[ 307 ] VII .
The Endemic Flora of Ceylon , with Reference to Geographical Distribution and Evolution By J. C. Willis , M.A. , Sc. I ) .
, Director of the Botanic , Rio de Janeiro .
Communicated by D. H. Scott , For .
.R.S .
( Received November 12 , \#151 ; Read December 10 , 1914 .
) Contents .
Local Commonness of Endemic and other Species Invalidity of the Theory of Natural Selection .
Commonness chiefly dependent on Age ... .
Relative Age of the Monocotyledons in Ceylon The Value of " Adaptations " ... ... ... ... ... ... Possible Size of Mutations ... ... ... ... ... ... Evolutionary History of a Species ... ... ... . .
Geographical Distribution in General ... ... . .
Summary ... ... ... ... ... ... ... ... ... ... ... ... .
The completion of my revised catalogue of the Ceylon flora* has enabled me to put together statistical notes on some of its more interesting features , and in the present paper I shall deal with the endemic flora .
Though only separated from India by a narrow strait , which at Adam 's Bridge is barely 28 miles across , sprinkled with islets and sandbanks , and from 6 to 12 feet in depth at low water , Ceylon contains an immense endemic flora , which enables it to rank with many oceanic islands : even the Sandwich Islands have not so many .
The flora of Ceylon consists of 2809 species of flowering plants , !
and of these no fewer than 809 are endemic , or a proportion of 28'8 per cent. And these are not mere varieties , or doubtful species , but good Linnean species , accepted by such authorities as Trimen and Sir Joseph Hooker .
On the other hand , the number of endemic genera is only 23 out of 1027 , so that the island does not rise to the rank of a widely separated province in geographical botany .
Incidentally , the facts about to be brought forward are an apparently insuperable objection to the theory of natural selection and adaptation .
In previous papers * 'A Revised Catalogue of the Indigenous Flowering Plants and Ferns of Ceylon , ' Colombo and London ( Dulau ) , 1911 .
t The 2808 of the Catalogue includes two Cycads , but three new species have since been discovered ( see 'Ann .
Perad .
, ' vol. 5 , pp. 167 , 217 , 387 ) .
VOL. CCVI.---B .
331 .
2 T [ Published , July 5 , 1915 .
PAGE dealing with the endemic flora of Ceylon I have endeavoured to prove that the endemic species have not been evolved in any kind of advantageous response to local conditions , and 1 have called attention^ to the fact that they are much rarer than the species of wider distribution that occur alongside of them .
But only in the present year have I completed the laborious task of extracting from the great Flora of Ceylon by Trimen and Hooker the statistics that are required .
This Flora is beyond doubt the most complete yet published for any tropical country , and thanks to 70 years ' work on this comparatively small island ( 25,000 square miles ) by such men as Gardner , Hooker , Thwaites , Trimen , and many more , the local flora is thoroughly well known , and it is unlikely that further work will do anything more than add half-a-dozen species or so to the list , without changing in any but a very few cases the very useful indications given as to relative abundance .
The Ceylon flora contains 149 families , with 1027 genera and 2809 species , of Angiosperms ( sensu Hooker ) , 58 with 105 genera and 196 species contain no endemic forms , but these are mostly small families .
The average number of species in a family in Ceylon is 19 , and the only family with more than 19 that contains no endemics is the Capparidacese with 20 .
It may be noticed here , as a fact that has possibly some bearing on the mechanism of evolution , that this family contains nine species endemic to Ceylon and South India , a very large proportion .
The only other families of any important size without endemics are the Menispermaceae and Solanaceae with 13 species each , the Ficoideae with 11 , the Lentibulariaceae and Polygonacese with 10 each , while there is 1 family with 8 , 1 with 7 , 1 with 6 , 2 with 5 , I with 4 , 10 with 3 , and 18 each with 2 and 1 species .
The families with the most endemic species , arranged in order , are :\#151 ; Orchidaceae .
ltubiaceae Dipterocarpaceae Euphorbiaceae .
Acanthaceae Melastomaceae .
Gramineae .
Myrtaceae .
Lauraceae .
Anonaceae .
Composite .
Geraniaceae .
Scitamineae .
Styraceae * " Some Evidence against . .
Table I. .
78 or 9-6 .
71 87 .
'47 ? ?
5*8 .
45 ? ?
5*5 .
39 5 ?
4-8 .
36 ? ?
4*4 .
31 ? ?
3-8 .
30 37 .
23 J ?
2*8 .
