in the Stock (Matthiola incanci). 471 
mannia is described as 4-lobed (‘ Pflanzenreich iv, 104, p. 144; see also c Bot. 
Mag.’, 1831, t. 3061). 
None of these facts are in good accord with a bicarpellary construction, 
and some are clearly incompatible with it. But in the light of what we 
have now learned from observation of certain Cruciferous genera, they can 
be satisfactorily harmonized and explained. There is no doubt that in 
Eschscholzia , as also in Dendromecon and Hunnemannia , we have not less 
than ten solid and ten other carpels present. 1 The processes of sterilization 
and consolidation, which have reached their height in the Chelidoneae 
among the Papaveraceae and in the Cruciferae generally, are here seen 
actually in progress. The ten solid carpels are indicated by the ten ribs of 
the ovary wall, and further, by the two 5-lobed stigmas seen in Payer’s 
figure of Eschscholzia and again in Dendomecon % In the exceptional 
condition (hitherto unexplained) represented in Baillon’s figure of a fruit of 
E. crocea , we have a stage in which division of labour among the carpels has 
reached a point at which six of the solid and eight, if not all, of the other 
carpels have already become infertile, though the former have retained their 
stigmas. The two solid members on either side of each median carpel, and, 
perhaps this same member as well, have remained ovuliferous, but show 
a decline of the stigmatic function, each median trio forming between them 
only a single stigma, which is less well developed than that of the middle 
member in each lateral group of three, but longer than that of its companion 
(outside) members. On dehiscence each triplet of sterile solid carpels, 
together with the contiguous valves, becomes detached as a single lateral 
compound valve. We see the next stage in the 4-sutured ovary with four 
stigmas, where the stigmas of the four outside solid carpels of the two 
triplets have now entirely disappeared. Finally, in the 2-stigma-bearing, 
2-compound-valved fruit dehiscing by a single split, the ovuliferous carpels 
have also lost their stigmas, leaving only the two of the lateral solid carpels. 
The range thus illustrated within this one species is exhibited throughout 
the genus. Among more than 120 forms listed by Fedde (‘Pflanzenreich’, 
loc. cit.) the enormous majority have four stigmas generally of unequal 
length, apparently the most stable phase to-day. But of the rest one is 
cited with a doubtful six, two with eight, one with eight to ten, one with 
a doubtful twelve, while at the other end of the scale some three or four 
have reached the limit of reduction (2). In face of these facts, whether these 
numbers are invariable or only predominant for each type, G 2 becomes 
a meaningless symbol. Dehiscence usually occurs in the median plane. 
As it appears to be very usual for the split to arise so that one median 
carpel is left attached to each compound valve, the latter would show more 
rows of ovules on the one margin than on the other in a form where the 
1 The total of twenty is based on the supposition that each median flat face is composed of but 
one carpel. Only if this were not so would the number be higher. 
