1916] Kroeher: Floral Belations Among the Galapagos Islands 217 



islands, no doubt because they are nearer, also share more, on the 

 whole, in the floras of Wenman and Culpepper than the western and 

 central groups. 



For the sake of completeness, though it does not seem to add nnich 

 that is new, I have included Table \\\. This is a computation, along 

 the lines suggested above, of the proportion which the number of 

 species common to each pair of islands bears to the total number of 

 different species known from the same two islands. Tlie percentages 

 are based on Dr. Stewart's data, with the islands rearranged in order 

 as in my Table VI. 



It is obvious that in a percentage table of this sort the high num- 

 bers will not all be at the heads of the columns and at the left of the 

 rows, as in Table VI, but must cluster about tlie intersections of 

 columns and rows. For instance, if all of the forty species on a 

 given island are found also on a second island whose total number 

 of forms is forty, and again on a third island whose wealth of species 

 however reaches 200, the percentage for islands one and two will be 

 100, and for one and three only twenty. I have indicated by heavy 

 type the highest percentage occurring in each column. It will be seen 

 that these bold-face numbers practically all occur about where they 

 should come as a matter of mathematical probability; namely, in close 

 proximity to the row of spaces which diagonally bisects the table. 

 (If corresponding entries had been made also in the horizontal rows, 

 the arrangement of the heavy-type numbers would of course have 

 been .symmetrical to this diagonal axis.) This distribution once more 

 corroborates mere floral wealth as the fundamental factor in island 

 relationships. At the same time, the notable perturbations from prob- 

 ability are practically all due to geographical situation. Compare the 

 \\\\x\\ figni-('s for Bindloe-Abingdon and Jervis-Dimcan, to which Nar- 

 borough-Albemarle just fails of being added. 



If the distribution of the heavy-type numbers were mathematically 

 regular, the entry of the one such number in each column should 

 result in Ihcir appearance also one in each row; which is appi'oxiinately 

 the case. The one conspicious deviation from this theoretical rule is 

 afforded by Gardner, whose horizontal row will l)e .seen to contain as 

 many as five heavy numbers. This can scarcely be an accident, and I 

 am inclined to attribute it to the sliuhtly pi-ei)ouderating iuHuence of 

 the southeastern islands on the remainder of the archipelago. A 

 similar influence appears deducible as regards southeastern Ilood. with 



