The Mutations in the Lamarckiana-F amily . 231 



gation sliows. If a fruit is pulled off without great care 

 the stem is usually broken in the process, an event which 

 has more than once caused me considerable annoyance at 

 harvest time. 



In all its other characters O. rubrinervis is a very 

 healthy plant — quite as strong as O. Lamarckiana at any 

 rate. It is the only one of my new species which is not 

 inferior to the parent type in richness in pollen and seed. 

 Apart from its brittleness it seems to be fully qualified 

 for the struggle for existence. I have not however yet 

 organized any experiments to determine this point. 



0. rubrinervis appeared in the main line of descent, 

 as the table on page 224 shows, in the third, fourth, 

 fifth and sixth generations. There were 32 examples 

 of it altogether. In the other families it was also ob- 

 served from time to time; and appeared as early as 1889 

 in the laez'ifolia-immly. In 1897 only three appeared in 

 the main line of the Lamarckiana-i2im\\y, whilst 10 ap- 

 peared in the branch lines of descent. Their ancestors 

 for at least five generations back were all 0. Lamarckiana, 

 or at least not of the rubrinervis type. 



O. rubrinervis appears each time without visible prep- 

 aration : and what strikes one most is the absolute con- 

 stancy of the characters although these were quite distinct 

 from those of its ancestors. 



When once I recognized a plant in its rosette stage 

 as being a rttbrinervis I could predict that it would have 

 a fragile and brittle stem and red calyx and fruit. This 

 constancy in character is a feature of all my new ele- 

 mentary species and is even more striking in the case of 

 O. lata than in rubrinervis. 



I first started experiments on the constancy of the 

 new rubrinervis in 1896 and 1897. In 1895 I covered 



