ON ARTIFICIAL ATAVISM. 19 



of the wild plant is evidently of a compound character and consists of a 

 series of single color forms, including two elements of yellow, one con- 

 fined to the underlip, which remains when the yellow is elsewhere lost in the 

 corolla. But these yellow characters I have not included in my experiments, 

 though, of course, their behavior in my hybrids was often observed. To 

 determine the number of these units would necessitate a much longer study 

 of the cultures than I was able to give them. 



The red color consists essentially of two characters, a flesh-color with 

 lively red lips, and a white "or nearly white tube. These two characters are 

 separately represented by two distinct varieties called respectively "flesh- 

 colored" and "Delila." Both of them I found fairly constant from seed. In 

 the following account of my experiments I will confine myself to these two 

 elements and leave aside further analysis. 



The statement here given that these two characters are the principal com- 

 ponents of the original red color rests upon results gained by crossing this 

 type with the white variety, and cultivating the second generation, viz., the 

 children of the hybrids. I observed that Antirrhinum majus yielded a good 

 crop of seeds when fertilized with its own pollen. It is an easy operation to 

 castrate and fertilize them, and I always enclose the whole raceme in a bag 

 of parchment paper to exclude the visits of insects. These bags are very 

 effective for such experiments, and are impervious to the weather for two or 

 three weeks or more.* I never used seed wliich had not been fertilized by 

 myself. 



My crossing was made in the summer of 1896. Some specimens of the 

 white variety were castrated and pollinated with the normal red type. Next 

 year I had a great many dark red flowering hybrids, and fertilized none of 

 them with their own pollen. For various reasons they did not give enough 

 seed to cultivate the next or second generation in sufficient number of indi- 

 viduals. Nevertheless, the splitting up of the red color in its various com- 

 ponents was to be seen in the different lots, each obtained, from the seeds of 

 one self-fertilized plant. I give the figures for a lot of forty-nine specimens, 

 all children of one hybrid mother, in which the separation of the different 

 units was sharp enough to count the principal types without any difficulty. 

 These were four, viz., red and white, the colors of the grandparents, and 

 besides the two types named above as the components of the original dark-red 

 color. 



I counted 



Dark-red 51 % 



Flesh-colored 16% 



Delila 31% 



White 2% 



A great many of these plants were self-fertilized in parchment bags, and 

 of some of them the seeds were sown next year (1899). Of course, the seeds 

 were gathered and sown separately for each plant. 



I chose the progeny of a dark-red colored hybrid of 1898, which split up 

 next year in the same manner as did the former generation. I had in all 



* They were made by Mr. P. J. Schmitz in Dusseldorf (Germany), and are made of 

 the same material as the bags commonly used for protecting grapes against the stings 

 of wasps. 



