Plants in their Relation to Others. 509 
islands like Madagascar, New Zealand, or Borneo, must have been connected 
with the mainland at the time when they received the bulk of their flora, 
though the connexion need not have been absolutely continuous, but might 
have been interrupted! by narrow straits. And it is also, I think, clear in 
the second place that endemic genera, with few exceptions, are really young 
genera in the earlier stages of dispersal. Just as it was all but inconceivable 
that endemic species, which showed as ‘ wheels within wheels ’, as in the 
map of Ranunculus in New Zealand given above, should be dying out in so 
regular a manner, so here it is extraordinarily hard to imagine that the 
genera can be survivals, and yet appear in a regular proportion to the 
totals of genera in their various groups. 
Tens of Fams. 
World. 
O/ 
Austr . &c. 
O/ 
W. Austr. &c. 
Islands. 
O/ 
1. 
/o 
5 j OT 9 
40.1 
/o 
1,579 
39 - 1 
459 * 
r 0 
4°*5 
606 
/o 
38-3 
2. 
1,868 
14.9 
592 
14.6 
176 
I 5*5 
285 
18.0 
3 - 
I , c 94 
874 
8.7 
360 
8.9 
86 
7.6 
r 44 
9.1 
4 - 
6.9 
325 
8.0 
78 
6-8 
115 
7.2 
5 - 
695 
5-5 
271 
67 
75 
6-6 
83 
5.2 
6. 
561 
4.4 
216 
5*3 
57 
5.0 
82 
5-i 
7 * 
456 
3*6 
83 
2-0 
x 9 
i-6 
55 
3-4 
8. 
355 
2.8 
hi 
2-7 
30 
2.6 
48 
3-o 
9 - 
296 
2-3 
99 
2.4 
24 
2-1 
29 
1.8 
10. 
233 
1.8 
79 
1-9 
29 
2-5 
37 
2-3 
Total 
ii 3 45i 
91.4 
3,7i5 
92.I 
1,033 
91. 1 
1,484 
93*7 
11 to 20 
9 X 9 
7-3 
278 
6-8 
90 
7*9 
86 
5*4 
21 to 29-2 
T 47 
i-i 
38 
0.94 
10 
0.88 
12 
o -75 
Grand Total 
12,517 
4,031 
I , I 33 
1,582 
The percentages may be plotted as curves (Diagram 6, p. 510), when 
their agreement is seen in a still more striking manner. 
These curves are very remarkable in their close coincidence, and after 
sight of them it is difficult any longer to maintain the position that endemic 
genera in general are survivals of old floras. Survivals are not likely to be 
nicely graduated in proportion to the size of the groups of families to which 
they belong. 
The first and principal part of the prediction is thus fully borne out, and 
comparison shows with equal clearness that the proportional representation 
of the different families among the endemic genera of islands decreases as 
one goes down the scale. There are 292 families of Spermaphyta in the 
world, and the first 100 of these have island endemic genera in 92, the 
genera being 12-9 per cent, of the total genera in the families. The inter- 
mediate 92 families are represented in the islands by 45 only, with 9*28 per 
cent, of their genera, while the final 100 are represented only by 13 
(4 families endemic) with 8*72 per cent, of their genera. This bears out the 
second part of the prediction, whose complete success may be taken to mean 
that in general, with few and comparatively insignificant exceptions, the 
islands were united to the mainland at the time they received their floras, 
