18 nORTICULTUKAL SOCIETY OF NEW YORK. 



In the case of simple flower-colors which are not built up of different 

 units the number of the color varieties is of course a limited one, and is 

 even often reduced to the existence of a white flowered form. But if the 

 colors are compound, and consist of two or more units, we ordinarily find a 

 greater array of differently colored types, in so much more as the degree of 

 composition of the original colors is greater. Besides the white variety, many 

 blue flowers have a red form ; so have many dark-red species a white and flesh- 

 colored form (Varieties carnea), etc. Some very beautiful flowers have a 

 darker color at the base of their petals, and such species often exhibit a variety 

 in which the so-called "heart" is wanting. Other flowers are mottled, and 

 have also spotless forms, etc. In all such cases the compound nature of the 

 original color-type seems to offer itself as the most plausible conception and as 

 a basis for further experiments. 



If now we go through the lists of color varieties named in the commer- 

 cial seed catalogues or in any descriptive horticultural work, we ordinarily 

 find the type of the original wild species and a white variety as the two 

 extremes of the series. Between these two we find a larger or smaller number 

 of intermediate color forms, and we may assume that these, or some of them, 

 are to be regarded as the elementary parts into which the original mixture may 

 be split up. And if we apply this reasoning to these intermediate forms, we 

 may come to two well-defined inferences, which it will be possible to test by 

 experiment. The first of these is, that by crossing the original with the 

 white variety the different intermediate types may be obtained. The second 

 inference is the reverse of the first, and assumes that by crossing the inter- 

 mediate types or at least by crossing the principal ones among them we could 

 return to the original color of the wild species. 



To test this argument I have chosen the common snapdragon {Antir- 

 rhinum niajus). But before giving the detailed description of my experi- 

 ments I must point out that Mendel, in his celebrated memoir on hybrids, 

 has briefly discussed the possibility of the compound nature of the colors of 

 most of our flowers, and has clearly formulated his conviction that by this 

 supposition it would be possible to explain their most curious behavior in the 

 few crossing experiments he had made with them. In fact, he says in the 

 introduction to his paper, "Experience of artificial fertilization, such as is 

 effected with ornamental plants in order to obtain new variations in color, 

 has led to the experiments which will here be discussed."* 



The experiments which I am now going to describe will give an entire 

 confirmation of Mendel's predictions.** 



The common snapdragon, Antirrhinuni majus, is cultivated in our gardens 

 under many beautifully colored varieties. The wild species or original form 

 is of a dark red color, but there are also a white variety and many inter- 

 mediate types. These varieties are constant from seed and remain pure as 

 long as occasional crosses with other forms are excluded. The dark-red color 



* A translation of Mendel's paper was published bj' the Royal Horticultural Society 

 in their Journal, vol. 26, 1901, and is to be found in Bateson's work on Mendel's princi- 

 ples of heredity, 1902, pp. 40 — 95. 



** For a fuller statement of my experiments with Antirrhinum and for the descrip- 

 tion of similar investigations with other species, I must refer the reader to the second 

 volume of my "Mutationstheorie" (Leipzig, Veit & Co.), and especially to the chapter 

 on "Die Zerlegung der Bluthenfarbe," pp. 194 — 206. 1902. 



