cii. VII] CONFIRMATION BY PREDICTION 71 



It may be applied to less if it be simply desired to gain an argu- 

 ment from greater or less probability to add to other arguments 

 in favour of some point, but when it is to form a main argument 

 it must be applied to at least ten allied species at once. By this 

 means the exceptional species, of which there are many, will 

 be lost in the crowd, and also a group of species will be obtained 

 M'hich react to their surroundings in much the same way, have 

 more or less the same rapidity of dispersal, and so on. On 

 averages there can be no question about the wider dispersal in 

 New Zealand of the Chatham plants, though individuals can be 

 found with little dispersal there. The herbaceous Compositae 

 may be enormously younger in the islands than the woody 

 Leguminosae, for example, and also younger in New Zealand, yet 

 by virtue of their better dispersal mechanism, and the fact that 

 they are herbs, may be much more widely distributed in the 

 latter, and may even have started much later from New Zealand 

 than the Leguminosae (which could hardly cross a strait) and 

 yet have reached the islands. Both groups, however, obey Age 

 and Area, though they cannot be compared with one another 

 as to relative age. 



If there were, again, a" great change of conditions between 

 New Zealand and the Chathams, or any serious barriers like 

 mountains, this would completely alter the hst of plants that 

 might arrive. One must remember all these provisos in dealing 

 with the distribution of plants, but none the less one finds that 

 by keeping to the Age and Area rule as enwiciated, and dealing 

 always with groups of allied species, results may be obtained 

 that are fairly reliable. 



To return to predictions, another upon the following lines (132) 

 was equally successful. A family will rarely arrive in a country 

 as a group of genera simultaneously; some will arrive sooner 

 than others. On the average, therefore, in any circle of affinity, 

 the families with several genera Avill be older in that coiuitry 

 than those M'ith one or two, as it is all but impossible that their 

 first genus should only arrive at the same time as the solitary 

 one of another family. This being so, we shall therefore expect 

 the larger families of New Zealand to he better represented upon 

 the outlying islands than the smaller, as being older. On Stewart 

 Island, at the south end of New Zealand, avc do in fact find this 

 to be the case, as the following table shows : 