21 5 ?
2'6 , .
19 5 ?
2-3 .
18 2*2 .
17 5 ?
2*1 .
17 J ) 2-1 Natural Selection .
, " 'Ann .
Per ad .
, ' vol. 4 , p. 12 ( 1907 ) .
PR .
J. C. WILLIS ON THE ENDEMIC FLORA OF CEYLON .
Anacardiaceae Table . .
15 I\#151 ; continued .
or 1*8 per cent , of all the endemics , Araceae ... . . .
14 " 1-7 ? ?
9 9 Dilleniaceae . . .
12 " 1*4 5 ?
99 Leguminosae . .
12 , , 1 '4 ?
9 99 Cyperaceae . . .
12 " 1*4 9 9 99 Ebenaceae . .
.11 " 1*3 9 9 9 9 Guttiferae . . . . .
11 " 1*3 99 99 Sapotaceae . . .
11 " 1-3 9 9 9 9 Palmaceae . . .
11 " 1*3 99 9 9 Lahiatae ... . . .
10 1 *2 99 9 9 Loranthaceae .
* . .
10 " 1*2 99 99 besides 3 families with 8 , 4 with 7 , 2 with 6 , 4 with 5 , 2 with 4 , 13 with 3 , 19 with 2 , and 19 with 1 species .
Of the families with endemics 6 , the Monimiaceae with 2 and the Ancistrocladaceae , Dipsaceae , Nepenthaceae , Proteaceae , and Valerianaceae with 1 species each , have all their species endemic .
The Dipterocarpaceae ( 47 out of 48 ) , Styraceae ( 17/ 20 ) , Dilleniaceae ( 12/ 15 ) , and Anacardiaceae ( 15/ 19 ) have over 75 per cent , endemic .
The Lauraceae ( 23/ 33 ) , Melastomaceae ( 30/ 52 ) , Triuridaceae ( 2/ 3 ) , Gesneracese ( 8/ 12 ) , Sapotaceae ( 11/ 17 ) , Bixaceae ( 7/ 11 ) , Geraniaceae ( 18/ 29 ) , Myrtaceae ( 30/ 49 ) , Pandanaceae ( 3/ 5 ) , Guttiferse ( 11/ 19 ) , Ternstroemiacese ( 4/ 7 ) , Anonaceae ( 21/ 39 ) , Palmaceae ( 11/ 21 ) , and Rubiaceae ( 71/ 138 ) have over 50 per cent. The Balano-phoraceae , Burmanniaceae , Connaraceae , Cornaceae , Myristicaceae , Pittosporaceae , and Thymelaeaceae have each exactly 50 per cent , of their members endemic .
Local Commonness of Endemic and other Species .
We may now go on to consider the distribution of these endemic forms within the island .
Trimen divides all species into six classes\#151 ; Very Common , Common , Rather Common , Rather Rare , Rare , and Very Rare.* Now as Very Rare is applied to any species which has only been found in one place , or in two close together ( or more rarely at some distance apart ) , and as there are 455 of these , it is evident that if these classifications were used with absolute uniformity they should show approximately equal numbers in each heading\#151 ; one-sixth of the total , or 468 in each .
But in actual fact they show , for the whole flora\#151 ; * Trimen died leaving no material for Gramineae , and I have myself calculated the rarities of tire species , adhering as closely as possible to the system he appears to have employed .
Table II .
Number .
Marks .
Percentage .
1 .
Very common ... .
285 285 10-1 2 .
Common 670 1340 23-8 3 .
Rather common ... .
555 1665 19-7 4 .
Rather rare 429 1716 15-2 5 .
Rare 415 2075 14-7 6 .
Very rare 455 2730 16-1 If we mark these from 1 to 6 and calculate the average rarity , we find 2809 species with 9811 marks , or an average rarity of 3'49 , when theoretically it should be 3*5 .
There can be little doubt that at some former period Ceylon and Southern or Peninsular India formed part of one land mass , even if they were not at the end of a long continental projection from Africa .
This being so , we have divided the plants of Ceylon into three groups to make the comparisons that follow .
In the first place we have taken those that are endemic to Ceylon only , of which , as we have already said , there are 809 .
In the second we have taken those that are only found in Ceylon and Peninsular India ; of these there are 492 , and it may reasonably be taken for granted that they are on the whole older than the first .
The third group , which comprises 1508 plants , is composed of those that have a wider distribution than this\#151 ; not necessarily very much wider , but still going beyond the limits we have marked out for the second group , e.g. , into Assam .
