DISCONTINUOUS VARIATION. 51 



Thus there were obviously two forms present, a 13 ray 

 form and a 21 ray form. The seeds from the 12 and 

 13 ray forms were collected and sown next year, the 

 flowers obtained therefrom having the following num- 

 bers of florets: 



Ray florets, 8 9 10 11 12 13 14 15 16 17 18 19 20 21 

 Frequencies, 2 107 13 94 25 771 2030 



That is to say, all trace of the 21 ray form had been 

 eliminated, and a nearly pure 13 ray form obtained. 

 That this was so was proved by sowing the seed of some 

 of the 12 rayed plants obtained on this occasion in the 

 following year. It was then found that the frequencies 

 of occurrence of flowers with various numbers of rays 

 remained practically unchanged. 



But how do these cases of what Bateson has termed 

 discontinuous variation arise ? In one or two of the in- 

 stances quoted we saw that the two humps of the curve 

 of variation scarcely overlapped at all. In the case of 

 Primula they were all of them sharply defined, but 

 there was still a good deal of fusion, whilst in the ray 

 florets of the Ox-eye Daisy the fusion was greater still. 

 Finally, in the frontal breadths of Naples crabs the 

 fusion was complete, and the existence of dimorphism 

 was shown only by the asymmetry of the curve. It 

 would be possible to multiply instances of such curves 

 as these, in which the fusion ran through all stages of 

 completeness and incompleteness, but those quoted are 

 quite sufficient for our purpose. They suffice to show 

 that all stages of fusion may be met with, and so incline 

 one to the opinion that the later stages, in which the two 

 or more humps of the curve overlap little if at all, are 



