THE ASSOCIATION OF CHARACTERS 321 



of the upper part of the petals. It produces the variety 

 known as the Danish flag and characterized by its broad 

 central white cross on the red field of the petals. 



It goes without saying that these splittings may not 

 only be produced by artificial crosses, but must occur also 

 whenever accidental crosses are brought about by insects 

 visiting the variety and the species when cultivated side by 

 side. If this takes place in a nursery, the horticulturist will 

 recognize the new types, isolate them as such, and put them 

 on the market. They will prove constant, or at least become 

 so after some additional splittings. In other words, many 

 intricately colored species must afford, besides their white 

 variety a larger or smaller number of types, in which the 

 constituents of their color-designs are more or less com- 

 pletely isolated. By this means we can study the units by 

 simply cultivating and comparing all the commercial varie- 

 ties of such a species. 



Having thus mentioned the different methods by which 

 unit-characters may become isolated, in order to give proof 

 of their real existence, we now come to the other side of the 

 question. Our conception of units was originally based on 

 the desire of having some principle which might explain 

 the internal causes of correlation, or at least some of them. 

 We have seen that botanical marks observed in the scales 

 of an ear of wheat or barley may be indications of hardiness 

 in winter, of fitness for definite commercial purposes, of 

 resistance to diseases, and other valuable qualities. In order 

 to explain such coincidences, we have assumed that unit- 

 characters are not productive of single marks, but may ex- 

 hibit their influence in different parts of the plant body. 

 The same unit may become visible in the color of the stem, 

 foliage, flower, and fruit, and so it must also be with other 

 units, which, when added to a type, not only change its 

 flowers or its leaves, but also affect other organs, physiologi- 



