294 KELLOGG AND BELL 



in the 502 individuals taken from the same spot at the same time 

 are as follows : 22 individuals have 16 lines ; 63 individuals 

 have 15 lines; 172 individuals have 14 lines; 146 individuals 

 have 13 lines ; 55 individuals have 12 lines ; 28 individuals have 

 ii lines; 21 individuals have the lines too much interrupted to 

 be counted ; 8 individuals have the lines too much netted, i. e., 

 connected with each other, to be counted ; 7 individuals have 

 the lines too much interrupted and netted to be counted. 



As indicated in the description of three classes in the fore- 

 going table, the lines show variations in degree of distinctness, 

 tendency toward coalescence, and tendency toward interruptions 

 or breaks in continuity. In some instances the lines are distinct 

 and nearly parallel. In most cases a tendency toward coales- 

 cence of the lines is manifest, a wide range in the specific char- 

 acter of this coalescence being present. Two or more lines 

 may coalesce in the middle or at one or both ends. In some 

 cases a network is thus formed and it becomes difficult or really 

 impossible to count the lines. In some cases the lines are wavy, 

 and along the posterior margin the lines are frequently con- 

 nected by irregular markings resembling those on the wings. 

 Interruption or obliteration of parts of the lines also occurs and 

 in a few cases certain lines are farther apart than in others. 

 In some specimens more or less regular, elongated spots occur, 

 one on each side near the median line half way between anterior 

 and posterior margins. They appear at first sight to be simply 

 interruptions of the transverse lines, but in some instances the 

 latter may be traced through the spots when the specimens are 

 held in proper light. 



Altogether the variation appears to follow pretty fairly the 

 law of error, the modal number of lines being 14 with 13 nearly 

 as frequent. The pattern is almost certainly distinctly protective 

 in character and maintained in its present condition by the action 

 of natural selection. The residue of variation, here tabulated, 

 is composed of characters evidently not of selective value. It 

 would indeed seem reasonable to expect that a variation of but 

 two or three lines on either side of the mode would not seriously 

 invalidate the protecting (water mimicry) value of the pattern. 

 Thus in this case where the pattern is obviously of direct pro- 



