A. St Clair CAroRN 267 



be the more readily observed. There is thus every indication that 

 this slight change in regard to the average glinne length of extracted 

 pure types as compared with the parents is a permanent one. 



Grain Colour. 



Before detailing the numerical results of this aspect of the e.xperi- 

 ment it will be convenient to describe here the difterent colour types 

 found in the F.^ generation. Full coloured purples resembling the 

 Eloboni parent, but of light and dark shades, were thrown. Besides 

 these, however, a number of streaked individuals appeared. The colour 

 of the streak is a dull brownish purple. The purple element is always 

 intensified by treatment with dilute suljjhuric acid, which turns it 

 bright crimson. In Plate XIV, fig. 8, the various colour types are shown. 

 Grain a is the ordinary Polish type, of a light yellow translucent appear- 

 ance, but with no purple colour in the girdle cells, and hence, as far as 

 this cross is concerned, classifiable as ' Non-coloured.' The grain 6 is a 

 full purple, or ' Flushed ' form, while c and d illustrate the ' Streaked.' 

 b', c', d' show the colour changes in acid. The streak shown in d is 

 by no means the minutest recognisable, which may be so extremely 

 faint, possibly because the place for the strongest streak on the plant 

 may be on an unripe secondary tiller gi-ain, as to be nearly invisible to 

 the naked eye, even when treated with acid and examined in the 

 brightest light. In such a case microscopic inspection reveals crimson 

 stained nuclei in a small cluster of girdle cells, but very little pigmen- 

 tation of the sap. (Plate XIV, fig. 10, x.) Care has to be taken, in 

 circumstances like these, to ensure that the colour is really crimson in 

 acid, not reddish or pinkish-brown ; for nuclei and their adjacent proto- 

 plasm often take on these tints in the gi-ains of non-coloureds which have 

 been much weathered or somewhat rusted. In Plate XIV, fig. 10, some 

 of these potentially misleading colours have been introduced for the sake 

 of comparison with the acid reaction of the real purple pigment. 



A noticeable feature in ears containing full coloured purple grains 

 is the way in which the same kind of flush stains the glumes. The 

 glume purple is most intense just before the grain hardens. As com- 

 plete maturity is slowly attained it fades away altogether. In pale or 

 particoloured forms its presence is less vividly marked, a faint purplish 

 margin to the glume being the sole indication. 



Development of colour in the grain seems to be dependent to a 

 variable extent on direct exposure to sunlight. This is especially the 

 case in streaked forms where the width of the gape of the paleae. 



