7(J 



Coo2>eratire Investigatioiis &n Plauts 



parental relationship of plant means. Oiir meagre results may then be sdinmcd 

 up iü thc foUowing table : 



TABLE IX. 

 Grandparental Inheritance. 



These results at least iudicate that, however determined, the grandparental 

 relationship is fairly dose to half the parental. It is much to be regi-etted that we 

 have not wider data, especially on a less luxuriaut crop than the Kidderminster 

 one, for the above results arc very variable. 



(10) Colour Inheritance. W. R. Macdonell made an claborate Classification of 

 the colour of 1604 flowers divided according to their parentagc into 24 groups. 

 He fornied 13 colour classes, which after consideration were classified into tbree 

 groups according to the intensity of red colouring matter in the flower. The first 

 group embraced all the ränge from dark red to red-whites ; the second the pure 

 piuks and the third all the piiik-whites to pure whites. In this Classification 

 attention was not paid to the base of the petal, but to the middle portion and 

 margin. The foUowing are thc frequencies of the original scale : 



Classes (8) and (11) had representatives in crops from Seed (ß) and (y). 

 Classes (1) to (5) foruied our first general class, (6) our second, and (7) to (13) our 

 third. This grouping enabled us to use the method for dealing with characters 

 not quantitatively nieasurablc, assuming that the distribution of red colouring 

 matter is an approximately normal one. If a be the Standard deviation of the 

 whole population of 1604 offspring flowers, and 2 the mean Standard deviation 

 due to the arrays having a common parent 2 = <tV1— r*, where r is parental 

 correlation. Now a was found in terms of p, the pink ränge, by means of tables 

 of the probability integial, and we have o-=^;(1021S). The staudard deviation 



