CH. vii] CONFIRMATION BY PREDICTION 73 



and also one of the three outlying groups so often mentioned, 

 the wides form 41 per cent., when two groups are reached they 

 form 64 per cent., and of those plants that reach Stewart and 

 all three, i.e. Kermadecs, Chathams, and Aucklands, they form 

 80 per cent. The result agrees exactly with the prediction, con- 

 firming the hypothesis in a very striking manner. 



Or we may predict that the far outlying islands will have a 

 large proportion of forms in common with one another and with 

 Stewart, all being old in New Zealand, and that the jjroportion 

 will be much larger than that in common with New Zealand. 

 In actual fact, one finds 81 per cent, of the Stewart families, 

 67 per cent, of the genera, and even 40 per cent, of the species, 

 on the other islands, Avhile of the plants that occur in New Zea- 

 land, but not on Stewart, only 32, 17, and 15 percent, respec- 

 tively occur, an enormous difference. The prediction is com- 

 pletely borne out by the facts, and it will suffice to quote one 

 or two instances. The Kermadecs have 30 per cent, of the genera 

 that occur upon the Aucklands, 1200 miles away, in a totally 

 different climate, and only 19 per cent, of those of New Zealand. 

 Of 52 species occurring outside the Kermadecs, as well as in 

 those islands, 30 occur in the Chathams, and even 5 in the 

 Aucklands; and so on. 



One may in the same way predict a great similarity between 

 the floras of the islands off the British coast, above mentioned. 

 On examination, one finds, in the five families before considered, 

 that their 70 genera have in the British Islands an average of 

 4-7 species, against 3-4 for the whole flora. Whilst about 37 per 

 cent, of the whole 175 species of these families are confined to 

 one island, 24 per cent, are foinid on two, 19 per cent, on three, 

 and 19 per cent, on all four, widely separated, and widely 

 different in climate, etc., though they be. The average occur- 

 rence of each species is upon 2-2 islajid groups of the four. 



Or we may predict that the genera which are common to the 

 islands and New Zealand, taking at least two groups of the 

 three, will be very old genera, and consequently in general will 

 be large genera in large families. This is so obvious when one 

 comes to make a list, and finds it composed of Ranunculus, 

 Cardamine, Lcpidium, StcUaria, Colobanthus, Geranium, etc., 

 that it hardly needs any further elaboration. The 32 genera 

 upon the islands in the first half of the New Zealand flora show 

 an average size of 144 species, against an average for the world 

 of only 12. Only five of them, Corynocarpus, Coriaria, Panax, 

 Samolus, and Calystegia, are below the average in size. 



