354 



THE JOUJINAL OF BOTANY 



figures suggest that increase or reduction in the number of parts of 

 the flower of V. persica is largelj^ dependent on nutrition, as is the 

 case in Papciver duhium (see Journ. Bot. 1922, 299). It maj be 

 noted in this connexion that 39 of the floAvers examined b}^ me ex- 

 liibited deviations from the normal number of parts in two whorls. 

 Increases in two whorls occurred in 38 flowers, and one flower exhi- 

 bited decreases in two wdiorls. No instance Avas observed of an 

 increase in one whorl associated with a decrease in another. This 

 points to a general cause of increase or decrease. 



Conclusions. 



The data regarding floral variation in Veronica persica are con- 

 sistent with the hypothesis of a primitively pentamerous flow^er. They 

 point to the calyx being less variable than the corolla, and to its 

 reduction having followed on that of the latter. Out of 108 flowers 

 observed by me in which the anterior petal was missing, only one 

 had the calyx reduced to three sepals. 



The frequent occurrence of anterior staminodes suggests that the 

 suppression of the anterior stamens did not long antedate — if, indeed, 

 it did not sjaichronize Avith — the fusion of the two posterior corolla- 

 lobes. The more recently a member has been lost, the more frequently 

 it may be expected to recur. Starting with a h^'pothetical flower of 

 the formula K-C^A''G^ V. persica appears to have undergone reduc- 

 tion in six stages, as illustrated b}^ diagrams of flowers actually 

 observed; the aestivation of the sepals is not shown, as the flo^vers 

 were all expanded when examined ; and the gynseceum being normal 

 in all cases has been omitted. Stage I, suppression of the two anterior 

 stamens ; II, fusion of the two posterior coroUa-lobes ; III, suppression 

 of the posterior sepal. These changes would have produced a flo\\er 