Now , if we take the distribution of these three groups in point of commonness , as given in Trimen 's Flora , we arrive at a somewhat unexpected and remarkable result .
I had long known , and in fact published , that the endemics were rarer than the others , but was agreeably surprised when the final totals came out with such arithmetical precision as they actually show .
Table III .
i j i Endemic to Ceylon .
Ceylon and P. India .
Wider distribution .
VC . . .
19 45 221 c . . . .
90 118 462 RC . . .
139 103 313 RR . . .
136 84 209 R . . . .
192 64 159 VR . . .
233 78 144 809 Marks 3518 492 Marks 1714 1508 Marks 4579 Rarity* .
4-3 3*5 3-0 * I.c.in figures from 1 to G. To reduce this rarity even to 4*2 would require that 80 species should each be placed a class higher in the scale .
DR. J. C. WILLIS ON THE ENDEMIC FLORA OF CEYLON .
The endemic species increase in number from VC to VR , while the species of wide distribution follow the reverse direction .
This result shows much more clearly* if one multiplies the figures by the necessary factors to bring them to the same number under each of the headings of VC , C , \amp ; c. , i.e. to a total of 468 ( one-sixth of the total number of species ) under each .
These factors are l-64 , 0'70 , 0*84 , P09 , 1T2 and 1*03 , respectively .
We thus get the following Table :\#151 ; Table IV .
| Endemic to Ceylon .
Ceylon and P. India .
Wider distribution .
YC . . .
31 74 362 C . . . .
63 83 323 RC . . .
117 87 263 RR . . .
148 92 228 R . . . .
215 72 178 YR . . .
240 80 148 Rarity . .
4-4 3-5 2-9 The result which was distinctly enough indicated in the first table now comes clearly out , and it is evident that the Ceylon endemics are on the whole much below the average in their degree of commonness , while the species of wide distribution are much above it , both of them at the same time showing figures that progress regularly from one end of the scale to the other .
The species common only to Ceylon and Peninsular India , on the other hand , are fairly normally distributed over the scale , as many under one head as under another , and with the average rarity of the whole flora .
Thus in general the rarest plants in Ceylon are the local endemics , and the commonest those of wide distribution .
This is not at all the result that one would expect had the endemics , as is usually supposed , been developed by the aid of natural selection to suit the local conditions .
One would at least have expected that they would have been as common as the species of wide distribution , or at any rate as those developed to suit Ceylon and South India , but they are much rarer than both , and of these two groups the latter is rarer than the former .
Only 2 *3 per cent , of the Ceylon endemics have reached the stage indicated by Very Common , and no less than 14*6 per cent , of the species of wide distribution have reached it .
Another interesting comparison may be made by taking the second table above mid placing alongside the numbers the theoretical figures , supposing all species to have reached their final distribution of one-sixth under each head .
We thus get : * It of course shows exactly the same result if calculated as percentages .
Dli .
J. C. WILLIS ON THE ENDEMIC FLORA OF CEYLON .
Table V. Ceylon Ceylon and P. India .
Wide .
Actual .
Theo- retical .
Difference .
Actual .
Theo- retical .
Difference .
Actual .
Theo- retical .
Difference .
YC .
31 135 - 104 74 82 - 8 362 251 !
+ 111 C . .
63 135 - 72 83 82 + 1 323 251 + 72 RC .
117 135 - 18 87 82 + 5 263 251 + 12 RR .
148 135 + 13 92 82 + 10 228 251 - 23 R . .
215 135 + 80 72 82 - 10 178 251 - 73 YR .
240 135 + 105 80 82 - 2 148 251 - 103 It is evident on comparing the numbers that a simple perfectly indifferent regrouping , by which some of the endemics would gradually become more common , and some of the " wides " more rare , would reduce all to much the same level of commonness , with approximately equal numbers under every heading .
It must of course be made clear that we are here dealing only with averages , but the numbers under each head are large enough to give averages that may be fairly well relied upon .
We may safely say that on the average the local endemic species are the rarest , those of Ceylon and Peninsular India next most rare , and those of wide distribution commonest .
Having thus traced , on the grand totals of species of each type of distribution , the result of the summing up of the endemic flora , as compared with the other two sections of which the Ceylon flora is composed , we may go on to show that this result is not one which simply comes out on the total , hut also holds in detail , and is not due to any such fact as that some one or t wo families are very rare , and might he regarded as being driven to the wall .
