THE VARIABILITY IN CULTIVATION OF HARDY FLOWERING PLANTS. 228 



in my garden different from what I planted, but which I know 

 to be the spontaneous offspring of them. 



Whilst in several genera many of the species cross so readily 

 that I have found it hardly possible to keep them true from seed 

 even for a single generation, other classes of plants go on from 

 year to year reproducing their parent without any suspicion of 

 a cross. The most habitual and persistent mixing of species I 

 have seen takes place amongst Columbines. Every kind seems 

 to cross readily with every other, long-spurred and short-spurred, 

 yellow, blue, and red ; I never can tell what home-saved seed 

 will produce. In their native land Columbines do not seem to 

 vary much. I have observed thousands of plants of Aquilegia 

 vulgaris in the Pyrenees which were all constant. So it is with 

 the American kinds. I have raised many a lot of seed collected 

 wild, and the plants are uniform in colour and character. So it 

 is if one kind is isolated ; a nurseryman at Forres, in the 

 Highlands, has for thirty years supplied the market with A. 

 glandulosa, which has proved these perfectly constant. But 

 where grown together the plants continue to change in every 

 generation, and, as all the hybrids are more or less fertile, 

 there is no limit to the degrees of mixture. Still I notice a 

 decided tendency in each successive generation to draw towards 

 the most robust type, which in my garden is A. vulgaris, and I 

 believe, if left for a few years to themselves, they would all be 

 absorbed in this. 



The genus Dianthus is quite as prone to cross as Aquilegia. 

 I have never tried to make a florist's flower out of it, or had 

 recourse to selection ; but I save the seed of the small alpine 

 species, and raise it, and allow the more robust kinds to sow 

 their own seed and to come up indiscriminately. Both the 

 dwarf kinds, of which I may call D. alpinus the type, and the 

 taller umbellate forms, of which the commonest is D. barbatus, 

 hybridise freely amongst themselves and with one another. 

 Some of the larger kinds intermix so as to make it hopeless to 

 determine the species to which their parents belong. Such 

 plants as D. ccesius, D. fragrans, D. petrceus, D. phmarius, and 

 one or two more seem to combine in a race of fertile hybrids, 

 the flowers varying greatly in size and colour, and in the fringed 

 or even margin, none of them, however, being constant from 

 seed. These kinds seem to tend more and more to a form, 



