32 
Clearwater, Feldhamer, Morgan, and Chapman 
of Natural History were used in the analysis. Females were excluded 
because of small, unequal sample sizes among groups. Adults only were 
examined and distinguished from juveniles according to tooth wear 
(Grau et al. 1970), closure of parietal sutures, and month of capture. 
Raccoon skulls were grouped according to capture location in 
Maryland into one of three physiographic regions (see Paradiso 1969): 
(1) western, which included the Appalachian Plateau and Ridge and 
Valley regions (n = 21); (2) central, which included the Piedmont region 
(n = 23); and (3) eastern, which included the Coastal Plain region (n = 
19). Skulls from the two initial groups represented P. 1. lot or. The third 
group represented P. 1. maritimus, based on all captures being from the 
type locality, Blackwater National Wildlife Refuge, Dorchester County, 
adjacent Fishing Bay, or similar marsh habitats in five other counties on 
Maryland’s eastern shore. Comparisons also were made with skulls of 
P. l.fuscipes (n = 16) from Texas and Mexico. 
Differences among groups were determined by analysis of variance 
of 12 skull measurements (see Table 2), with Duncan’s Multiple Range 
test, and by stepwise discriminant function analysis, after elimination of 
correlated variables (r > 0.70). The selection criterion was the maximum 
Mahalanobis distance. Analyses were conducted using programs of 
SPSS (Nie et al. 1975). 
Table 1. Actual and predicted group membership for male raccoon skulls from 
western and central Maryland ( P . /. lotor), eastern shore of Maryland 
( P . /. maritimus), and Texas ( P . l.fuscipes). 
Actual group 
(N) 
W. Md. 
Predicted group 
C. Md. E. Md. 
Texas 
W. Md. 
(21) 
13 
6 
2 
0 
C. Md. 
(23) 
5 
15 
3 
0 
E. Md. 
(19) 
1 
5 
12 
1 
Texas 
(16) 
2 
0 
4 
10 
RESULTS AND DISCUSSION 
In the discriminant function analysis of the three Maryland groups 
and P. 1. fuscipes (Fig. 1), width of braincase had the highest standardized 
coefficient (-0.849) in the first function, followed by basilar length 
(0.830). The first function explained 69.0% of the variation among 
groups. In the second function, width of rostrum (0.505) and width of 
