REVISION OF THE KING SNAKES. 197 



County), Knoxville, and Nashville; Tennessee; Bethel, Vermont; Lou- 

 doun County, and Staunton (Augusta County), Virginia; Q]iQ?Lt River, 

 ^est Virginia; Beaver Lake (Dodge County), Cedarburg, Cedar 

 Lake (Washington County), Madison, Mauston, Milwaukee, Newport 

 (Door County), and Waukesha, in Wisco7isin; Aylmer (Quebec), 

 Hull (Ottawa), near Ottawa, Point Pelee (Ontario), Spence's Lake 

 (Bracebfidge, Ontario), and nea,r Toronto, in Canada. 



Published records, that appear to be reliable, for other localities, 

 are as follows: Huntington, Connecticut, (Linsley, 1843, 43); Nor- 

 way, Maine, (Verrill, 1863, 197); Springfield, Massachusetts, (Allen, 

 1868, 180); near Waveland, Montgomery County, (Butler, 1887, 

 148), and Greencastle (McLain, 1899, 3), Indiana; London, Ohio 

 (Morse, 1901, 127), Hamilton County, Akron, and Nelsonville, Ohio 

 (Morse, 1904, 130); Alma, Michigan (Euthven, 1912, 113); Antrim, 

 Montcalm, Kent, Ottawa, and Barry Counties, Michigan, (Clark 

 1905, 110); Eocldand (Wallace, 1904, rl40), Rossie, St. Lawrence 

 County, New York (Hough, 1852, 23), and Cold Spring Harbor, 

 Yaphank, Orient, Greenport, and Southold, on Long Island, New 

 York (Engelhardt, 1915, 4); Stony Run, Cecil County, Maryland 

 (Fowler, 1915, 40); St. Clair township (Allegheny County), Pennsyl- 

 vania (Atkinson, 1901, 150); Shower Hill, New Jersey (Fowler, 1907, 

 181); Sunburst, Haywood County (Brimley, 1915, 202), and Chest- 

 nut Knob Mountain, North Carolina (Dunn, 1917, 630). 



Variation. — In spite of the apparent constancy and distinctness of 

 triangulum in the northern tier of States, the form as a whole is one 

 of the most variable. The diagram showing geographic differences 

 in number of ventrals (fig. 56) brings out some notable variations with 

 locality. The dependability of the figures varies somewhat with the 

 number of specimens upon which they are based. It is evident, how- 

 ever, that in the northern and western portion of its range the ventrals 

 average higher, more than 200, and the extremes are higher than for 

 the central and northeastern parts. The few specimens from the 

 southeastern States, except in the mountains, show the high num- 

 bers of the northwest. From New England to Virginia and in the 

 Alleghenies the averages and the extremes are low, and distinctly 

 lower in New Jersey and Delaware, and near the District of Columbia. 



This variation in ventrals should be compared with tlie table of 

 scale formulae. It will be noted that in the states close to the range 

 of syspila the scale rows average about the same as for that form, 

 that is, a formula lower than 21-19 is very common in males and a 

 little less so in females. In New York and Pennsylvania the only 

 change is a slightly lower average, but in New England, the most dis- 

 tant region, only one male is recorded with a formula of 21-19, more 

 than half the females have a lower formula than this, and the smallest 

 proportion of specimens reaches 23 rows. This compares closely with 

 the southern Alleghenies. The formulae for New Jersey and the 



