208 TLORA INDICA. [SabiacecB. 



The cohesion of the carpels in Sabiacece is so very slight, even in the ovary, and dis- 

 appears so rapidly as the fruit advances towards maturity, that the connection is proba- 

 bly chiefly vrith apocarjjous orders. Blume and Miers, as we have seen, place the Order 

 in the immediate neighbourhood of Menispej-macets, indicating at the same time an 

 affinity with Lardizabalacea. To us it appears intermediate between Schieandra- 

 cetB and Menispermaceie, agreeing with the former in the subscandent habit, in the 

 persistence of the bud-scales at the base of the branches, in the synchronous evolu- 

 tion of flowers and leaves from the same buds, the dotted flowers, two-celled ovaries, 

 and the amphitropous or campylotropous ovules, and with the latter in the oblique 

 development of the ovary, by which the style becomes basilar, and the drupaceous 

 fruit, and diilering from the ordinary structure of both in the pentamerous flowers, 

 in the opposition of the sepals and petals, the presence of a disc, the partial cohe- 

 sion of the ovaries and styles, the inferior radicle, and the exalbuminous seeds. The 

 last character, however, is present iu some 3fenispermacece. 



The quinary arrangement of the flowers at first sight appears a great obstacle to 

 the association of Sabiacece with Menispermacece or Schizandraoeee ; but this diffi- 

 culty loses much of its force in consequence of the occurrence of pentamerous flowers 

 in Odontocari/a in the one Order and in Schizandra in the other. Odontocarya^ 

 from Mr. Miers' description and drawing, appears to have many points in common 

 with Sabia, and deviates considerably from the normal structure of the Order to 

 which he has referred it. ScMzandra has been well illustrated by Dr. Asa Gray, 

 who has shown that, though the number of stamens is always five, the petals and 

 sepals vary from five to sir. 



The frequent transition from trimerous to pentamerous flowers in certain genera 

 of Ranunculaceee, and the close affinity of Banunculacece, which are usually penta- 

 merous, to Berieridem, which are always trimerous or tetramerous, tend still further 

 to weaken the force of this objection. It may be observed that the transition is 

 usually from trimerous flowers arranged in three or more rows, to pentamerous 

 flowers in two rows only. This is also the case with the similar transition in Pohj- 

 gonacece, in which Order some genera have pentamerous flowers in a single series, 

 while others have trimerous flowers in a double verticil. An exception, however, 

 occurs in Helleborus and some other pentamerous Eanunciilaceiey in which the petals 

 are about twice (or three times) as numerous as the sepals. 



The most remarkable character of Saiiacea is undoubtedly the opposition of the 

 sepals and petals, because the alternation of succeeding verticils both of leaves and 

 flowers is so universal, that any exception has come to be regarded as next to impos- 

 sible. To this rule, indeed, we believe it will be found that Sabia offers rather an 

 apparent than a real exception ; for though the opposition of each member of the two 

 verticils is very evident, we believe the explanation to be tliat a portion only of the 

 outer verticil belongs to the calyx, the two outer segments being lateral bractiets. 



In all the species of Sabia which we have examined, a single anterior bract is 

 found usually in close contact with the calyx. The two lateral sepals (as they are 

 usually termed) arc exterior in aestivation, and are in most of the species a little 

 longer and broader than the three inner sepals. The estivation of the petals is the 

 same as that of the sepals, the two lateral being exterior, one anterior, and two pos- 

 terior, interior and overlapping each other by one margin. 



A very similar structure exists in KeliantheTrium^ but the small size of the two 

 bract-like lateral sepals, or more properly bracteoles, and the great breadth of the 

 three inner sepals, prevent the opposition of the two verticils from being so decided 

 as in Sabia. In Oistus, where the lateral bractiets are wanting, no evident relation 

 can be traced between the position of the sepals and petals. In Amaranthacece, 

 where the calyx is usually very much imbricated, the struclure is possibly analogous, 

 as is indicated by the reduction of the number of sepals to three in several species 

 of Amaranthus. 



Sabia is entirely an Indian genus. Blume indicates three species, aU seemingly quite 

 distiuct from those described below, so that the number known amounts to ten. The 

 dehiscence of the anthers i-equires to be observed carefully iu the living plant. In 



