274 Willis. — The Floras of the Chet lying Islands of 
former, or 67 per cent., occur in the islands, while of the latter only 26 
(the genera in capitals) occur, or 17 per cent., an enormous difference. The 
prophecy made above is thus fully borne out by the facts. 
Further, of the 26, it occur only in New Zealand and the Kermadecs, 
7 in the Chathams, 7 in both Kermadecs and Chathams, and one only 
(Rostkovia) in the Aucklands and the South Island of New Zealand. 
This was probably a late arrival of the southern invasion. 15 of the 
26 genera belong to families enumerated in the northern invasion (10, 
p. 35b), and all but Rostkovia are distinctly northern types, most of which 
probably took part in the northern invasions. 
(9) Just as with the families, we may predict that those genera will 
on the whole be the largest in number of species, whether in New Zealand 
as a whole, or in Stewart Island, which reach three of the outlying groups 
of islands. Then will follow those reaching two islands, one island, and 
no islands at all. Testing this we find : 
Table VIII. 
Reaching 
Gen. 
Spp. in Stewart. 
Average. 
Spp. in N.Z. 
Average. 
Stewart and 3 groups 
17 
78 
4*5 
2 93 
17.2 
>> 2 >j 
39 
132 
3*3 
396 
IO-I 
„ „ 1 group 
58 
99 
i -7 
294 
5.0 
)> » n o „ 
55 
74 
i-3 
162 
2.9 
Neither Stewart nor other 
islands 
J 33 
204 
i *5 
Other islands, but not 
Stewart 
27 
43 
i*5 
(10) Just as with the families, one may predict that the islands will 
have proportionately more genera in common with one another than they 
have with New Zealand. 
Table IX. 
Kermadecs. 
Chatha?ns. 
Aucklands. 
New Zealand. 
Kermadecs in common with 
— 
32 
21 
62 ) 
* 
32 % 
30 % 
*9% i 
Chathams „ ,, „ 
3 2 
— 
44 
98 \ 
5i% 
64% 
29% 1 
Aucklands ,, ,, ,, 
21 
44 
■ — 
68 J 
33 % 
44% 
20% 
Again the facts bear out the prophecy. 
(11) One may predict that the genera which occur in Stewart and do 
not occur in the islands will on the whole be the smaller genera of Stewart. 
The average number of species in a genus in Stewart is 2*2, and of the 
55 genera that do not occur in the islands 44 have one species in Stewart 
Island, six have 2, and the others 3, 3, 3, 4, and 5, or an average of 1-3 
species per genus. 
(12) We may now go on to the actual species, and predict that these 
will also be largely selected from the Stewart list. It would occupy too 
