

118 MH. G. liEXTIIAAI OX MTrtTACE^, 



liferae, Cornacca?, and liubiacea?, although constant in species, k 

 seldom, if ever, available for generic distinction. 



The number of ovules in each cell of the ovary is useful to attend 

 to only where there are one, two, or few, as it is otherwise vari- 

 able, even in species ; but placentatiou and the mode of insertion 

 of the ovules is, within certain limits, and especially in the cap- 

 sular tribes, one of the most constant generic characters in the 

 Order, and only recently much attended to. It exercises some 

 patience in its investigation, but it vv^cll repays the trouble, and has 

 the advantage of beii)g supplied by most herbarium specimens. 

 As a general rule the character of the order is to have anatro- 

 pous, or more or less amphitropons ovules, attached in two or 

 more rows to an axile placenta; but the minor modifications are 

 innumerable. The placenta itself is attached to the middle of the 

 axis, or adnate to nearly its whole length, or arises from near or 

 from quite the base, or descends from the upper angle of the cell. 

 In many (if not in all) Cham?elauciere, the 1-cclled ovary beii^ 

 probably monocarpellary, the placenta is moi*e or less approximate 

 or adnate to one side of the cavity. 



Oalorma and HJiodamiua become 1-celled from the non-develop- 

 ment of the placenta-bearing dissepiments ; and the placentas are 

 almost basal in Oslorma^'parietal in Hhodainnia. The ovules, more 

 or less curved, or nearly straight, and more or less amphitropous 

 in the great majority of Myrtese, as in many genera of Lepto- 

 spcrmcae and Lccythidca^, become nearly anatropous and descend- 

 ing (unless when closely packed) in Leptosjyermitm^ Kunzea, Tris- 

 taniay HacWiousiay and Grias^ ascendiug (unless w^hen closely 

 packed) in Melaleuca^ Adonis, Phj/matocarpiiSy Calothammts, Ere- 

 mda^ Sijncarpus, Lysicarpus^ Cloezia, T('picalia,aiid Coiiratdri. The 

 laterally attached ovules are usually superposed in two or rarely 

 four vertical rows, which are parallel when the placenta Is narrow ; 

 but in Bceckea and some allied genera the placenta is sometimes 

 dilated, assumes more or less of a peltate form, and the two rows 

 of ovules form a regular ring round its margin, and in Xantno- 

 stemon {Frewj/a^ Brongn. and Gris) the placenta elongates with 

 the ring of ovules below its apex. AVhere the ovules are aua- 

 tropous or nearly so, as well as in a few genera where they are 

 very numerous, although more or less amphitropous, they often 

 occupy the whole surface of the placenta without forming distinct 

 rows. In these cases, when closely packed, they are mostly 

 horizontal, and the tendency to ascend or to descend, as above dis- 

 tinguished, may be difficult to ascertain. In Beavfortia there i^ 



