OBOLARIA VIRGINICA. 29 



from the relation borne by what are called the sepals of the solitary 

 flower to the bracts of the cluster, as well as from their want of 

 agreement in number with the divisions of the corolla. But, on the 

 other hand, the Gentian family, with which Obolaria is to be par- 

 ticularly compared, presents one of these same exceptional cases ; 

 their carpels being commonly (although, as in Obolaria itself, not 

 uniformly) right and left of the axis, and consequently opposite 

 the lateral sepals. 



It is now evident, that the only natural orders to which Obolaria 

 has any obvious relationship are those two to which it has been va- 

 riously referred, namely, Gentianacea; and Orobanchacese. The 

 smgular introflexions or processes analogous to the double (or 

 rather separated) placentae of some Orobanchaces would seem to 

 favor its alliance with that order, in which, perhaps, the distribution 

 of the ovules over the whole surface of the cell might the rather 

 be expected. The (commonly) lateral position of the carpels 

 would favor the association of Obolaria with Orobanchacese as 

 much as with Gentianaceae, if Lindley and Endlicher have cor- 

 rectly attributed this character to the former order. But against 

 this is the much higher authority of Brown, who believes the 

 carpels to be anterior and posterior in Orobanchefe.* The wei<^ht 

 of this character, therefore, foils in favor of GentianaceEe. The 

 regular corolla, equal and isomerous stamens, and opposite leaves, 

 are also points of difference from Orobanchaceje, and of agree- 

 ment with Gentianacese ; to which may be added the green foli- 

 age and terrestrial growth, which would be altogether anomalous 

 in the former family (although, on the other hand, one Gentiana- 

 ceous genus is parasitic). The only point essentially at variance 



* Planta Jman. Rariores, p. 1 12, note. — The carpels are certainly anterior 

 and posterior in Epiphegus. 



