Breeding experiments which show tliat hybridisation and mutation etc. 213 



rosette stage and come into bloom during the summer. Thus my Fg 

 plants all bloomed last summer; but many of the F^ plants grown 

 in the hot climate of St. Louis remained rosettes. 



The Fj plants showed a remarkable uniformity of type, as will be 

 explained later, but the F2 presented an equally great diversity. In 

 dealing with the F2 cultures, the greatest care was exercised in all stages, 

 to prevent unconscious selection of the seedlings. The seeds were sown 

 on January la'**., 1912, and the pricking out was not begun until about 

 two months later. In pricking out, every reasonable precaution was 

 taken to prevent unconscious selection, the process always beginning 

 at one side of the seed pan, and all the seedlings, both large and small, 

 being taken as they came. Wlien the numbers in a culture were net 

 excessive, all the seedlings were pricked out, the seed pan being kept 

 for further possible germinations. Wlaen such germinations occurred 

 the fresh seedlings were pricked out afterwards. In certain cases where 

 the number of seedlings in a pan was very large, some were left at one 

 side of the pan. The necessity for these precautions to prevent selection 

 on account of a possible differential germination rate or other cause, 

 will be clear later, when it is shown that in different cultures the pro- 

 portions of certain types were significantly different from each other, 

 i. e., they varied from famil}' to family. 



In most cases the germination was nearly or entirely completed 

 before the pricking out began. This was not always the case, however, 

 for sometimes germination continues for three months after the seeds 

 are sown. The rate of germination probably depends to a large extent 

 upon the degree of maturity of the seeds when harvested, but germi- 

 nations are often erratic and impossible to predict. I wish to emphasize 

 the fact that ev-ery precaution has been adopted to prevent any uncon- 

 scious selection, and I am assured that the results are as free from 

 any such source of error as they can be made. I am indebted to Mr. E. 

 J. Allard for his care in taking several of the photographs which illu- 

 strate this paper. 



B. The ancestors of the crosses. 



We will consider first the parents of the cross grandiflora x rubri- 

 calyx. My cultures of 0. grandiflora Solander were derived from seeds 

 obtained in 1909 from Prof. S. M. Tracy, who obtained them from 

 plants derived from seeds collected by him in the original locality at 

 Dixie Landing, Alabama. The species apparently survives only in this 



