292 Willis . — The Floras of the Outlying Islands of 
The islands must have had connexion by land with New Zealand. 
The Kermadecs contain a number of species which do not occur in 
New Zealand, but occur in Polynesia ; it therefore seems probable that 
they received part of their flora direct from Polynesia. 
The distribution of the Kermadec species in New Zealand is considered, 
and it is shown that it does not agree with any hypothesis but that of a land 
bridge between the two. 
The Kermadecs probably were not in the track of the main northern 
invasion of New Zealand, but were upon the route of a minor invasion from 
Polynesia. 
25. The Kermadec species which reach the Chathams also reach in 
general as far south in New Zealand as Dunedin. 
The Aucklands show a great proportion of Monocotyledons, and the 
distribution of this group in New Zealand is briefly considered* it being 
shown that it probably took part in both invasions. 
The wides of New Zealand nearly all range to one or the other end of 
the islands, if not to both. Those which do not (Table XVI) include the 
doubtful determinations, species of probable recent introduction, &c. 
26. The proportion of Monocotyledons is greatest in the Aucklands, 
least in the Kermadecs, and intermediate in the Chathams. 
27. Stewart Island shows a greater proportion of Monocotyledons 
than New Zealand. 
28. Species reaching both Aucklands and Chathams reach also in 
general to Auckland city. 
The evidence is against the probability of carriage by water of South 
American forms to New Zealand, or of New Zealand forms to South 
America, though there are a few that may have been so carried, and favours 
the former existence of a land bridge, which probably passed to New 
Zealand nearer to the Aucklands than to the other southern islands. 
29. The families missing in the Chathams are the smaller families of 
the Stewart list, in general ; (30) similarly the smaller genera. 
31. A great part of the Chathams’ flora can be predicted by aid of age 
and area. 
32. The Chathams have more species in the highest classes (widest 
distribution in New Zealand) than the Kermadecs and Aucklands. 
The hypothesis has now been successfully used to make no fewer than 
67 predictions, which have proved to be correct on examination of the facts, 
and increase of area with age may thus perhaps be considered as being the 
chief positive factor in determining the distribution of plants about the globe, 
the chief negative factor being the presence of barriers. 
Finally, a restatement of the hypothesis is made. 
