To study geographic variation in color a method was devised as follows: A single skin (KU 42407, from 1½ miles east of Buckhorn in Weston County, Wyoming) was selected as a representative of the paler mice and arbitrarily given the number 2. A single skin (KU 17491, from 3 miles east of Moran in Teton County, Wyoming) was selected as a representative of the darker mice from the western part of Wyoming and arbitrarily given the number 4. These mice were selected so that they were respectively paler and darker than the estimated average of the total variation within the populations to be studied, but the two mice were not at the extremes of paleness and darkness. Comparisons were based on visual inspection of the dorsal pelage as a whole. Skins were compared with these two mice and given whole numbers from one to five. If paler than the standard for 2, the skin was numbered one; if not distinguishably paler or darker, it was given the number two; if intermediate in color to the standards for 2 and 4 and not definitely more nearly referable to one than the other, it was given the number three; if it resembled the standard for 4, it was numbered four; and if darker, it was given the number five. In this manner skins from a given locality could be evaluated one by one and the results plotted, averaged, and treated statistically. On Figure 1 the average values for color of 32 series are mapped to show the geographic variation of color. The following series of adults are the basis for Figure 1 (abbreviations for collections other than at the University of Kansas are included in parentheses): Each locality is followed by the month (or months) of capture, the number of specimens, and the average value for color.
Montana: Glacier County, August, 6, 1.8; Hill and Chouteau counties combined (Mich), July, 24, 1.5; Malta, Philips County, August, 14, 1.5; Sheridan County, August, 6, 1.5; Fergus County (USBS), August, 5, 2.4; Ravalli County (KU and USBS), August, 12, 2.8; Silver Bow County, August, 7, 3.0; Sweet Grass County (Mich), June and July, 7, 2.7; Park County, August, 10, 2.6. Idaho: Pocatello and vicinity, November and December, 5, 3.4. Wyoming: Park County, August, 6, 2.8; Sheridan County, September, 9, 1.2; Johnson County, August, 12, 1.5; Campbell and Crook counties, July, 11, 1.4; Weston County, July, 7, 1.6; Teton County, September, 8, 3.4; Teton County (Mich), June, 17, 3.1; Afton and vicinity, Lincoln County, July, 10, 4.2; Sage, Lincoln County, July, 5, 5.0. South Dakota: Pennington County (Chi), June, 14, 2.1; Pennington County (Mich), December and January, 8, 1.1; Walworth County, July, 4, 3.7; Buffalo County, July, 6, 3.2. Colorado: Loveland and vicinity, Larimer County (KU and USBS), July, 13, 2.8; Boulder County (Chi), September, 34, 2.6; Park County (Denv), March, 8, 1.9; Colorado Springs (ERW), March, April, and May, 5, 2.8; Saguache County (USBS), August, 46, 3.0; Conejos County, June, 4, 3.0; Wray, Yuma County (USBS), 3, 4.7. Nebraska: Dundy County, August and November, 14, 4.6. New Mexico: Colfax County, June, 8, 3.2. Variation in color is discussed in the accounts of the subspecies concerned.
Figure 1. Geographic variation in color in Microtus pennsylvanicus in the Rocky Mountains. Paler colors are represented by smaller numbers. Numbers are derived from the series of specimens listed in the text by the method described there. The subspecies that occur in the region studied are as follows:
| a. M. p. pullatus | e. M. p. modestus |
| b. M. p. insperatus | f. M. p. aztecus |
| c. M. p. uligocola | g. M. p. drummondi |
| d. M. p. finitus | h. M. p. pennsylvanicus |
For each of the series listed in Table 1 all adult mice having skulls that measured more than 24.0 mm. in condylobasilar length were studied. Total length, length of tail, and length of hind foot were taken from the collector's field labels. The measurements of the skulls listed below were taken by means of dial calipers reading to one-tenth of a millimeter, and in the same fashion as described previously (Anderson, 1954:492). Measurements of specimens in each series were averaged (the arithmetic means were computed). If the averages differed noticeably the significance of the difference was tested statistically. Averages referred to in the text as significantly different differ by as much as, or more than, the sum of two times the standard error of each of the two averages. Linear measurements are in millimeters; color values are in the arbitrary units described in a preceding paragraph. Measurements taken of the skulls are: condylobasilar length, zygomatic breadth, interorbital breadth, lambdoidal breadth, prelambdoidal breadth, depth of braincase, and alveolar length of upper molar tooth-row.
