1899] COL O URING MA TTERS OF FL O WERS 1 49 



standing that in the latter case the process of oxidation has been more 

 energetic and more complete than in the former case. Even where 

 the blue is apparently the result of a combination of a tannic acid 

 with a base, the effect cannot be produced save under circumstances 

 especially favourable for oxidation. The flowers comprised within the 

 great division Corolliflorae having a gamopetalous corolla, enjoy, by 

 reason of the great expansion of cellular surface, a respiratory and 

 transpiratory activity which the Polypetalae cannot exhibit. We see 

 that all or most of our decidedly blue flowers are gamopetalous, e.g. 

 gentians, bell flowers, Jacob's ladder, convolvulus, speedwells, various 

 labiates, borage, etc, Linum percnne is an example of a bright cobalt 

 blue-coloured polypetalous flower ; but it may be readily observed that 

 the petals are very broad and generally large in comparison with the 

 thin and wiry stem and the small strap-shaped leaves ; and I know of 

 no valid reason why, under such propitious circumstances, the quercetin 

 of the cell sap may not in the absence as here of any phlobaphenic 

 tannin, develop to the full, and sowise evolve a high oxybenzoic acid 

 which, being immediately transformed into pigment, may not be 

 readily detectable in the free state. 



In view of the researches already published, and of my own experi- 

 ments, I think it may be concluded, that — 1. A blue flower is unpro- 

 ducible in species which contain or are capable of forming phlobaphenic 

 tannin, no matter what the development of the inflorescence may 

 amount to. 2. A blue flower is more likely to be produced in a species 

 having a gamopetalous corolla or perianth, and therefore liable to 

 evolve by higher oxidation a certain quantity of a high oxybenzoic 

 acid. 3. In species wherein the tannin natural to the organism is 

 iron-greening and non-phlobaphenic, a blue flower may possibly be 

 producible in a polypetalous corolla, provided always that the petals 

 or perianth be large relatively to the height of the plant and to the 

 size and robustness of its stem and leaves : in this case it is uncertain 

 whether gallic acid is necessary for the production of the effect, but 

 anyway an alkaline compound of an oxybenzoic acid would seem to be 

 indispensable. I hope in a concluding paper to specifically examine 

 these three propositions, and to supply full illustration of their scope 

 and tenor. 



11 Ql'Een's Square, 



London, W.C. 



