Inheritance of the heptandyo-lorm. of Digitalis f>iirpurca L. 201 



ciently reduced to show them, but jiurtly also because the dorsal 

 segments of the accessory corolla are not as generally present even 

 in flowers showing the highest development of the /ie/>fa>icira-cha.Ta.cter 

 as are the ventral lobes. De Chamisso's figures 2 and 3I) (repro- 

 duced in my Plate I as figs. .\ and B) show the complete accessory 

 corolla, to the plarts of which De Chamisso gave distinctive names as 

 follows: The median dorsal lobe = "(c) lacinula postica imparl"; the 

 lateral dorsal lobes -- "(f) alac laterales"; the ventral lobes =^ "(i) 

 lacinulae coroUinae stamina primaria stipantes". The median dorsal 

 component of the accessory corolla is also shown in two of Miss 

 Saunders's figures (her figs. 4 and 5) though in both the flowers 

 represented by these figures it was very small. 



The morphological significance of these five petaloid structures 

 forming the accessory corolla, seemingly occupying the sinuses of the 

 normal corolla and becoming visible only when the lobes of the 

 normal corolla have been completely modified to stamens, is not 

 clear. Perhaps they are in some way related to the filaments of the 

 stamens which they subtend. The filaments are normally adherent 

 to the tube of the corolla at base, and this may indicate such close 

 approximation of the fundaments of petals and stamens that the 

 morphogcnic activities ojierating to produce these two sets of organs 

 maj' become slightly confused, the activities belonging normally to 

 the one cycle being transferred in some degree to the adjoining cycle. 

 In other words, it may be conceived that the "growth-wave" which 

 in tlie stamiual cycle gives rise to a stamen, affects to a slight degree 

 the adjacent coioUar cycle and there produces a petaloid out-growth 

 lying in the same radial plane as the stamen. 



The nature of the mutation by which the hcptandra variety was 

 produced from the normal Digitalis purpurea may be thought of in 

 similar terms as a still more extensive confusion of morphogcnic 

 activities which are normally limited to two separate, closely adjacent 

 cycles of development, namely, the staminal and the corollar. This 

 confusion is still further evidenced by the frequency with which 

 union of normal stamens with the accessory stamens appears in the 

 flowers of the kept and ra-ioxm. De Chamisso's fig. 3 (my fig. B, 

 Plate I) shows a striking example of this phenomenon, three of the 

 four normal stamens being each united with an adjoining accessory 

 stamen. Three of my figures (figs. 7, 8, and 9) show similar con- 



1) [,innaoa 1: 571 — 575, I'late \'I. 1826. 



