twice each year, at least in the northern populations of Mustela frenata and in the other species of more northern distribution. Schwalbe's (1893) work, including sectioning of the skin and study of the hair follicles, led him to conclude that the underfur was molted twice each year in Mustela erminea.
In Mustela frenata noveboracensis, M. f. nevadensis, and M. f. nigriauris, measurements taken on adult males show the overhairs to be longer in the winter pelage than in the summer pelage of specimens from the same locality. For example, in M. f. nigriauris from Berkeley, California, the overhairs of the summer coat (July and August) average 8 millimeters in length on the hinder back and 7 mm. on the belly, but average 9.5 mm. and 8 mm. respectively in January-taken specimens possessing the full winter coat. At Ann Arbor, Michigan, in the summer coat, the longest hairs on the hinder back average approximately 12 mm., and those on the belly, 9.5 mm., against 13 mm. and 9.5 mm. respectively in winter. Although general observations initially led me to believe that the black, terminal hairs of the tip of the tail are longer in the winter pelage than in the summer pelage, actual measurements fail to show a difference in length.
The change from one coat to the other in the long-tailed weasel has been described among others by Miller (1930, 1931A), Hamilton (1933) and Glover (1942) on the basis of captive specimens. In a general way, the progress of the molt in their specimens agrees with that which I have been able to make out from examination of skins taken in the wild. There is, however, this difference: Their specimens show a more spotted pattern when in process of hair-change than do specimens taken in the wild. Probably the more or less unnatural conditions under which these captive animals lived modified the normal progress of molt.