SUPPLEMENT 123 



explained. JOHANNSEN'S (1903) researches at least form an important advance 

 towards the interpretation of the subject. This investigator shows that it is 

 possible by selection of the largest seeds of a certain type of bean to incr ease 

 rapidly the average weight. When, however, he employed not the offspring 

 of a whole field but that of a single individual, he always obtained the same 

 average value whether he started from small or from large seeds. He con- 

 cludes from these experiments that an apparently pure variety still includes 

 sub- varieties QOHANNSEN calls them ' lines '), which differ in the average 

 weight of their seeds. When specially large beans are selected from a mixture 

 of such ' lines ' as the point of departure for further cultures, they may be 

 extreme forms of the small-seeded lines, but probably they belong to the large- 

 seeded lines. If selection be carried on in the same way for two or three genera- 

 tions a selection of the large-seeded lines will be the result. Similarly, in the 

 case of other selections there will not be any isolation of varieties which have 

 newly arisen, but of already existing races from a general mixture of these. 

 This explanation is not satisfactory in all respects, because it does not account 

 for rapid reversion, associated with the cessation of selection. 



393, 1. 3, for third read second 



I. 9, for or a single bud read or in a single bud. Still it is not the amount 

 of the leap but its hereditary constancy that is the essential characteristic of 

 a mutation. 



II. 33-4, for (e.g. white-flowered . . . characters, read ; they have either 

 lost a feature possessed by the type species (retrogressive varieties), such as 

 a pigment, hairiness, &c., or exhibit a character which occurs in allied forms, 

 and which they once possessed and have again assumed (degressive varieties). 

 These sub-species, however, are distinguished from each other in all features, 

 and arise by the formation of new characters (progressive varieties). 



1. 40, after characters, read [That DE VRIES'S mutation theory is really 

 founded on a garden escape is for many reasons unfortunate. The purity of 

 the breed may be all the more questioned since the plant is no longer to be 

 found in its original home (Texas).] 



394, 11. 9-11, for predominant ; . . . previously read predominant, if MENDEL'S 

 laws of prevalency and segregation hold good. In the case of the Oenothera 

 mutations, and in progressive mutations generally, that is not the case, how- 

 ever (DE VRIES, 1903) ; the hybrids with the principal type or with another 

 mutation even in the first generation take on many forms. 



In the following generations the forms arising in this way remain quite 

 constant ; in other words, the mutations are not obliterated by crossing. In 

 an extreme case it is obvious, however, that the first hybrid generation may 

 be uniform, so that the new mutation alone makes its appearance. 

 1. 13, for 1891 read 1791 



395, 1. 19, after itself read 



We have regarded mutation as something totally distinct from variation ; 

 this is true, no doubt, for certain cases, but not for all. DE VRIES has observed 

 a great number of races which must have arisen by mutation but which never- 

 theless transmit their characters entirely ; into the discussion of these half 

 and intermediate forms we cannot enter, but content ourselves with noting 

 the existence of transitions between mutations and variations. Again, this 

 must be the case from another aspect, viz. in relation to origin, in so far as 

 the variations appearing may be the result of external factors. 



The better known cases of interest in this relation occur among micro- 

 organisms. It has been found possible with the aid of high temperature 

 and by use of poisons to deprive certain Bacteria of the power of forming 



