516 



ANNUAL EEPOET SMITHSONIAN INSTITUTION, 1908. 



are known to be direct derivatives of simple-leaved ancestors, and are 

 therefore capable of easy observation. With these facts in hand the 

 possibility suggests itself that we might be able to distinguish between 

 a structure recently arisen and one which was borne by a related 



sjjecies for thou- 

 sands of years, but 

 the exactness of 

 such estimates is 

 a matter of con- 

 jecture. 



It seems to be 

 taken for granted 

 by a great number 

 of workers inter- 

 ested in this sub- 

 ject that sj^ecies 

 showing wide vari- 

 ations of the or- 

 gans are the ones 

 most likely to offer 

 a high frequencj^ 

 of mutations, and 

 my correspondents 

 in various parts of 

 the country are 

 constantl}^ calling 

 attention to these 

 forms under such a 

 mistaken impres- 

 sion. As a matter 

 of fact we may con- 

 fidentl}'^ expect that 

 the species which 

 show the greatest 

 variation, or are 

 ever ajoorting, are 

 the youngest. 



Now, having ob- 

 tained the result 

 just described, we 

 find ourselves face 

 to face with one of the most interesting and difficult questions in 

 heredity. If the newly arisen mutant forms are more widely variable 

 than the older ones, how do they ultimately become narrowed? 

 If the greater number of species originated by mutation, as we 



40 ^ so 55 



Fig. 4. — Range of variation of tiie ratio of width to length 

 of leaves in the parental Lamarckiana and one of its 

 mutants. The dotted line shows that the range in the 

 parent is 28-53, with a coefficient of variability of 9.53 ± 

 0.22 per cent, while in the mutant represented by the 

 solid line the range is 20-48, with a coefficient of vari- 

 ability of 10.30 ±0.22 per cent. 



