502 Conclusion. 



the type have, so far, escaped me altogether. I have 

 hitherto only experimented with ordinary sowings ; and 

 my object has been rather to become familiar with the 

 principles of mutation, than to bring to light a multitude 

 of novelties. 



I gradually came to see that the method of searching 

 for mutations was capable of improvement in many re- 

 spects. There seem to be two main ways of doing this : 

 the choice of seeds and hybridization. If the harvest 

 turns out to be meagre for one reason or another, or the 

 fertility of the seeds diminishes greatly (that is to say 

 only a small percentage remain fertile) the prospect of 

 getting mutations in general, or at any rate of getting 

 particular forms, seems to be considerably increased. 

 For example in a sample of seeds which had been kept 

 for 5% years the fertility went down from 70 to 5 seeds 

 per cubic centimeter; but the percentage of mutations 

 went up from 1-5% to 40% (p. 263). In another cul- 

 ture only about 30 seeds germinated out of the whole 

 harvest sown; but of these, 12 gave rise to mutants which 

 formed therefore 40% of the population. And the view 

 that crossing increases variability is generally held and 

 seems to be supported by some of my experiments. 



My cultures were conducted on the following lines. 

 From seeds or plants which I gathered at Hilversum I 

 derived my so-called families as follows : Seeds were 

 collected every year from a few (e. g., 4-10) individuals. 

 These were chosen as being typical examples of the fam- 

 ilies in question and were either left to be fertilized by 

 insects on some isolated spot (1887-1894), or (as in 

 later years) were protected from the visits of insects by 

 parchment bags and artificially fertilized with their own 

 pollen. It was only in /a/a-families that crossing always 



