270 The Pedigree Families. 



carried out in the field. Of such families I possess three, 

 as I have already stated. The Laniarckiana-i2im\\y arose 

 from rosettes, the /a/a-family from seeds gathered in 

 1886, and the laevifolia-i^mxXy, with which this section 

 deals, from the small group of that species that was 

 found in the field (1887). The hereditary phenomena 

 which they exhibited afforded a clearer insight into these 

 processes than could be derived from the observations in 

 the original locality. 



I propose to ofTer now some remarks on the incom- 

 pleteness of field-observations and on some difficulties 

 of experimenting in a garden. If every mutation was 

 a favorable one and gave rise to individuals which had 

 a good chance of surviving in the struggle with their 

 neighbors in the field, and with unfavorable climatic 

 conditions in the cultivated state, many a new form 

 would have come under observation, which, as it is, must 

 have perished in its youth. But most new species are 

 weaker than their parent, for example by having smaller 

 leaves, and therefore grow more slowly ; some are partly 

 or absolutely sterile in one sex and therefore cannot be 

 perpetuated at all except by crossing; whilst others come 

 to nought for unknown reasons. 



Variability in the duration of life is especially un- 

 favorable ; and may become positively dangerous in cul- 

 tivation. Normal annual and biennial individuals can 

 be easily made to flower and to set seed, although in- 

 ability to resist cold has been the cause of the loss of 

 some of my most interesting rosettes. On the other hand 

 there often occur plants which behave as if they were 

 going to be annual but grow their stem too late to set 

 seed or even, in many cases, to flower at all, before they 

 are carried off by the winter. In the case of several of 



