592 PLANTS JAVANICiE RARIORES.
inflected parts, or, as they are commonly called, placentae, it not unfrequently appears to be bilocular. But this cohesion only occurs when the production of ovula is con- fined to the upper or inner surface of the carpel, which is the case in the greater part of the genera with elongated capsules ; for Avhere both surfaces are ovuliferous, as in the baccate and most of the genera with included capsules, no such cohesion can take place. The general direction of the margins of the placenta of each component part of ovarium and pericarpium may at first appear a deviation from the ordinary structure, the general rule being that the margins only unite to form a complete cell, wdiereas the completion of the cell in that manner is incompatible with the direction of these margins, which in each carpel are turned from, not towards each other. This difference, however, is more apparent than real, and the structure in Cyrtandrece may be justly compared with that of such genera of other families as have the placenta of a multi- locular, or that of the single distinct, carpel projecting considerably into the cavity.
The great elongation of pericarpium in many of the genera having capsular fruit, is the more remarkable in Cyrtandrece, as there is no instance of similar elongation, or any approach to it, in either of the two other tribes of Gesneriacece. In this elongation of capsule, however, they approach to Bignoniacece, where it is both more general and exists in a still greater degree.
In most of the Cyrtandrece with elongated capsules, the valves, though membranaceous, are perfectly straight, but in a few others they are spirally twisted, though nearly of the same texture. The spiral torsion of the valves certainly does not depend on the length of the capsule merely, the greatest length being found conjoined with straight valves, as in JEschynantliiis ; nor is it the consequence of drying, for the twisting in all cases commences long before the ripening of the fruit. The mechanism explaining these ni] differences is, however, in general obvious. In the twisted valve the endocarp consists of a stratum of vertically elongated fibres, with an extremely thin or hardly manifest
�� �
