wir 
1-seeded carpels, may, it is suggested, have been derived from 
the Helleboroideae ; or more likely it may have had a separate 
origin from some (at present) unknown group. At any rate 
the Helleboroideae approaches the Ranunculoideae closely in 
the neighbourhood of the genera Isopyrum and Enemion, wherein 
the ovules are sometimes reduced to a very small number. elow 
is given a diagram showing the approximate relationships of 
Sg principal groups and linking genera of the family Ranuncu- 
Bontham & Hooker’s classification began with the genus 
Clematis and ended with Paeonia. We now reverse this arrange- 
ment. Tribes Paeonieae, Helleboreae and Clematideae remain 
more or less as defined by Bentham & Hooker. As noted above, 
genera Glaucidium and ‘H ydrastis in the tribe Paeonieue, deter- 
mined mainly on characters derived from the relative lengths 
of the outer and inner seed coats. I think this feature has been 
much over-emphasized and in this case has led to an unnatural 
collection of very different types. Paeonieae is therefore treated 
here as being generically monotypic. Worsdell,* mainly on 
anatomical grounds, considers it to be worthy of taking family 
rank, From a taxonomic point of view, however, I think this 
is scarcely practicable. For similar reasons Engler’s removal 
of Hydrastis to Berberidaceae is not acceptable. Its numerous 
carpels and longitudinally dehiscent anthers seem to destroy 
entirely the homogeneity of that family (I exclude from 
Berberidaceae the Lardizabalaceae) and it is much better placed 
in the Ranunculaceae, to which most previous authorities have 
referred it. I have ventured to recast tribes Ranunculeae and 
Anemoneae, separated by Bentham & Hooker on the position of 
the ovule and merged by Prantl into one tribe, on more natural 
lines, determined by the absence or presence of the involucre of 
leaves below the flower, so constant a characteristic (associated 
with apetaly) of the large and natural genus Anemone. For this 
reason the genera T'halictrum, Knowltonia, Adonis, Callianthemum 
and Myosurus are transferred from Anemoneae to Ranunculeae, 
to which they seem more closely related by characters other than 
the position of the ovule. The genus Syndesmon links these two 
tribes just as Clematopsis (Kew Bull. 1920, 12) connects Anemoneae 
through Anemone § Pulsatilla with Clematideae. Delphineae, 
hitherto regarded as a subtribe of Helleboreae, is here raised to 
tribal rank. In a primitive group like the Ranunculaceae the 
remarkably specialised and advanced type of floral structure 
shown in the zygomorphic flowered Delphinium and Aconitum 
should be regarded as a character of primary importance in 
classification. The seeds, too, are very distinctive in the group. 
The most highly specialised types of perianth occur, then, 
in the genera Aquilegia, Aconitum and Delphinium, whilst the 
* W. rsdell, ‘‘ The Affinities of Paeoni . Bot. 
114-116 (1908); see also the same author on WAS Mears pigs the Vasculas 
System in certain Orders of the Ranales ” in Ann. Bot. 22: 663 (1908). 
