TRICEROS. 607 
simis. Pet. longissim. acuminatis, discus styloram basin con- 
jungens planus carnos. Styli 3 sepalis oppositi, duobus in- 
curvis simplicibus. Stigmatibus 2 setaceis recurvis. 
.. Fructus late clavatus, apice quasi truncat. cornubus tribus 
(estylis induratis) recurvis, capsularis, extus venosus, in- 
terveniis celluloso-reticulatis. 
Semina 2 pendula, 3tio abortient. oblongo-quadrat. com- 
pressiuscule utrinque rapheos margine clavato rugoso. 
Raphe completa, simplex? Testa coriaceo chartaceo. Em- 
bryo ordinis. 
Sepalis evasculo-cellulosis. Pet. vasculo-centrale complete, 
lateral. incomplete dimidiatis. 
Planta volubilis cirrhifera, habitu Cis, fol. carnosiusculis 
pedalis, foliolis angustiuscule dentatis lanceolatis. Racemis 
flor. fem. nutant. Ducis: bracteis linearibus subsimplici- 
bus, flor. masculor. longe et anguste epp 
Stam rudim 0, in flora femineo. Styli 2, 1 sepalis 4 petalis 
.. 9pposit. 3tio petalo omnino opposito. 
. . Oss. On what principle is an anther like thia explainable ? 
- The bilocillate form is riducible to the genuine bilocular (qua- 
. drilocellate) by assuming the secondary septa to be deficient. 
And the only manner I know to reduce the genuine uni- 
. locular anther to the ordinary type, is to assume the anthe- 
. Tal leaflet margins as entirely coalescent. 
The true modification of anthers requires examination. It 
48 at once obvious, that a quadrilocellate anther cannot be well 
*Xplained if we assume the lines of dehiscence as correspond- 
ing to the lines of inflection of the margins of the leaf. To 
explain them in that way, 4 leaflets are required for each 
and all on the same plane, which is contrary to analogy. 
_ The septa are the last things formed probably y having a 
3 different origin and analogy. 
