METAMORPHOSIS. 165 



Velenovsky figures a similar instance in G. hifiora.. It 

 was observed that in G. sagittata (PL XLIV, fig. 1) and 

 C'. dionseaefolia (fig. 127) the leaf possesses two quite 

 free basal lobes which are infolded and adpressed to 

 its upper surface ; in the last-named species and in 

 0. appendioilafa {G. limbata) these lobes were, more- 

 over, completely adnate a third or more of their length 

 to the leaf -surf ace (PL XLIV, fig. 2). 



Other species in which such adpressed basal lobes on 

 all the leaves were observed as a normal feature are 

 G. introlob't, C. marginafa, and (7. Novse-Zelandice* In 

 Saxifraga ciliata a medianly-placed, stalked, inversely- 



FiQ. 127. — Caltha dionassefolia. TransTerse section of foliage-leaf 

 showing ventral lamellae, or the upturned basal lobes {vl', in illus- 

 tration of anther-struoture. The black part is the upper surface 

 of the leaf. Diagrammatic. 



orientated, basal leaflet was seen lying almost flat 

 against the upper surface of the leaf. 



The large single basal pocket, or, when this and the 

 main lamina are about equal in development, the 

 pitcher-shaped character of the foliage-leaf (as in 

 Saxifrwia), of the leafy, or of the petaloid anther, 

 represents merely the stage in anther-formation im- 

 mediately preceding that in which the median lamella 

 are formed ; the next step, as shown by so many struc- 



* It may be mentioned that in most, if not in all, of these species the 

 apex of the main lamina was conspicuously emarginate, a fact which is of 

 considerable interest for comparison with the forked apex of Celakovsky's 

 abnormal ffieracium-leaf. It is suggested, . therefore, that in the case of 

 Caltha a, correlation exists between the presence of median lamellbe (or 

 adpressed basal lobes) and an emarginate apex. 



