l!^l<i| h'rofhrr: Floral lielaiionf; Among the Galapafjos Islands 203 



into the table. The three highest numbers in each column have been 

 j)iiiit('<l in black type. A glance shows that these heavy numerals 

 all come in rows for Charles, Chatham. ;m<i Alhf-marle, tli<- three 

 islands which with respectively 267, 2G1, and 205 species are the 

 richest in the group, pjven the three vacancies caused by the inter- 

 section of the row and 1li<' imIhiiui for each of these islands are in- 

 structive; their place is taken in each case by a black number in the 

 row for James, the next richest island, with 153 species. The only 

 two exceptions afi- for Gardner, (iT pf-r cent of whose species recur 

 on neai'l)\ Hood as against only 52 on riclur but much more distant 

 Albemarle; and isolated and scantily vegetated Tower, for which 

 neai-l)\- At)ingdon with 58 \>cr cmt also rejilaces Albemarle with 41. 

 These two exceptions are interesting, it is true, because they indicate 

 the influence of geographical position, contrary to Professor Robin- 

 S(jn's statements. But they are too few to be of nnidi significance; 

 and in general, the results reveal too little, other than the overwhelm- 

 ing influence exerted on the results by the absolute number of species 

 growing on each island, to endow Hh- nn'lliod cmployi-d with much 

 validity as a means of determining inter-island relationships. 



The same is true when tin- largest numbers encountered in each 

 horizontal I'ow twc selected, as lias bi-i-n doni- in iln- table by the use 

 of italics; only in this case it is of course the [xxn- or small islands 

 that appear most frequently. Thus it will be seen that Charles, 

 Chatham. Albemarle, and James are not represented at all by italics; 

 while Tower, Gardner, Abingdon, and Jervis, with only nineteen, 

 thirty-three, fifty, and twenty-two species respectively, appear from 

 ten to five times. 



It therefore occured to me to combine the two sets of percentages 

 given in the rows and columns of Table II into their means. ^lathe- 

 maticall.N' this procedure does not seem justifiabli'. as this mean does 

 not express anything intrinsic. It would have been preferable, per- 

 haps, to give the percentage which the number of species, common 

 to each pair of islands, formed of the total number of distinct species 

 found on the two islands. But this plan, besides involving some com- 

 putation, seemed open to the objection that after all its results would 

 depend too directly on the wealth of the various floras. Thas, only 

 3 per cent of all the species found on Albemarle (205) and Tower 

 (19) are common to both; but 11 per cent of those occurring on 

 Albemarle and Seymour ("47), and as mnr-h as 27 per cent of those on 



