Two Lata-Families. 287 



in 1889. Its seeds, some of which were sown in 1890 

 and some in 1894, gave rise for the most part to the 

 same form. The cuhure of 1890 consisted partly of 

 annual plants (10 of which w^ere scintillans) and partly 

 of biennial ones (comprising 26 scmtillans) , the former 

 flowered well but too late; the others were frozen in the 

 winter, so that I got no seed from this experiment. The 

 1894 crop w^as entirely biennial and contained 11 plants 

 of scintillans which set seed in 1895. 



Let us now return to the main line of descent of our 

 family. On the 18th of April 1889 I sowed the seed 

 which had been harvested, on a bed (about 3 square 

 meters in size) which was fairly thickly covered with 

 plants towards the end of May when the latas could easily 

 be distinguished from the Lainarkiauas. Most of the 

 latter were removed. Towards the end of July, 12 annual 

 specimens of lata flowered : the rest of the latas, which 

 had either developed stems too late or still remained in 

 the rosette stage, were weeded out. Of those that flow- 

 ered eleven set seed, and were harvested together. 



Part of the seed was sown in the following year : 

 a smaller sample was kept till 1894, in which year they 

 gave rise to 340 seedlings of which 52 were lata : these 

 were however not cultivated further. 



I sowed the seed to produce the fourth generation 

 on the 5th of May 1890; as before, on a bed of about 

 3 square meters. At the beginning of Jnly there were 

 on this bed 79 specimens of lata and many of La- 

 marckiana. The former were partly annual, partly bi- 

 ennial. The annual plants did not flower before the 

 middle of September and only six latas set seed which 

 ripened very late and could not be harvested before De- 

 cember. 



