VARIATION AS LIMITED BY THE ASSOCIATION OF CHARACTERS. 601 



thenogenic process of germination upon the placenta, and (3) by offsets 

 or buds. Some species of Orinums are said to reproduce themselves 

 occasionally by simple division of the rootstock,* and other genera 

 produce droppers, axillary buds, or underground stems capable of 

 rooting from cuttings at their nodes, &c. But I wish only to draw 

 attention to the first three methods of reproduction. Now under the 

 third section offsets or buds generally produce plants which are true 

 to their parent^ — that is, in which all the parental characters are asso- 

 ciated. I believe that this rule obtains in all cases of bud reproduc- 

 tion in good species, for in all good species all the characters are linked 

 together, and it is this very association (by some innate necessity) 

 which constitutes the species. But when we propagate mongrel plants 

 by the same means we often find that the progeny is by no means true 

 to the parent, because in the latter case we are dealing with plants in 

 which all the characters are no longer linked together by necessity.! 



Plants propagated by sexual self-union resulting in seed exhibit these 

 traits in a more marked manner. The variability of some (mongrels) 

 and the comparative constancy of others (species) is foreordained by the 

 elemental character of the plant, and would be known beforehand if 

 these elemental characters were known in their entirety. But when 

 we consider the results of parthenogenesis I wish to draw special 

 attention to certain observations, the importance of which seems to 

 have escaped attention. Take it for granted that the beauty of 

 flowering-plants and the splendour of their coloration has been 

 attained during the period within which such plants have reproduced 

 themselves by sexual methods ; and T have noted many cases in which 

 parthenogenic reproduction of highly bred plants has been associated 

 with the production of ancestral types of perianth in the progeny, that 

 the change in the means of reproduction to an asexual condition has been 

 associated with a carrying back of the perianth to the condition of some 

 remote ancestor. In other words, that these characters are associated. 



On four separate occasions have I examined the progeny of an 

 alleged hybrid between some garden Hippeastrum and Vallota purpurea. 

 In every case the result was a Hippeastrum of sorts, but far less beauti- 

 ful than the parent Hippeastrum. In three cases the result was for 

 all practical purposes Hippeastrum rutilum, a species which I have 

 always regarded as the oldest, and hence the most rudimentary, in the 

 genus. Now, if in all these cases the attempted hybridization failed to 

 result in cell-union, but exerted sufl&cient irritation to cause some form 

 of parthenogenesis, then the association of character (ancestral type of 

 perianth) with function (ancestral means of reproduction) would account 

 for the depauperated offspring. It should not be thought that these 

 depauperated plants were merely weaklings, and hence carried poor 

 flowers, for such was not the case. In fact, the batch raised recently 

 by Mr. James Hudson at Gunnersbury are exceptionally vigorous 



* This is on the authority of the late Sir Charles Strickland. 



t The garden Cineraria cannot be relied on to come true even from root 

 divisions. I have known double pink individuals produce nothing but single 

 bicoloured blue and white flowers from root divisions. 



