462 Saunders. — A Reversionary Character 
exceptional circumstances, as e. g. when the solid carpels are lacking (if this 
condition occurs, see above, p. 458) or when they are so small that the two 
edges of each lateral valve become conjoined at the point of contact, each 
thus forming a fertile closed cell to the exclusion of the solid median pair, as 
in Menonvillea , Hexaptera, Biscutella (Fig. 36). It is over the valves, how¬ 
ever, in the Stock that the stigmatic papillae are centred, 1 though in the 
young siliqua before the sutural knobs enlarge they appear as an almost 
complete ring. In fruits of glabrous plants the sutural lines, as stated 
above (p. 452), can sometimes be seen to be continued over the shoulders of 
the knob, which in such case is derived in part from the edges of the two 
neighbouring valves (Fig. 37). We have, then, now to regard the typical 
siliqua as composed of four carpels, of which the lateral pair alone retain 
the valve form, the median pair having become solid. As a rule, solid and 
valve carpels regularly alternate, but we have already seen that in fruits in 
which both valve and solid carpels occur, but in which reduction has pro¬ 
ceeded asymmetrically, juxtaposition of solid carpels and sometimes even 
of valve carpels may occasionally be seen. When two valves are contiguous 
their edges join where they meet in such a way as to show no sign 
externally of the soldering, the line of junction forming merely a blunt 
angle. Less rarely fruits were found exhibiting juxtaposition of solid 
carpels. In two instances (Cases 2 and 5) the interpolation of a valve part 
way up the siliqua showed that in reality the two well-developed solid 
carpels must have been separated by a very reduced third member, which 
assumed valve form as it became successful in the competition for space. 2 
In another, however (Case 4), the triple vascular supply and triarch contour 
line of the cross-section of one commissure proved that a small solid carpel 
had been formed in each diagonal plane on either side of and in contact 
with one solid median carpel. Although the central member of the trio 
got as far as forming a projecting rib of tissue, it did not attain to valve 
form. We find in fact that the symmetrical alternate arrangement, though 
characteristic, is not invariable. In the ancestral form all the carpels must 
undoubtedly have been similar and valve-like, and the processes of reduction 
and consolidation here assumed to have taken place may well have pro¬ 
ceeded by symmetrical stages. But a partial return to the primitive state 
is very likely to occur through some favourable condition so localized as to 
be one-sided in its effect. 
1 This arrangement, though very exceptional, is not confined to Matthiola. It occurs in the 
nearly allied genus Moricandia ( M . divaricata , Coss.), where also the stigmas are sessile, and 
probably elsewhere (see below, p. 466). 
2 Occasionally in a wide commissure with two main fibro-vascular bundles the two bundles may 
diverge so much that a considerable breadth of non-vascular tissue is developed between them, 
simulating an interpolated valve. That such tissue does not repre-ent a true valve carpel is shown 
by the entire absence of vascular tissue and by the outline, which is V-shaped at the base instead of 
having the rounded contour of the true valve. 
