BY THE SELECTION OF SOMATIC VARIATIONS. 63 



are shown to be direct products of the plant. The important point is 

 that while the pigments are themselves of highly complex chemical 

 substances, the changes which produce marked differences in color are 

 very slight. 



It is also significant that marked differences may exist between the 

 parts of a single flower. The simplest of color patterns in Antirrhinum 

 treated by Wheldale (1914, p. 110) illustrates this condition. Flowers 

 of the yellow-flowered variety possess a pale yellow pigment in the tube 

 of the corolla, a deep yellow pigment in the lips, and a still deeper 

 patch on the palate. In attempting to harmonize these varied con- 

 ditions with a Mendelian factorial analysis, the power to produce these 

 pigments in spatial relation is assigned to a factor Y. In the "ivory" 

 flower a pale pigment is found in the tube and Ups and a yellow pigment 

 only in the epidermis of the palate. Power to produce pale pigment 

 quite generally in the flower, and to inhibit the formation of deep yellow 

 every^^here but in the epidermis over the palate is assigned to a single 

 factor, /. The conditions of color distribution are in themselves color 

 indicators that different processes occur in different parts of the same 

 flower, giving different kinds of substances in different amounts and 

 with different distribution. 



Studies of oxydase reactions in different tissues indicate that inter- 

 cellular relations are of much importance in determining the distri- 

 bution of pigments. By means of micro-chemical tests Keeble and 

 Armstrong (1912) obtained evidence that the distribution of oxydase in 

 various types of Primula is closel}^ correlated with the development of 

 anthocyanin. In general they find the oxydase most abundant in 

 epidermal and in vascular tissue. The extent of ox3'dase distribution 

 differs much in different varieties. They make the highly interesting 

 observation that action of chloroplasts seems to act as an inhibitor 

 of oxidase formation or of the production of chromogen. Varied types 

 of color pattern in flowers, and even anthoc^-anin development in 

 stems and leaves, is, according to these investigations, closely related 

 to distribution of oxydases and chromogens. 



It is of further interest to note that patterns resulting from such 

 qualitative and quantitative reactions which depend in considerable 

 degree on flow of substances in a plant may be quite uniform not only 

 in the flowers of a plant, but among the various members of its seed 

 progeny. Such phenomena have led to the assumption that patterns 

 are represented as such in germ-cells by hereditar}- units. A^Tien, how- 

 ever, hybridization occurs between varieties having even the simplest 

 of patterns, the F2 generation more often than otherwise presents a most 

 remarkable range of types. Usually, this sort of diversity results in 

 crosses betw^een varieties, especially when the color patterns of flowers 

 and leaves are concerned and gives results that could not be predicted 

 with any degree of accuracy. 



