CHAPTER XVIII 



THE HOLLOW CURVE OF DISTRIBUTION 



-Dy far the most remarkable feature that stands out through all 

 the work described in the preceding pages is what may be termed 

 the " Hollow Curve of Distribution." It was first noticed in 1912, 

 when working up the flora of Ceylon for the first paper upon Age 

 and Area (123). This flora of 1028 genera was composed of 573/1 

 (573 genera of one species in Ceylon), 17G/2, 85/3, 49/4, 36/5, 

 20/6, and so on, the niunbers becoming somewhat irregular after 

 six, but decreasing fairly regularly if taken in twos (genera of 

 7 and 8 species, 9 and 10, etc.). Having already the knowledge, 

 familiar to systematists, that genera of one and two sj^ecies were 

 the most numerous, it was thought that the regular decrease of 

 the numbers might be accidental, and time did not then permit 

 of comparisons with other floras^. The hollow curve, however, 

 appeared again in 1916, in coimting up the areas of distribution 

 of the endemics of New Zealand. Unlike Ceylon, New Zealand 

 was treated by actual measurement, and when the endemics 

 were divided into ten classes by area, it was found that the lowest 

 class, though it occupied much the smallest area (40 miles by 

 length of New Zealand against 120 for most classes), contained 

 much the largest number of species, having 168 out of 902, while 

 the ninth class came next with 128. The two classes contained 

 32-8 per cent, of the whole number of endemics (of New Zealand 

 proper), although their area was only barely 15 per cent, of the 

 total. 



1 In actual fact, as may be quickly verified, all (or most) local floras 

 show the same thing, with their jrenera arranged in hollow curves when 

 grouped by the number of their (local) species. This is what one would 

 expect if genera are produced from other genera at a more or less uniform 

 rate, and in a more or less "casual" way. The subject will be treated in 

 greater detail in another place, and it will suffice for the present to call 

 attention to the fact that the hollow curve is regularly shown, as by the 

 Ceylon local flora (above, and cf. curve 4 on p. 237). The British flora shows 

 223/1, 90/2, 35/3. 32/4, Ifi/o, la/fi, 5/7, 7/8, 2/0, fi/lO, and so on to 71, the 

 numbers becoming rather irregular after (i. The flora of (.;aml)ridgeshire 

 (Babington, omitting Rubus, Uicraciuyn, and Salix) shows 210/1, (il,2, 3G/3, 

 21/4, 14/5, 6/6, and so on {I.e. curve 6). That of Wicken Fen, which is only 

 a very small area in the same county, is graduated in the same manner. 

 The flora of Italy (I.e. curve 9) shows the same thing, and so do the floras 

 of Greece, British India, New Zealand, the Bahama Islands, and others 

 that have been tested. 



13—2 



