VARIATION 33 



ates around the number 21, while the dry pasture popu- 

 lation B, characterized by blossoms which were in 

 general noticeably smaller, fluctuates around the num- 

 ber 34. The habitats of the two lots were so near to- 

 gether, however, that there was probably a considerable 

 intermixture of the two types, as shown by the tendency 

 of each polygon to produce a second mode. Thus the A 

 polygon shows that there is an increasing tendency 

 or variability in the twenty-one floret type toward the 

 thirty-four floret type, due probably in this particular 

 instance to invasion resulting from the proximity of 

 the B colony. 



8. GRADUATED AND INTEGRAL. VARIATIONS 



It is comparatively simple to treat statistically 

 integral variations, illustrations of which have been 

 given in the case of beech-leaf ribs, starfish rays, and 

 daisy florets, all of which are characters that can be 

 readily counted. In the same way any measurable 

 character, such as the size of snail shells, may fall into 

 easily limited groups, as, for example, 10 to 11 mm., 

 11 to 12 mm., 12 to 13 mm., etc. It is somewhat 

 more difficult to classify variations when color or 

 pattern is the character in question, since it then be- 

 comes necessary to define certain arbitrary limits for 

 each class of the series within which to group the indi- 

 vidual variants. 



Tower, in his famous researches on potato-beetles, 

 encountered variations in the pigmentation of the pro- 

 notum all the way from entire absence of color to com- 



