IN THE TROPICS : III. 147 



Here and in similar cases the average of the different percent- 

 ages is of little moment, since the range in variation is far 

 beyond the ordinary limits of error. 



Assuming that the 4 cobs with 1 to 4 blue grains 

 are cases of accidental "xenia," a supposition which is 

 rendered probable by the fact that from 1 to 5 blue grains 

 appeared in about 10 per cent, of cobs of the white dent 

 in neighbouring rows, we get 73 per cent, or 55-5 out of 76 as 

 the proportion of white F 2 grains which were homozygotes ; * 

 the expectation being 50 out of 76, as has been already pointed 

 out. 



The essential point to be noticed is that grains which were 

 to all appearances white showed themselves, upon a study 

 of the offspring, to have been of impure constitution : such a 

 number of them however bred true as to indicate that Mendel's 

 law of purity of the gametes still holds good in such a case. 

 We therefore conclude that in this instance the heterozygote 

 does not necessarily exhibit that character, namely blue, 

 which might have been expected to be dominant, but may ap- 

 pear either blue or white. We may here refer back to Expt. 

 34, in which a similar circumstance was actually observed in 

 the first generation of a cross (blue < white) between definite 

 strains. 



Expt. 64. — The plants arising from the blue grains of the 

 same generation , and pollinated by the same white dent strain , 

 produced a number of cobs — F 3 — -containing from 22 to 40 

 per cent, of blue grains, the remainder of the grains in each 

 cob being white. Altogether 4.034 blue and 10,300 white, or 

 28 T per cent, of blue. 



* If on the other hand these plants represent the lowest terms of 

 the series properly showing various numbers of blue grains, this pro- 

 portion becomes 62 per cent, or 47 out of 76. The point could have 

 been tested by growing a further generation an a very large scale, 

 but in either ease, in consideration of the small number of plants 

 obtained, the agreement with Mendelian expectation is reasonably 

 close. 