The detailed figures show the results given in Table VI .
Looking this over , we see that the whole range of the rarity of the Ceylon endemics is only from 3'9 to 4,9 ' " , working with not less than 14 species in a group ; and in fact a great proportion even of the families with only a few endemics , which are put together in the list , also come within these figures .
The species common to Ceylon and South India similarly vary oidy from 2*0 to 4'0 , and if a minimum of 20 species be taken , vary only from 3*0 to 4'0 .
The species of wide distribution vary from P0 to 4'G , and again if groups of not less than 20 he taken , the variation is only from 2*2 to 3 *8 .
f The means of these extremes correspond very nearly with the actual means of the whole summing up .
The variation in rarity between the different families or groups of families of Ceylon endemics is small , and goes to show that no one family has any particular advantage over another , and that there is no case that can he pointed to as showing that any ot * l.e. they are never so common us the mean rarity ( 3-5 ) of the whole flora , f I.e. never so rare as the commonest endemics .
DR , J. C. WILLIS ON THE ENDEMIC FLORA OF CEYLON .
Table VI .
Family Endemic species .
Marks .
Rarity .
Ceylon and P. India species .
Marks .
Rarity .
Wides .
Marks .
Rarity .
Orchidaceae 78 343 4-4 39 153 3-9 45 175 3-8 Rubiaceae . .
71 306 4-3 25 75 3-0 42 122 2-9 Dipterocarpacef e . . .
47 233 4-9 1 3 3-0 \#151 ; \#151 ; \#151 ; Euphorbiaceae .
45 201 4-4 26 83 3-1 59 184 3-1 Acanthaceae 39 170 4-3 30 111 3-7 24 54 2-2 Melastomaceae 36 165 4-5 11 35 3-1 5 17 3-4 Gramineae .
31 146 4-7 20 80 4-0 189 557 2-9 Myrtaceae . . !
' . .
' 30 133 4-4 6 17 2-8 13 34 2-6 Lauraceae . .
23 94 4-0 4 10 2-5 6 16 2-6 Anonaceae .
21 89 4-2 12 32 2-6 6 28 4-6 Compositae . .
19 !
78 4-1 22 71 3-2 37 92 2-4 Geraniaceae 18 71 3-9 5 14 2-8 6 12 2-0 Scitamineae 17 77 4-5 4 16 4-0 16 47 2-9 Styraceae . .
17 78 4-5 2 4 2-0 1 1 1-0 Anacardiaceae .
15 62 4-1 2 6 3-0 o 4 2-0 Araceae . . .
14 63 4-5 4 9 2-2 15 41 2-7 3 families with 12* .
36 163 4-4 43 170 3-9 304 958 3-1 1 J\gt ; nt . .
44 189 4-2 16 51 3-1 21 60 2-8 2 10 } . .
20 78 3-9 21 77 3-6 26 79 3-0 7 8 or 7S 52 226 4-3 34 126 3-7 111 333 3*0 6 6 or 5|f 32 128 4-0 23 84 3-6 50 139 2-7 15 4 or 351 47 190 4-0 43 158 3-6 152 449 2-9 38 2 or 1** 57 235 4-1 63 204 3*2 218 710 32 91 families . .
809 3518 4-3 456 i 1589 3-4 \#166 ; 1348 4112 3-0 * Dilleniaceae , Leguminosae , Cyperaceae .
t Guttiferae , Ebenaceae , Sapotaceae , Palmaceae .
} Labiatae , Loranthaceae .
S Apocynaceae , Gesneraceae , Urticaceae , with 8 ; Bixaceae , Tiliaceae , .
Asclepiadaceae , Eriocaulonaceae with 7 .
|| Loganiaceae , Celastraceae , with 6 ; Ampelidaceae , Sapindaceae , Gentianaceae , Piperaceae , with 5 .
51 Ternstroemiaceae , Meliaceae , with 4 ; Yiolaceae , Polygalaceae , Linaceae , Rosaceae , Cucurbitaceae , Myrsinaceae , Convolvulaceae , Orobanchaceae , Yerbenaceae , Amarantaceae , Liliaceae , Commelinaceae , Pandanaceae , with 3 .
