of Lythracece and Vochyacece. 213 



Here we find stipules 0, flowers regular, corolla hypogy- 

 nous, stamens hypogynons, and not fewer than the corolline 

 divisions ; carpels free from the calyx. The hypogynous sta- 

 mens are also normal and without exception, through the se- 

 quence JEsculales, Hypericales, and Aurantiales. 



An unsupported resemblance to Violacece has been claimed 

 for Vochyacece, but it is rebutted by the characters of the Vi- 

 olates, in which, among other points, the ovules are many, 

 and albumen rarely absent. The coincidences with Violates 

 lie almost wholly with the deviations of the different families 

 from the normal structure of that alliance. 



Were we to suppose the Violacece in affinity with the Po- 

 ly galacece, the case for a joint affinity towards Vochyacece 

 might be stronger, but this seems now to be surrendered. — 

 The affinity of Polygalacece to Violacece is however much clo- 

 ser than the supposed affinity to Sapindacece among the JEs- 

 culales. Von Marthas, in the artificial arrangement of his 

 Conspectus, places Vochyacece near Sapindacece and other 

 families among the JEsculales and Euphorbiales : — 



jESCULALUS. Leaves petioled; stipules rarely present, and then 

 small, (some Malpighiacece). Sepals in a broken series, imbricate, odd se- 

 pal superior. Torus discoid. Petals, (if present), issuing from the edge of 

 the disk, not connate with each other, deciduous. Stamens hypogynous, 

 filaments subulate or flat, (Milling tonieee), in a single or double row ; an- 

 thers introrse, bilocular, cells parallel, not opening by pores. Carpels 2, or 

 3, or 4, forming independent cells, free from the calyx,, connate with each 

 other. Stigmas not sessile. Trophosperms central ; ovules definite. Albu- 

 men 0, or extremely thin, (Millingtoniece) ; radicle next the hilum. 



Polygalacece cannot be there ; and I at last agree with 

 Agardh's "Nullibi alias collocandce" in placing this most a- 

 nomalous family near the Brassicales. The affinity to Rese- 

 dacece seems to be admitted ; neither can I throw Treman- 

 driacece to the Rhamnales, which are Micranths, nor overlook 

 their passage to Capparidece- With all these Vochyacece has 

 no relation, nor does it shew any tendency to parietal or sub- 

 parietal placentation. The alliance of Polygalacece not being 

 definitely fixed, the test so far fails ; but in Polygalacece there 

 are no stipules ; the stamens are adelphous, hypogynous, and 

 not facing the petals ; ovary free from the calyx; ovules pen- 

 dulous ; albumen usually present. 



Whatever may ultimately be the place of Vochyacece, it will 

 be nearest to its recognised relations, if stationed among CE- 

 notherales, not far from CEnotheracece, (Lopezia), Rhizopho- 

 racece, Combretacece, and Memecylacece, to which Reichen- 

 bach adds Chrysobalanacece. To these we may also add 

 Lythracece, which thus give mutual support to each other ; 

 there is the same tendency to tetragonal branches j Cuphea 



s4 



