Dr. Arnott on the Genus Torreya. 127 



in the axis, but then divide into two divaricating branches, 

 like the top of the letter T (thus H:H ), which branches be- 

 come involute, and have each a single ovule attached to their 

 extremity. Thus the Torreya belongs to the Corolliflorae, and 

 has four didynamous stamens and an ovarium with four ovules, 

 a circumstance which Sprengel must have entirely overlooked. 

 Of the two orders, Labiatce and Verbenacece, which contain 

 genera with these characters, it is obviously most allied to the 

 latter, on account of the nearly equal lobes of the corolla. But 

 this order (as well as the Labiates) has a four-celled, not uni- 

 locular ovary ; this however does not appear to be of any 

 consequence, and is more a difference of words than of facts ; 

 for, from the proximity of the extremities of the two half dis- 

 sepiments, the parallelism of the divaricating branches, and 

 the fleshy nature of the ovarium, it is highly probable that in 

 a more advanced state the approximate parts would either co- 

 here, or have the interval filled up with a fleshy substance. 

 The ovaria of all the drupaceous Verbenacece which I have 

 examined exhibit the same structure, which is distinctly de- 

 scribed by Roxburgh in his Flora Indica, iii. p. Q 1 ], where he 

 says of Siphonanthus hastatus, "germ four-cornered, seem- 

 ingly four-celled, with one ovulum in each lobe attached to 

 the concave side of the wings of the parietal fungiform recep- 

 tacle meeting in the centre but not uniting, hence seemingly 

 four-celled." The same structure is readily traced in the ma- 

 ture fruit of Clerodendrum Siphonanthus, Br. (Siphonanthus In- 

 dicus, L.), where the shell of each nut or pyrena is not of one 

 solid piece, but is a convolute lamina formed of the inner wall 

 of the drupe, the dissepiment, and one of its divaricating invo- 

 lute branches ; or, to speak more philosophically, it is the con- 

 volute half of one of the two carpellary leaves, thus C7) (Tt> 

 of which the ovary and fruit is composed. ^ < ^~ > 



Sprengel's Torreya, then, belongs to the Verbenacece, and the 

 enlarged calyx, tubular corolla, with an equally five-lobed 

 limb, much protruded stamens and style, bifid and acute 

 stigma, as well as the structure of the ovary, point out its 

 identity with Clerodendrum : it belongs to Mr. Brown's first 

 section, but I have seen no species that agrees with it in the 

 inflorescence and size of the flowers (about half an inch long), 



