SIR J. LUBBaCK — PIIYTOBIOLOGTCAL OBSEUYATIOXS. 381 



In the Bignonlacere, again, a large number of species have 

 emarginate cotyledons ; and this would appear also, as for instance 

 in PiiJiccoctemum AiihlefiL to be due to the chalaza, thouo-H in a 

 different manner. The seeds themselves are transvcrselv oblouf>", 

 much compressed dorsally, surrounded on all sides except tlio 

 base by an extremely thin, transj)arent, membranous wing, which 

 is traversed by nerves radiating from the central part of the seed; 



• Fig. 102. 



Seedling of Chclranihus Chelri, Two thirds uat. size 



and is uneven at the margin. The raphe is ventral, extending 

 from the hilum to the centre of the embryo. The chalaza is 

 attached to the embryo-sac a little above the middle of the embryo. 

 The radicle is veiy small, distant from, but pointing to, the liiium. 

 The embryo is straight and flat; the cotyledons grow until 

 they come to the point of attachment, of the chalaza, when they 

 extend forwards on each side, forming two lobes. 



In Oroxylum indicnm the general structure of the seed is \^rj 

 similar, but the growth of the two lobes of the cotyledons is even 

 more luxuriant, so that they actually overlap. A structure more 

 or less similar occurs in other genera of this family. 



The emargination is very much deeper in other groups, and 

 due to other conditions, for instance in the Convolvulacese. In 

 Convolvulus Soldanella (fig. 103) the embryo, which iss eventually 

 very large, lies at first straight in the seed embedded in a clear 

 jelly-like perisperm, and rests on a solid, ovate, grooved, white 

 ridge (figs. lOi and 105, «), which rist?s from close to the micropylc. 

 This tongue-like ridge grows with the embryo. At the opposite 

 end of the seed the raphe and chalaza form a somewhat pro- 



