416 REVIEW AND CONCLUSION 



in having numerous stamens and separate carpels 

 arranged spirally on a convex thalamus ; the CAPPA- 

 RIDE^E and CRUCIFER^E in the parietal placenta- 

 tion of their ovules ; the MALVACEAE, STERCULIACE^E 

 and TILIACE^E in their numerous stamens ; the GERA- 

 NIACE^, RUTACE.E and MELIACE^E in their promi- 

 nent disc, which is still larger in the RHAMNE^:, 

 SAPINDACE.E and ANACARDIACE.E. The connexion 

 between the PAPILIONACE^:, (LESALPINE.E and MIMO- 

 SE.E has already (p. 324) been pointed out ; the 

 MELASTOMACE.E, MYRTACE.E and LYTHRACE^ resem- 

 ble each other and differ from the foregoing in their 

 completely inferior ovaries. 



And in all these orders the petals are free or, if 

 connected, only slightly so and mostly above, not at 

 the base. We may therefore class them together in 

 one large group of families, the Polypetalae, as distinct 



from the RUBIACE.E, COMPOSITES, MYRSINE.E, SAPO- 

 TACE^E, APOCYNACE^:, ASCLEPIADACE^, CONVOLVU- 

 LACE.E, BORAGINE^E, ACANTHACE^, and LABIATES, 

 which are Monopetalae, and differ also from the 

 former class in the greater definiteness of the floral 

 parts (in none for instance are the stamens indefinite) 

 and in other ways. 



Of these orders the first two have inferior ovaries, 

 and stamens equal in number to (isomerous with) the 

 corolla lobes; the ASCLEPIADACE^E and APOCYNACE.E 

 also have isomerous stamens, but a superior two- 

 carpelled ovary and thick white latex. There are, of 

 course, many more natural orders than the few studied 

 in this book, and with them the groups would be 

 found not only larger but sometimes better defined. 



