32 Willis . — The Flora of Stewart Island (New Zealand): 
Cook’s Strait, and which include two wides, so that there really only remain 
1 6, in classes 7 to 9. Of the 5 in class 9, 4, 1 Gunner a Hamiltoni , Olearia 
angustifolia , Veronica amabilis , and especially Atropis novae zealandiae , are 
more or less closely confined to the coast, and I must leave it to the 
ecologists to settle whether they can be in any way regarded as relicts, or 
whether they are really coast species which arrived (or evolved) late, and by 
water carriage. 
In any case, however, these last 16 species belong all of them to fairly 
large (old) genera (i. e. large in number of species in New Zealand). The 
genera of the 5 in class 9 contain in New Zealand 147 species, an average of 
almost 30 ; those of the 7 in class 8 contain 140 species, or an average of 
20 ; those of class 7 are 4 with 38 species, or an average of 9. All are 
much above the average size of genera in New Zealand (4*2 species). It 
is obvious that these species belong to the same class of genera (the larger 
and older) as do the actual endemics of Stewart itself, which are dealt with 
below (10 genera, 249 species). As one approaches the centre of New 
Zealand one finds endemics in increasing numbers belonging to smaller and 
smaller genera, which, as we have seen above, are on the whole younger and 
younger genera. If these are relicts, then it is clear that the younger 
and smaller (in the country) a genus is, the greater its chance of giving rise 
to relicts ; but the very awkward problem at once crops up, to explain why 
the relicts are crowded together in the centre of the main country of New 
Zealand, where there are also the most wides, as we have already shown 
(6, p. 201). There is not, it seems to me, any evidence about these species 
to give any ground for supposing them relicts. 
A comparison of the figures of rarity given at the bottom of Table II 
will also prove of little encouragement to the upholders of the hypothesis 
of dying out. All these species are on the average far more wide-ranging 
in New Zealand than the averages of the groups to which they belong. 
Even the endemics show a greater average range in New Zealand than the 
average of the wides of that country ( 7 , p. 331). 
Examination of the 14 unpredicted ferns (where if anywhere one might 
expect relicts) gives even less cause to believe in extinction. Six belong 
to class 1, ranging New Zealand from end to end, 3 to class 2, 5 to class 3, 
and none to any more localized class. 
There yet remains one objection which may be brought up against the 
whole prediction and discussion given above. It may be said that I have 
drawn my predictions so wide that I have included the whole flora, and have 
simply picked out from this list the species belonging to Stewart. As the 
bulk of the predictions are based upon the actual floras of the other islands, 
this can hardly be the case, but it will be well in conclusion to give a list of 
1 The fifth is Gentiana lineata , occurring in LongWood Range and Blue Mountains, Otago, and 
an unknown locality in Stewart. 
