Given any one of the above structural features, say, dorsal outline of the skull, several characters may be provided by it. For example, weasels of the species Mustela frenata have the dorsal outline of the skull convex in southern Louisiana, straight in Missouri and concave in North Dakota, thus providing three characters. This is geographic variation. These variations, characters in zoölogical parlance, when plotted on maps, reveal the geographic occurrence of, say, the convex shape of the skull. In combination with other characters, for example, dark color and short tail, basis is provided for recognizing a subspecies, in this instance Mustela frenata arthuri of Louisiana. Because the change from convex to flat skull takes place geographically at about the same place (in eastern Texas) as does the change from short tail to long tail, and the change from dark color to light color, it is easy to draw a line there marking the western geographic limit of occurrence of the M. f. arthuri. This same line marks also the eastern margin of the geographic range of the subspecies Mustela frenata frenata, the subspecies next adjacent to the westward. On this line and for several miles to either side of it weasels show varying combinations of these three characters or an intermediate condition as regards one or more of the characters, or both. For example, from a locality in eastern Texas a weasel may have (1) a facial pattern exactly intermediate between that of the unicolored face of arthuri and that of the bicolored face of frenata, (2) the long tail of frenata and (3) the convex skull of arthuri. In the sum of its characters this specimen is exactly intermediate between typical arthuri and typical frenata. Another specimen from the same place may differ from the first specimen only in having the tail slightly shorter. The total "score" for the two specimens is, therefore, by a very slight margin in favor of arthuri. Let us suppose that we obtain a third specimen from the same place and that it has the face marked like that of arthuri but the tail fully as long, and the skull as lacking in dorsal convexity, as in frenata. Now the score is definitely for frenata. For convenience of handling, the population is referred to frenata, providing that the average of specimens from a nearby locality to the westward is not in favor of arthuri. In event the average of specimens from a locality next adjacent to the westward is in favor of M. f. arthuri, the total evidence from the two localities may be weighed together and appropriate decision as to subspecific status of weasels from the area is made according to what the average is for the area as a whole.
The three individual animals of an intermediate sort are ordinarily termed intergrades. This implies that their characters are the result of mixed parentage—perhaps a female of M. f. arthuri and a male of M. f. frenata but probably each parent itself was an intergrade and the offspring, of which we examined three, owe their characters to reproductive processes operating in obedience to Mendelian laws of inheritance.
The two kinds of animals, Mustela frenata arthuri and Mustela frenata frenata, are identified as subspecies because of the intergradation between them. If at this and all other places where the geographic ranges of arthuri and frenata met there was no crossbreeding (no intergrades), the two kinds would be treated as distinct species. Intergradation, and the lack of it, are accepted as the criteria of subspecies and species, respectively.
These criteria suffice for animals, in this instance weasels, which have a continuous geographic distribution. Some kinds of weasels are confined to islands, as for example the islands off the coast of Alaska and British Columbia. Because weasels are land animals, crossbreeding in nature between the weasels of two islands is, of course, impossible. A modified test (used in the study here reported upon) in deciding on specific versus subspecific status in these instances can be made as follows: On the adjacent mainland, ascertain the degree of difference between two subspecies whose geographic ranges meet (for example, M. e. richardsonii and M. e. alascensis). Next ascertain the degree of difference between the insular kind of animal and the kind on the mainland. If the degree of difference is greater when the insular kind is compared than when only the kinds of the mainland are compared, the insular kind is to be regarded as a species. If the degree of difference is no greater between the insular kind and the mainland kind than it is between the two adjacent mainland kinds, the insular kind is to be regarded as a subspecies. In short, for insular kinds, the criterion is degree of difference, with the limitation of geographic adjacency, rather than intergradation.
The geographic variation (subspecific characters) found could be spoken of as two kinds: First, there is the variation which is expressed in a general trend for a long distance, producing, in general, a cline of even slope; and second, that of inconstant trend in any one direction. In his "The Rabbits of North America" Nelson (1909:34-35) has commented on the latter type of variation as follows: "While studying series of specimens from all parts of the vast range occupied by the geographic races of such species as Sylvilagus floridanus and S. auduboni, I have been impressed with evidences of fluctuation of both external and skull characters. These fluctuations are somewhat wavelike in character and rise to central points of extreme development and then sink away to intermediate borders beyond which new waves rise. Where the waves of differentiation are pronounced they mark recognizable geographic races. Within the area covered by the larger or geographically broader waves of differentiation (recognized as of subspecific value), smaller waves of differentiation are included, which may represent local variations in intensity of characters of the subspecies, or these characters may diminish and the variation tend in other directions, sometimes even closely reproducing the characters of another subspecies occupying a distinct area." In Mustela frenata, much of the geographic variation at first inspection appears to be of this nature. Closer scrutiny, however, reveals that the repetition, at geographic intervals, of several features of color and structure are closely correlated with environmental features which are repeated only at these same places.
In Mustela erminea, much of the variation is of the first kind, namely, that which can be expressed as long clines of relatively even slope. As several authors have said, zoölogical classification based on this kind of variation is like dividing the spectrum and depends largely upon the standards set, for, theoretically, the possibilities of subdivision are unlimited. Actually, however, none of the clines has an even slope and the possibilities for subdivision therefore are limited. Also, when several features are used, instead of only one feature, the classification is more satisfactory even if the basis is more complex.
Some features of structure which provide subspecific characters are mentioned below.