430 The Systematic Value of the New Species. 



from these seeds will be like. The result can indeed be 

 predicted pretty accurately by means of the law of re- 

 gression (see pp. 72>, 120 et seq.). The seeds of the 

 biennis-^ov^^r will give plants whose flowers revert to 

 the mean of the type of hicnnis ; the seedlings of the La- 

 marckiana-flower will revert to the normal of that spe- 

 cies. 



In other words : if we are in doubt as to the nature 

 of individuals which stand at the boundary between re- 

 lated species, the offspring produced after the self-fertili- 

 zation of the individuals in question will settle the diffi- 

 culty. Two plants which are absolutely identical in re- 

 spect of any particular character may be proved by their 

 progeny to be fundamentally different. And if, as often 

 happens, two related groups only differ in a single char- 

 acter their extreme variants may be indistinguishable. 

 Yet their seed will prove them to be intrinsically differ- 

 ent. 



The study of the limits of species is by no means 

 solely a descriptive one. Classifications based on no more 

 than an examination of a series of forms have no more 

 than a transitory value. ^ Statistical methods^ will re- 

 veal where the boundaries are : experimental methods 

 must be called in to decide in individual cases. 



§ 25. TRANSGRESSIVE VARIABILITY. 



The general conclusions arrived at in the foregoing 

 section may now be illustrated by numerical data. The 

 determination of the nature of the limits between related 

 species is one of the most difficult parts of the task of the 



^ De Candolle, La Phytographie, p. 80 



^ C. B. Davenport, Statistical Methods with Special Reference to 

 Biological Variation. 



