PORTULACEJE, AND THEIR ALLIES. 57 
tulacee, Caryophyllacee and Phytolaecacez in one cohors among 
his Dialypetale, and Chenopodiaces, Amarantacee, Polygonee 
and Nyctaginacee in another cohors far away among Apetale. All, 
however, are ready to suggest that in a really natural system all 
the above orders ought to be brought together, which cannot be 
done without entirely rejecting the above-mentioned great Candol- 
lean classes ; yet no substitute has been proposed for these classes, 
except a vain endeavour so to modify the linear series as to bring 
allied orders into close approximation. Thus Grisebach, one of 
the most able advocates for this arrangement (which, with any 
one for whose views we had less respect, we should be tempted to 
call a disarrangement), brings indeed all our Curvembryonous 
orders together, but places them between Euphorbiacee and Mal- 
vacex, which in our view have quite as much right to be placed in 
close proximity as Caryophyllee and Chenopodiee *. 
In considering how to deal with these various proposals, we 
must observe that none of the classes, groups, or alliances so 
formed are limited by any character that does not undergo many 
exceptions among the genera placed under them; nor are we able 
to devise any other that shall be thus strictly and absolutely de- 
fined. Even the curvature of the embryo round a farinaceous 
albumen, the chief character of the whole group, can scarcely be 
traced in Dianthus, in some Polycarpe@, in Anacampseros and its 
allies, in some Polygonee, &c.; and the position of the leaves, 
the presence or absence of stipules and petals, the number and 
insertion of the stamens, the relative position of sepals, petals, 
stamens and carpels, the degree of combination or reduction of 
the carpels and ovules, are characters so variously combined or 
dissevered, as always to leave small anomalous genera invalidating 
or uniting any groups we can form. Our object has therefore 
been to seek out such limitations as may bring together genera 
having the greatest general resemblance, and united by such 
tangible characters as should have the fewest exceptions. 
Our first great group is that of the CARYOPHYLLEE, the normal 
characters of which (besides those common to all the above orders) 
are opposite leaves ; sepals, petals, and one or usually two series of 
stamens, all isomerous ; a free one-celled ovary with several ovules 
in the centre, and formed by the combination of two or more car- 
* When in this and other similar papers 1 make use of the plural we, with 
reference to any general views on the principles of distribution and limitation of 
genera, I refer to those of Dr. Hooker and myself as adopted for the * Genera 
Plantarum we are preparing. 
