85 
either naked or crested ; sometimes 3-lobed, and then destitute of a crest. Stamens hypo- 
gynous, 8, usually combined in a tube, unequal, and ascending ; sometimes 4, and distinct ; 
the tube split opposite the upper sepal; anthers clavate, innate, mostly 1 -celled and open- 
ing at their apex, sometimes 2-celled; very rarely the dehiscence is longitudinal. Disk 
either absent or present, regular or irregular. Ovary superior, compressed, with 2 or 3 
cells, which are anterior and posterior, the upper one occasionally suppressed ; ovules soli- 
tary, very rarely twin, pendulous ; style simple, curved, sometimes very obHque and cucul- 
late at the apex, which is also entire or lobed ; stigma simple. Fruit usually opening 
through the valves ; occasionally indehiscent, membranous, fleshy, coriaceous, or drupa- 
ceous, winged or apterous. Seeds pendulous, with a caruncula next the hilum, naked or 
enveloped with hairs ; the outer integument crustaceOus, the inner membranous ; albumen 
abundant, fleshy, rarely reduced to a thin gelatinous plate, very seldom wanting ; embryo 
straight, or slightly curved, with the radicle next the hilum. — Shrubs or herbaceous plants. 
Leaves generally alternate, sometimes opposite, mostly simple, and always destitute of sti- 
pules. Flowers usually racemose, very often small and inconspicuous, but showy in many 
Polygalas. Pedicels with 3 bracts. 
Anomalies^ Flowers generally gamopetalous. Ovary sometimes 1 -celled by abortion. 
Fruit indehiscent in Mundia, Monnina, Securidaca, and Krameria, The latter has also no 
albumen. 
Affinities. The structure of this order has been explained by Aug. de 
St. Hilaire and Moquin-Tandon, from whose memoir above quoted, the fore- 
going character and almost all that is said here is extracted, and to which I 
refer those readers who wish to study the subject more intimately. Before 
adverting to the affinities of this order, it will be useful to consider what is the 
nature of the irregularity of the flowers ; an irregularity which is such as to 
obscure, in a great measure, the relative position of the sepals and petals. The 
calyx apparently consists of but three pieces, which are usually green, and like 
sepals in their common state ; but their real number is 5, the two coloured 
lateral petal-like bodies, sometimes lying within the apparent sepals, being in 
reality part of the series of the calyx. The corolla is mostly monopetalous, 
and, if carefully examined, formed of 3 pieces ; namely, the keel and two pe- 
tals, all soldered together. We have, therefore, an abortion of two petals, 
according to the laws of alternation : but this is not all ; there is not only an 
abortion of two petals, but of those two which would, if present, be found 
right and left of the keel. The monopetalous corolla is, therefore, formed by 
the cohesion of the two posterior and the one anterior petal of a pentapetalous 
corolla, of which the two lateral petals are suppressed. The keel has an ap- 
pendage of an anomalous character, called technically a crest, and often con- 
sisting of one or even two rows of fringes or divisions, originating not from 
the margin but from within it, and sometimes cohering in a common mem- 
brane at their base. Aug. de St. Hilaire has shewn that this crest is nothing 
more than the deeply-lobed middle segment of a keel, with these lobes in such 
a state of cohesion that the central lobe is pushed outwards, while the lateral 
ones cohere by their own margins and with its back. The stamens are only 
8, two therefore are suppressed ; or in Krameria 4, one being suppressed. I 
may remark, in addition, that the relative position of the flfth sepal and petal 
respectively, was first indicated by Brown. Denham 31. 
Polygalaceee are stationed by De Candolle between Droseraceae and Tre- 
mandraceae, and in the immediate vicinity of Violaceae. With the latter they 
are related on account of their hypogynous stamens, irregular flowers, and cu- 
cullate stigma ; and with Tremandraceae on account of the caruncula of their 
seed. To Fumarieae they approach in the general aspect of their flowers ; 
but if my theory of the structure of that order be admitted, their resemblance 
would not be so great as it appears to be. Leguminosae are, notwithstanding 
their perigynous stamens, an order with which Polygalaceae have great affinity ; 
the irregularity of coroUa is of a similar nature in both ; there is in I^egumi- 
nosae a tendency to suppress the upper lateral petals in Erythrina, as in Po^ 
