JOURNAL OF THE ROYAL BORTICULTURAL BOCIET1 



and their grand-parents, I felt bound to come to the conclusion that the 

 colour of the dorsal sopal could not in any sense be regarded as a single 

 hlendeliail character, but was undoubtedly a composite one, made up of 

 three distinct characters, viz. : — 



(1) The ground-colour ; (2) tho various markings upon the ground- 

 colour; and (8) the median band, bar, or stripe. 



These three characters seem to be inherited independently of one 

 toother, with hardly any correlation, and the different combinations of 

 these three characters serve to make up the colour of tho dorsal sepal of 

 the flower. 



Having finally selected these three single characters, each of course 

 to l>e considered separately, on its own merits, the next question was to 

 see it they fulfilled the other conditions laid down by Mendel. In 

 Mendel's experiments be found that, to test the matter fairly and 

 thoroughly, the characters must not only be siriglc, but also differential 

 and constant. 



That each of these three characters fulfils both these conditions is 

 evident lot the following reasons: — 



1. They are differential because in their nature they are specific, i.e. 

 peculiar to the species, and for that reason are perhaps even more suit- 

 able for investigations in heredity than the more or less artificial races 

 used by Mendel. At the same time it must be admitted that they have a 

 oertain disadvantage in being of the blended or intermediate type of 

 inheritance in the first generation, rather than the dominant or exclusive 

 type of Mendel's races. 



But this disadvantage, 1 find, is more apparent than real, and though 

 no doubt more complex to work out, yet the results are equally reliable if 

 tb<> investigator possesses an intimate knowledge of the material with 

 which he is working. 



k 2. That these specific characters are constant as well as differential 

 and single is also obvious because they belong to natural species, whose 

 parents and ancestors have practically been the same specifically for 

 many generations, and thus they should be less liable to reversion or 

 change than mere racial or varietal characters whose origin and fixation 

 must have been comparatively recent. In other words, the question of 

 ancestry is practically eliminated. If necessary, a further proof of this 

 is to be found in a study and comparison of the different varieties of 

 the three natural species with which 1 have been working, viz. : — 

 P, S})wcruinum, P. imigne, and P. Boxallii ; all three being clear-cut 

 and well-defined spocies. Of the three, P. insigne is certainly the most 

 variable, covering apparently a wider area of country than the other two ; 

 but however much its varieties vary among themselves, they still all 

 retain their specific characters intact, as do the comparatively few 

 varieties of the other two spocies. 



Having satisfied ourselves that tho characters selected fulfil the con- 

 ditions laid down by Mendel, we can at once proceed to the details of the 

 investigation. 



It will bo remombored that we set out with the idea of testing 

 Mendel's theory as to the purity of the single character determinants in 

 the germ-cells of hybrids. 



