110 



is to be placed next to the Bixacese (Parietales). Ifc is also to some extent 

 comparable to the Simambaceae (Geraniales). Moreover, the resemblance 

 between the KoeberHmaceae and the Capparidace?6 is so close that the former 

 is by some authors incorporated into the latter. 



Dichapetalaceae^^ (Geraniales) : Some genera with gamopetalous and 

 zygomorphons flowers of the same family are assignable to the Metaclilamydese. 

 But in other genera with polypetaloiis flowers, it agrees perfectly with the 

 Euphorbiacese — PhyUanthoideae (Geraniales). Its natural position therefore, 

 shiffcs from one place to another, accordiug to the difierent criteria applied. 



Cyrillacese^^ (Sapindales) : This group is related to^he Ericacese in the 

 habit and peculiar anthers of Costcea (Cyrill.). In another respect, it is 

 undeniablly allied to the Aquifoliacese. Hooker regards it as referable to the 

 Anacardiaceee, while Englek looks upon it as a disfcinct family assignable to 

 the Sapindales. 



Corynocarpaceae^' (Sapindales) : This group is referred to the Berberidacese 

 by JussiEU and Speengel, while it is assigned to the Myi-sinacese by others. 

 VanTieghem places it near the Geraniacese. 



Salvadoracese''^ (Sapindales) : This should find its place among the 

 Sapindales near the Celasfcraceae, if we lay great sfcress upon the choripetalous 

 genera, Azima and Dobera. But on the contrary, if fche gamopetalous genera 

 of the same family preponderate, then it should be placed close to the Oleacese. 



Fouquieriaceee^^ (Parietales) : This is related to the Polemoniaceee by tke 

 3-celled ovary, by the more or less connate styles, and by the tube-like corolla 

 with stamens at its base ; bufc, in ofcher characfcers besides those just givan, ifc 

 is quite close to fche Tamaricace». 



Winfceranacepe''^ (Pariefcales) : In fche presence of oil-ceUs in the cortex, 

 pith and leaves, in fche general strucfcure of fche xylera, especially in the 

 tracheids with bordered pits in wood-prosenchyma and in the inconstant 

 number and spiral aiTangement of the perianth-segments, this group comes 

 very close to the MagnoHacese (Ranales). On the other hand, if the connate 



1) Nat. Pfl.-fam. ni.— 4. p. 347. 2) Nat. Pfl.-fam. in.— 5, p. 180. 



3) 1. c. Nachtr. I. p. 2X7 ; Naclitr. IIL p. 197. 4) Nat. Pfl.-fam. IV.— 2, p. 17. 



5) Nat. Pfl.-fam. m.— 6, p. 298, and Nachtr. in. p. 228. 6) Nat. Pfl.-fam. ni— 6, p. 316 . 



