6 i Us 
1913] EAST—CROSSES OF NICOTIANA 181 
In the F, generation plants were produced with a corolla breadth 
identical with each parent. In fact, there were four individuals 
as small as the smallest specimens of N. forgetiana, and there was 
one individual with flowers within 3 mm. of the size of those of the 
largest N. alata grandiflora growing in my cultures. The mean 
of this distribution was 48.57 +o.19 mm., the standard deviation 
8.07 0.13 mm., and the coefficient of variation 16.62+0.28 per 
cent. 
TABLE II 
STATISTICAL CONSTANTS FOR FREQUENCY DISTRIBUTION OF TABLE I 
Designation Mean Standard deviation | Coefficient of 
N. peg AA eee 25.60.12 2.27+0.08 8.86+0.33 
meme OF., $97) ie 78.80.28 5.380. 20 6.820. 25 
F; (314 S328) ice Wis eel eis 44.30.23 3-67%0.17 8. 280.38 
Wa tata 39%) 1-65.06. 49.90.26 I1.26+0.19 22.57=0.39 
Examination of the F, generation of this cross indicated a 
correlation between the length of the corollas and the lengths of 
the filaments and the styles that for all practical purposes was 
perfect. By this statement I mean that the anthers were invari- 
ably just above the stigmas and the stigmas were invariably 
at the throat of the corolla. Of course absolute measurements 
would not show an integral coefficient of correlation, nevertheless 
one may assume, I think, that the fact is simply obscured by slight 
fluctuations. It seems as if the numbers were adequate from which 
to conclude either that the determiner or determiners of corolla 
length are also determiners of the length of the style and the fila- 
ment or that these factor complexes are perfectly coupled in 
inheritance. 
Corolla spread is also correlated with corolla length. It is by 
no means uncommon to find a sharp break in the correlation, as 
is witnessed by the individual with the very broad corolla and 
comparatively short tube pictured in pl. X, fig. 8; but one never 
s inverse extremes in the same individual. Just what the 
correlation coefficient 0.610.015 would prove to mean if indi- 
vidual analysis of later generations were made, is doubtful. We 
cannot go back of the gross statement that such a correlation exists 
