New Zealand and their Distribution . 285 
coast species, which may have reached the Chathams by water, and Urtica 
australis finally, which only reaches the small islands in Foveaux Strait, is 
also probably a water-carried species. The prediction is thus borne out by 
the facts as well as can be expected. Even if all the six species be con- 
sidered as exceptional, the prediction is within 20 per cent, of accuracy. 
Both the southern invasion, and the floras of the Aucklands, Camp- 
bell, &c., contain a large number of South American forms, and the 
question has been much discussed as to how they have reached New 
Zealand, or whether indeed they did not reach South America from New 
Zealand. Guppy ( 5 , p. 294) discusses the possibilities of water carriage, 
and shows that seeds drifted from South America could only reach the 
extreme north of New Zealand. But an examination of the New Zealand 
forms with South American affinities shows that those which do not range 
the entire length *of the islands are mainly concentrated in the south. The 
only ones with northern location are Sicyos angulata (range 0-540), 
Mesembryanthemum aequilaterale (360-500), Gratiola peruviana (0-940), 
Cyperus vegetus (60-500), and Scirpus sulcatus (80-700, Tristan da Cunha, 
not South America proper). Of these we have seen that the first probably 
entered from the Kermadecs, it also occurring in Polynesia and Australia, 
and the last may have been drifted by water from Tristan. Mesembryan- 
themum and Gratiola also occur in Australia, so that there remains only 
Cyperiis vegetus , and it is permissible to suppose that this may have arrived by 
water carriage, perhaps when the Lord Howe I. bank was dry land. 
Water carriage, however, will hardly explain how the others reached 
New Zealand from South America, and there remains the possibility that 
the transport was the other way, and that these species reached South 
America from New Zealand. Guppy shows that transport is possible from 
the south of New Zealand to Chili. The genera involved are mostly so 
cosmopolitan that one cannot argue from the generic distribution, and 
must take other matters into consideration. The great argument against 
this supposition, to my mind, is based upon age and area. These species, 
in South America, mostly reach not only Chili, but also Fuegia, the 
Falkland Islands, South Georgia, &c., and in some cases also Kerguelen 
and other islands of the Antarctic Ocean. It is, therefore, clear that they 
must be enormously old in South America, and if they had gone there 
from New Zealand they must be yet older in that island, and should 
therefore be very widely distributed there, which is exactly what does not 
occur. Their distribution shows clearly that they must be older in South 
America, and must therefore have gone to , not from, New Zealand. 
Or again, we may take single examples. It would seem very strange 
if Ranuncirius biternatus spread from Macquarie to South America (it does 
not occur in New Zealand), or Cardamine glacialis , found only in the 
Aucklands, Campbell. Macquarie, and South America, or other species 
