Plants in their Relation to Others. 5 1 1 
nexions, but that statement does not exclude the possibility that some few 
may have been oceanic, such for instance as the Sandwich Islands, or 
St. Helena. These far outlying islands have so few endemic genera that 
they do not appreciably affect the totals. 
Confirmatory evidence may be obtained in various ways. One may, for 
example, point out that since many families have been long enough upon 
islands to have given rise there to endemic genera, these families must be 
very old. But now, to have reached both the Old and the New Worlds in 
the course of its dispersal, a family must also be very old. One will there- 
fore expect that a very great proportion of those families that are repre- 
sented upon islands by endemic genera will also reach both worlds. 
Examining the facts, one finds that 5 families with endemic genera upon 
islands are themselves endemic there. Of the remaining 145, no fewer than 
1 31 (90 per cent.) reach both Old and New Worlds, while of the other 142 
families that have no endemic genera upon islands, only 75 (52 per cent.) 
occur in both land masses. 
Or again, on the whole the islands that are farthest out from the main- 
land should have almost solely the very oldest families (of the nearest 
mainland) in their various affinity circles, and we have seen that these are 
usually also the largest families. Thus, on the whole, the endemic genera 
of the far out islands should belong to larger families, averaging larger than 
the islands as a whole, though these should average larger than those of the 
mainland. We may test this on the flora of Madagascar. This island has 
endemic genera in 80 families, while the islands of the world have them in 
150. Of the 70 families thus missing, no fewer than 57 belong to the 75 
smaller, and only 13 to the 75 larger families. In New Zealand the pro- 
portion of families with endemic genera is 16 in the first 75 larger to 4 in 
the 75 smaller families, while in the Sandwich Islands the proportion is 13 
to 1. The prediction is thus fully borne out, and it is confirmed by the 
fact that the bulk of the small families occur only on the nearer islands. 
Or reference may be made to the comparison between the floras of New 
Zealand and the outlying Chatham Islands ( 22 , p. 287 ). 
If the island endemic genera were really survivals, they should, in virtue 
of their total number of 1,582, or 12*6 per cent, of the genera of the world, 
contain at least that percentage of the monotypic families, but in fact, of the 
55 such families, they contain only 3. Of the other 52, 12 are found in 
both hemispheres, 6 are palaeotropical, 6 neotropical, 6 in Africa, and 
smaller numbers elsewhere. 
The oldest families of all will in general have reached most islands. 
One will therefore expect the bulk of the endemic genera to belong to 
families that reach many islands. This is easily found to be the case, so 
there is no need to go into detail, but give simply one instance. The West 
Indies have 195 endemic genera in 43 families that also occur in the islands 
L 1 2 