Secondary sexual variation was not detected in the material studied. Variation with age is important to the taxonomist even among specimens designated as "adults", because growth and changes in various proportions continue throughout the life of the mice. The possibility that differences detected in the statistical treatment or observed directly could be the result of differences in average age within the samples of "adults" was considered in each case.
In order to study certain variations, the following "method of pairs" was used. Skulls of two series to be compared were matched in pairs so that they corresponded in size and ontogenetic stage of development. Then the two skulls of each pair were examined for differences in each of the following features: size of circle inscribed by the upper incisor teeth, width of nasal bones relative to their length, curvature of the zygomatic arch, elongation of the braincase relative to its width when viewed from the dorsal aspect, degree of indentation in the anterior edge of the zygomatic arch near the rostrum, degree of depression of the nasal bones when viewed from the side, width in the vertical plane of the zygomatic arch at the suture between the maxillary and jugal bones, length relative to width of the prominent fenestra in the posterodorsal part of the squamosal bone, size of the meatus of the auditory canal, distance between the internal margin of an occipital condyle at its posteriormost point and the tip of the paraoccipital process of the same side of the skull, size of the foramen magnum, vertical height of the supraoccipital bone from the dorsalmost point on the margin of the foramen magnum to the midpoint of the lambdoidal crest, constriction posteriorly or narrowness of the incisive foramen relative to its length, distance across the premaxillary bone from the anteriormost point of the incisive foramen to the posteriormost point of the margin of the alveolus of the upper incisor, area of the maxillary septum (Howell 1926:112, or "zygomatic plate" of Ellerman 1941:1), acuminateness of the anterior border of the palatine opening (internal nares), size of auditory bullae, size of foramen ovale, acuteness of the angle between the basioccipital and basisphenoidal bones at the suture between them (degree to which the area of the suture is raised between the bullae when viewed from the ventral aspect), width of first upper molar tooth, least distance between alveoli of first upper molars. Any differential feature present in more than 75 per cent of the pairs of animals is reported in the discussion of the subspecies concerned. The significance of each difference reported was calculated by the Chi-square test and the confidence limit is given in each case. The probability used in the Chi-square formula is one-half of the percentage of all pairs compared in which the skulls were different in regard to the character being considered. For example, in 68 per cent of the total number of pairs of skulls compared in this study a difference in the size of the auditory bullae was noted. Therefore the probability that a specified skull of a pair will have larger bullae than the other skull was taken as 34 per cent. A different probability for each feature compared was derived in like manner.
This study is concerned primarily with mice from Wyoming and Colorado; I realize, however, that the physiographic and ecological conditions important to the distribution and subspeciation of Microtus pennsylvanicus do not correspond to political boundaries. Geographic variation within these two states can be seen in proper perspective only when related to the neighboring areas and to previous studies. I have attempted to do this in the accounts of the subspecies.
Approximately five months in the field in Wyoming and Colorado in the summers of 1950, 1951, 1952, and 1953 gave me a familiarity with the region that has helped to clarify the pattern of distribution. My study was based, in addition, on 762 specimens that are listed under "specimens examined" in the accounts of subspecies, and on comparative material from other states. Most of these specimens are skins with skulls but some are skins only and others are skulls only. Some localities are represented by too few adult individuals to permit significant comparisons. Owing to damaged skulls, certain measurements of some specimens were omitted from the calculations. If it seemed that the damaged skull was exceptionally large or small or a deviant in any other regard it was not used, in order not to bias the computed averages, which might be used in comparing proportions of the skulls. In the lists of specimens examined, localities that are omitted from Figure 2 because overlapping or undue crowding of the symbols would have resulted are italicized.