CRUCIFERJE. 217 



15. The number of seeds varies from one or two to fifty and up- 

 wards. Sometimes subbasilar or subapical, descending or ascending, 

 they are attached by a funicle of variable thickness, which may be 

 quite free, or remain adherent for a very variable extent to the false 

 septum. There are three seed-coats ; the inner membranous ; the 

 middle, more or less testaceous, often prolonged into awing; the super- 

 ficial, often thin and epidermoid, whose cells often sw r ell up in water 

 into a thick layer of mucilage, as in the Mustards, &c. The presence 

 of a thin layer of albumen is exceptional ; the embryo usually fills 

 the seed-cavity. Frequently when the fruit is broad there are two 

 rows of seeds in each false cell, while in narrow elongated fruits there 

 is but one. However this point, on which great stress has been laid 

 in classification, may vary from species to species in a single genus, 

 and even from siliqua to siliqua on a single individual. 1 



10. The embryo is an organ to which the highest importance has 

 been ascribed in the classification of this order. Its radicle, often 

 ascending, 2 is usually folded on the cotyledons. 3 If these are flat and 

 it is applied to their commissure they are termed accumbent ; 4 if it lies 

 on the back of one of them they are termed incumbent? Or the cotyle- 

 dons are con duplicate? folded across one inside of the other, with the 

 radicle inside the groove between the two halves of the former. Or 

 they may be biplicate, 1 or folded twice transversely ; or else coiled in 

 a spiral. 8 In intermediate positions the radicle is more or less ob- 

 lique. Moreover, the cotyledons may be entire, emarginate, bilobate, 

 or bifid, as in Schizopetalon. 



Which then of these variable characters have been used by 

 botanists in their subdivision of this order ? First the form of the 

 fruit. Linnjsus and his followers divided Cruciferce into Siliquosa and 

 SiliculosfS ; and so did A. L. de Jussieu in his Genera Plantarum. 

 Adanson went much further in his Families des Plantes. He more 

 skilfully distinguished the form of siliqua that dehisces longi- 

 tudinally, from that which is lomentaceous and divides across, and 



1 See Wedd., Chlor. Andina, i. t. 85. — J. 5 Hesperis, Sisymbrium, 'Erysimum, &c. 

 Gay, in Bull. Soc. Bot. de Fr., x. 9. 6 Brassica, Moricandia, Succowia, JEruca, 



2 " Radiculam ssepe ascendenteia a placenta Crambe, Rapistrum, Raphanus, &c. 

 remotara." (B. H., Gen., 57.) 7 Heliophila, Chandra, &c. 



3 Straight in Leavenworthia. 8 Bunias, Erucaria, &c. 



4 E.g. : Arabia, Muttliiola, Cheiranthus, Nas- 

 turtium, Cardamine, Lunaria, Yesicaria, &c. 



