No. 517] A BIMODAL VARIATION POLYGON 



25 



Certain "teratological" eases are not represented by 

 figures, but are described merely. 



We may now turn to the important question of the fre- 

 quency of the several morphological types in the Cold 

 Spring Harbor collection made in the spring of 1909. 



In classifying these inflorescences all which were in- 

 jured in any way so that the number of primary divisions 

 of the leaf could not be made out with certainty were dis- 

 carded. Especial care was used in cases in which a leaf 

 was apparently divided into two leaflets, since such often 

 results from the breaking off of one of the lateral leaflets 

 of a temately compound leaf. In some cases inflores- 

 cences were included in which the lamina? of the primary 

 leaflets were not intact, so that the data can not be trusted 

 for the frequency of secondary divisions. In counting, 

 only completely divided laminae were considered as 

 leaflets. 



Petiolate leaves were not infrequently found, but no 

 special record was kept of them. 



The data are set forth in minutely analyzed form in 

 Table II-III. The light-faced numbers separated by 

 dashes represent the number of primary leaflets into 

 which the leaves are divided. The black-faced numbers 

 show the number of flower-buds— terminal and axillary— 

 which might normally have developed, and the number 

 immediately following in parentheses shows the number 

 actually developing. The / (= frequency) column gives 

 the number of occurrences of each type. 



Grouping the data according to the number of leaves 

 per inflorescence, we find : 



Two leaves, in 544 cases. 



685 altogether. 



Here is an example of what De Vries 19 has termed a 

 half-Galton curve. Whether it would be possible to in- 

 crease the number of leaves by selection as De Vries was 



