c3"> Fig. 3. Number of subcaudals in females and males of T. gracilis.

Ventrals minus subcaudals

The value for number of ventrals minus the number of subcaudals varies from 80 to 96 (mean, 88.8 ± 0.39) for 79 females, and from 67 to 83 (mean, 73.8 ± 0.28) for 109 males.

Additional scale characteristics

Of the 246 specimens examined, four (1.6%) have the rostral in contact with the left prefrontal, and hence the internasals are separated from one another. One snake has approximately one-fourth (anteriorly) of the suture between the parietals fused. One specimen has the posterior edge of the frontal fused with the left parietal. One specimen has the parietal shield on the right side abbreviated posteriorly. One specimen has both prefrontals fused with the frontal, although an incomplete suture is present.

Maxillary characteristics (Figs. 4 and 5)

Maxillary teeth anterior to fangs 10-14 (mean, 12.1; N = 20); tooth-socket number 6.5-8.5 (mean, 7.2; N = 20) directly opposite distal tip of suborbital process; lateral anterior foramina 1-3 (mean, 1.8; N = 20); tooth number 6-9 (mean, 6.9; N = 12) directly below center of posterior foramen (if only two foramina present). Three specimens lack a diastema and 17 specimens have a diastema. The anterior and posterior edges of the suborbital process are parallel in two specimens and not parallel in 14; the process is broken in four specimens. The lengths (anterior to posterior) of the lateral anterior foramina (if only two are present) are approximately equal in six specimens, and in six the anteriormost one is the longest; six specimens have one foramen and two specimens have three foramina. All specimens examined have two fangs, with the posterior edge of the lateral flange forming an obtuse angle.

Fig. 4. Lateral view of right maxilla of T. gracilis (KU 83484), × 39. Actual length of bone is 2.44 mm. A. Detail of bone. B. Outline sketch of bone to illustrate characteristics examined. Lateral anterior foramina (LF); fangs (F); "prediastemal" maxillary teeth (MT); lateral flange (PLF); perpendicular lines, A-B and C-D, are used to determine which tooth-socket occurs directly below the center of the posterior foramen (the specimen illustrated is atypical because there are three foramina present).

Fig. 5. Ventral view of right maxilla of T. gracilis (KU 83484), × 39. Actual length of bone is 2.44 mm. A. Detail of bone. B. Outline sketch of bone to illustrate characteristics examined. Lateral flange (PLF) forms an obtuse angle (ADC) with main axis of bone (AB); a reference line (2), perpendicular to the main axis of the bone, is used to locate the tooth-socket directly opposite the tip of the suborbital process (SOP); the anterior and posterior edges of the suborbital process are not parallel at any point (reference lines 1 and 3).

Discussion and Conclusions

Generic characteristics of Tantilla are as follows: size small, body slender, head not noticeably distinct from body; preoculars one; postoculars one or two; nasals two; loreal absent; smooth dorsal scales in 15 rows; anal plate usually divided but sometimes single; subcaudals paired; usually two enlarged, posterior, grooved fangs on the maxilla separated by a small diastema from the other maxillary teeth (Baird and Girard, 1853:131; Cope, 1900:1110; Dunn, 1928:24; Blanchard, 1938:369; Schmidt and Davis, 1941:268; Wright and Wright, 1957:722; Conant, 1958:180). Of these characteristics, only the fol

lowing were invariable in our sample: size small, body slender, head not noticeably distinct from body; postoculars one or two; smooth dorsal scales; posterior grooved fangs on maxilla. The dorsal scales, anal, and subcaudals were "correct" 92 per cent of the time; the preoculars and nasals were "correct" 97 per cent of the time.

Tantilla reportedly has two nasals (Baird and Girard, 1853:131; Cope, 1900:1110; Wright and Wright, 1957:722). Tantilla gracilis usually has a single nasal that is divided below the naris; variation in this characteristic is discussed above.

In addition to the generic characteristics, specific characteristics for T. gracilis are as follows: supralabials 5-8 (usually 6); supralabials 3 and 4 entering orbit; infralabials 5-7 (usually 6); temporals 1 + 1; ventrals 115-138 in females, 106-132 in males; subcaudals 33-53 in females, 40-57 in males; tail length 13-27 per cent total length in females, 15-30 per cent in males (Baird and Girard, 1853:132; Cope, 1900:1111-12; Force, 1935:653-54; Taylor, 1936:337-38; Blanchard, 1938:371-72; Kirn, Burger, and Smith, 1949:240-49). Excepting the number of temporals and the supralabials entering the orbit, the characteristics of the specimens in our sample are within the ranges of variation mentioned above. Of the 10 maxillary characteristics studied, no variation was observed in number of fangs or angle of lateral flange.

Because our data have some bearing on the problem of geographic variation in T. gracilis and the recognition of subspecies, we comment briefly on the status of subspecies in T. gracilis. Kirn, Burger and Smith (1949) proposed the recognition of two subspecies of Tantilla gracilis (T. g. gracilis Baird and Girard and T. g. hallowelli Cope). These subspecies were diagnosed on the

basis of differences in ventrals, subcaudals, and the ratio of tail length to total length; sexual dimorphism in each characteristic was considered. We do not recognize these subspecies for the reasons given below.

The data of Kirn et al. (1949) appear to be presented inaccurately in part. Both of us independently recalculated the mean given for each characteristic for each subspecies, using the data in figures 2-4 of Kirn et al. (1949:242, 244-245). Of their 12 means presented (table 1, p. 247) only two agree with our recalculated means, although the means calculated by each of us independently are in complete agreement (Table 5). Also, we independently calculated the percentages of specimens of each "subspecies" that are included in the ranges of variation given in their diagnoses (Table 5); again, our independent calculations are in complete agreement. In our opinion the differences between the populations for the characteristics analyzed do not