44 Willis.— The Flora of Stewart Island ( New Zealand ) ; 
This result again emphasizes the fact that the endemics occur almost 
only in the very largest families (the average size of a family in New 
Zealand is 15 species), i. e., as we have explained, the families which are 
on the whole the oldest in New Zealand. If we add together the two 
lists of endemics, we find that 17 occur in the first seven families, and 
only 2 in the second eight, while the 76 smaller families contain none 
whatever. 
The addition of these nine to the list of endemics of Stewart also 
confirms prediction 8, for the genera to which they belong, 9 in all, contain 
249 species, or an average of 27 each, a slightly higher figure than that 
given for the genera containing the first ten endemics. 
The other predictions given also remain as stated, with slight numerical 
corrections. 
Nothing, therefore, in this great addition to the species found on 
Stewart Island compels any modification or real revision of the general 
thesis of this paper, and in this respect it may be compared with the 
revision of the figures for Ceylon which I have elsewhere published (11), 
and which also showed no differences in the final results. 
Summary. 
The flora of Stewart Island is dealt with in the light of age and area as 
regards its taxonomic distribution. Numerous predictions are made as to 
what should be expected under that hypothesis, and verification is made 
upon the facts, all the predictions proving to be correct. 
An attempt is made to predict the actual composition of the flora 
from what is known of the floras of the Kermadecs (1,200 miles away, at 
the other end of New Zealand), Chathams, and*Aucklands, and 139 species 
of Angiosperms out of 383 are correctly predicted, and no less than 46 out 
of 62 ferns (74 per cent.), while only 18 species are predicted that have not 
been found. 
Verification of the other predictions shows that— 
Stewart has a greater proportionate representation of families than 
of genera, and of genera than of species, as compared with New 
Zealand. 
Families are represented in proportion to the number of genera 
contained in them (in New Zealand). 
Genera are represented in proportion to the number of species in 
them (in New Zealand). 
Wides (the oldest forms) are best represented in the flora of 
Stewart. 
The plants of Stewart are very widely spread in New Zealand, 
about twice as widely as the average. 
The proportional representation of ferns is greater. 
