EXPERIMENTAL EVOLUTION 527 



9 inches the frequency was zero. When the ears were 

 arranged in piles according to their length, the tops of 

 the piles indicated the curve of frequency (Fig. 393). 



The curve of frequency indicates the quantitative dis- 

 tribution of any character or quality when its occurrence 

 is dependent largely upon chance. This is strikingly 

 illustrated by the grouping of bean seeds rolled down a 

 smooth inclined plane, and collected in receptacles at the 

 bottom (Fig. 394). The seeds are started rolling midway 

 between the edges of the plane; the chances are about 

 equal for some of the seeds to fall into the outside compart- 

 ments, but the odds are vastly in favor of their landing at 

 or near the center. Thus they group themselves so that 

 the tops of the piles form a curve of chance variation. 

 When the result is influenced in one direction more than 

 in another the crest of the curve will be nearer one extreme 

 than the other, and the curve is to that extent skew. The 

 curve of bean seeds in Fig. 394 is slightly skew toward 

 the right-hand extreme. Suggest one or more reasons why. 



450. Fluctuating Variation and Inheritance. — When 

 the ancestry is not mixed or hybrid the curve of frequency 

 of any character in one generation ordinarily tends to 

 recur in successive generations of descendants,^ providing 

 the environment remains essentially the same. 



451. Discontinuous Variation. — Long before Darwin, 

 students of plants and animals had observed a different 

 kind of variation than continuous — one which was not 

 quantitative but qualitative, resulting in the expression of 

 new characters, or of a new curve oj frequency; that is, in 

 fluctuation about a new mean. Plants from some of the 



^ The behavior of hybrid descendants is a special case described in 

 ::hapter XXXVII. 



