240 | Dr, J. D. Hooxzn on the Vegetation 
nigh levels, where the tropical proportion no longer obtains, it is probable 
that Mr. Brown’s results are the more accurate. If the Galapagos number of 
Monocotyledones is small as compared with that of the continent, it is even 
more so with regard to that of other tropical islands: thus in St. Helena they 
equal nearly $ of the Dicotyledones, in the Society Islands 12, and in the 
Sandwich Islands 3. This paucity is not due to the sterility of the soil or 
dryness of the climate, for the Cape Verd proportion is 1, these islets being, as 
I have mentioned above, equally barren with the Galapagos. 
Although I can offer no explanation of this apparent anomaly, it may not 
be out of place to notice here, that the tropical islands in general possess 
proportionally more Monocotyledones than do the continents. This is no 
doubt due to the same causes which determine the increased proportion which 
the temperate zone shows over the tropical, and is the more striking from this 
circumstance, that the nearer the tropical islet is to a great continent, the 
greater does the proportional number of Dicotyledones become, as is shown by 
the Galapagos and Cape Verds, both adjacent to great continents, possessing 
more than the Sandwich Islands, Society group, St. Helena, or Ascension. 
The individual species of the very largest Monocotyledonous families being . 
more widely dispersed than any of equal extent amongst the Dicotyledonous, 
is also a reason why the insular proportion of the latter should be different 
from the continental *, 
If the insular tropical proportion be assumed to be 4, it is larger than that for 
the islands of the Atlantic immediately beyond the tropics. Thus for the Cana- 
ries (lat. 28°) itis probably 1, and for Madeira 3$ ; whence they again increase 
* And for the same reason great caution is required in deducing the continental proportions from . 
small local collections, 
almost necessarily erroneous if applied to the greater one of which it forms a part, although the general 
features of vegetation may be well displayed by the latter. To this I attribute the remarkable dis- 
