ON THE ORIGIN OF SPECIES. 493 



fertilization. The plant is purely feminine. Male or hermaphrodite 

 specimens I never saw, though I have often cultivated hundreds of 

 roundheads. On account of the broad leaves, and thick buds, Oeno- 

 thera lata was chosen as the systematic name. 



Encouraged by these results, I continued my investigation, partly, 

 during the same year, by a closer study of the locality where the seed 

 was collected, partly by sowing experiments on a large scale in the 

 spring of the year following. The former made me acquainted with 

 two new types, which had remained unobserved in 1886, but which, as 

 rosettes of root-leaves, must have been present at the time, since Oeno- 

 thera Lamarclciana is, in that locality, biennial, with hardly any ex- 

 ception. The one was glabrous, more delicate and more graceful, but 

 as robust as the common form, the other had so short a style that the 

 stigma, instead of protruding far above the stamens, was situated at 

 the base of the flower. Both forms were formerly absolutely unknown, 

 and, later on, in the sowing-experiments, proved to be as constant and 

 true to seed as the mother species. That they originated in the local- 

 ity where they were found may be considered as certain, but how this 

 happened could not be investigated. 



The seed, sown in the spring of the following year, again yielded 

 two new forms. The one was a dwarf form, such a one as occurs from 

 time to time among all kinds of culture plants but a few decimeters 

 high, whereas the mother species attains a height of l 1 /* to 2 meters 

 and more. The other was a form with shiny leaves, about half the 

 size of Oenothera Lamarclciana, narrow, dark green, and very graceful. 

 Both were quite fertile and produced a large quantity of seed. 



Dwarf forms are ordinarily described along with the species to 

 which they belong, as varietas nana or nanella, and my dwarf forms 

 agree with them in every respect. They offer a good contrast with the 

 other types, which cannot be termed varieties in the ordinary sense. 

 For, in the first place, they deviate from the parent species not in a 

 single character, but in all, and, in the second, they do not have their 

 parallel in other genera. For repetition, such as appearance of white 

 flowers, glabrous leaves, thornless stems and fruits, unbranched stems, 

 variegated leaves, double flowers, etc., is one of the most common 

 characteristics of true varieties. 



Later the dwarfs proved to be constant to seed. Not so the shiny 

 variety. Though I did not sow the latter each year, I did it fre- 

 quently; its characters reappeared as a rule in but about one third of 

 the individuals. 



By sowing I obtained in 1888 nearly 15,000 plants, among which 

 there were five dwarfs and five latas, that is to say, of each about 1 on 

 every 3,000. In later years, when I became familiar with the most 

 favorable methods of treatment, the percentage increased considerably, 



