Ranales , Rhoeadales, and Rosales . 
729 
habit is primitive for Angiosperms, he may be said to be in favour of 
tetrarchy in general as being more primitive than diarchy. Miss Sargant 1 
in 1903 presented a well-reasoned statement in support of her view that the 
varieties of seedling anatomy met with in the Monocotyledons were derived 
from a tetrarch condition, and pointed out that there were suggestions of 
this condition in the Ranunculaceae, which also furnished so many other 
points of similarity with the Monocotyledons. I agree with Mr. Tansley, 
who pointed this out in his criticism of the hypothesis that the elements 
instanced in Eranthis in support of the presence of tetrarchy are probably 
secondary. Nevertheless the work of the last ten years has shown that the 
tetrarch condition is to be found very widely in the Dicotyledons, forming 
about 50 per cent, of the Leguminosae and about the same of all known 
forms of the Rosales. Similar very rough calculations reveal some 30 per 
cent, in the Cactaceae and in the Compositae, 15 per cent, in the Ranales, 
11 per cent, in the Tubiflorae, 6 per cent, in the Centrospermae, while none 
at all have been recorded either in the Piperales or in Rhoeadales. 
As the total percentage of tetrarch forms recorded in these groups 
taken together is about 30 per cent., we see that the Cactaceae and Com- 
positae present about the average distribution of the types, the Rosales 
have an excess of tetrarch forms, the Ranales, Tubiflorae, and Centrospermae 
an excess of diarch, while the Piperales and Rhoeadales so far as known are 
exclusively diarch. 
These figures of course count for very little in view of the comparatively 
few forms described (about five hundred in the cohorts enumerated) and 
their unequal representation of groups, due to the practical difficulties of 
obtaining the most desirable forms. Thus the preponderance of Ranuncu- 
laceae in the Ranalean species recorded in this paper and the comparative 
fewness of the representatives of the other families probably account for 
the low percentage of tetrarch forms. 
The existence of the tetrarch condition within the Ranales — as recorded 
in this paper — as well as the prevalence of the type in other parts of the 
Natural System, notably in the Rosales, removes any special importance 
that may seem to have attached to Eranthis and places the connexion 
between the seedling anatomy of Monocotyledons and Dicotyledons on 
a broader basis. Furthermore, the ‘ Anemarrhena ’ variety of the tetrarch 
form is quite common in Dicotyledons (witness Berberis Lycium and 
B. aristata , Lauras, Pr units Cerasus, Pyrus Malus , Medic ago, &c., described 
in this paper), and no special significance such as has been claimed by 
de Fraine and Lee attaches to its existence in the Cactaceae and Tubi- 
florae. 
In addition to these definitely diarch and tetrarch forms there are 
a number of intermediate conditions recognized as such by Compton in the 
1 Loc. cit. 