** Malvaceae , Rutaceae , Burseraceae , Olacineae , Rhamnaceae , Connaraceae , Rhizophoraceae , Samydaceae , Begoniaceae , Umbelliferae , Cornaceae , Boraginaceae , Scrophulariaceae , Podostemaceae , Myristicaceae , Monimiaceae , Thymelaeaceae , Burmarmiaceae , Triuridaceae , with 2 ; Ranunculaceae , Pittosporaceae , Caryo-phyllaceae , Ancistrocladaceae , Sterculiaceae , Ochuaceae , Haloragiaceae , Combretaceae , Lythraceae , Araliaceae , Valerianaceae , Dipsaceae , Oleaceae , Nepenthaceae , Proteaceae , Balanophoraceae , Hydrocharidaceae , Dioscoreaceae , Naiadaceae , with 1 .
the larger families is being driven to the wall in the struggle for existence .
Within the limits marked out by the degrees of rarity indicated by 3'9 and 4'9 practically all families are alike ; some are probably a little better suited to the local conditions than others , and are commoner , but even here we have not the least idea how much of the greater commonness is due to age or other factors .
This fact , that there is little difference between the different orders , also comes out if we take the rarities of the orders as a whole , in those cases where they have more than 19 species , the average number in a family in Ceylon .
This shows Anonacese 3'8 , Capparidaceae 3*9 , Dipterocarpaceae 4*9 , Malvaceae 2'9 , Tiliaceae 3*6 , Geraniaceae 3'3 , Rutaceae 3'3 , Ampelidaceae 2*9 , Leguminosae 3*4 , Myrtaceae 3 *7 , Melastomaceae 4*1 , Cucurbitaceae 3T , Rubiaceae 3*6 , Compositae 3*0 , Ebenaceae 4T , Styraceae 4T , Apocynaceae 3'1 , Asclepiadaceae 3'9 , Convolvulaceae 3*5 , Scrophu-lariaceae 3*0 , Acanthaceae 3*G , Verbenaceae 3T , Labiatae 3T , Amarantaceae 2*9 , Lauraceae 3*G , Loranthaceae 4*0 , Euphorbiaceae 3'G , Urticaceae 3'0 , Orchidaceae 4T , Scitamineae 3*7 , Commelinaceae 3*6 , Palmaceae 3T , Araceae 3*4 , Cyperaceae 3*2 , and Gramineae 3'2 .
Examination of this list shows that the whole range of rarity is only from 2*9 to 4 9 , and if the Dipterocarpaceae ( 47 out of 48 endemic ) be omitted only from 2*9 to 4*1 , or equal distances on either side of the averages .
And of this range a large part is due to the greater or less percentage of endemics in the orders .
Table VII .
1 Ceylon .
Ceylon\#151 ; P. India .
Wider .
Total .
Ceylon endemics .
Those families with rarity above 3-5 show\#151 ; 1 Per cent. 462 2075 209 733 298 951 969 3759 47-6 Rarity . .
4*27 3-50 3*19 3-87 Those with rarity 3 5 ( one family ) or under show\#151 ; 166 699 !
168 608 806 2386 1140 3693 14-5 Rarity . .
4*21 3*61 2-96 3-23 !
It is thus evident that the greater rarity of some families is largely due to the endemics that they contain , for the difference in the third column is only 0*23 , and in the fourth 0*64 .
If allowance be made for this , the range of variation is considerably reduced .
We may even follow the comparison into the genera .
Taking the first 22 that come , with reasonable numbers of endemic and non-endemic species , we find Table VIII .
55 endemic species with 224 marks .
5G 11011-endemic species with 154 marks .
Rarity 4*0 .
Rarity 2*7 .
There are more individual exceptions among the genera than among the families , but on the average the rule comes out in an unmistakable manner .
DR. J. C. WILLIS ON THE ENDEMIC FLORA OF CEYLON .
Another very striking comparison may be obtained by taking those genera in which there occur species of all the different types of distribution\#151 ; endemic , Ceylon and India , and wider .
In this way we obtain the following :\#151 ; Table IX .
Genus .
Species and Marks .
Ceylon .
Rarity .
Ceylon and P. India .
Rarity .
Wider .
Rarity .
Uvaria 2/ 8 4-0 2/ 4 2-0 2/ 10 5-0 Polyalthia 3/ 11 3-6 3/ 5 1*6 1/ 4 4-0 Alphonsea 1/ 6 6-0 1/ 2 2-0 1/ 6 6-0 Garcinia 2/ 7 3-5 2/ 4 2-0 1/ 2 2-0 Calophyllum 7/ 30 4-2 1/ 2 2*0 3/ 13 4*3 Biophytum 3/ 8 2*6 1/ 5 5-0 1/ 1 DO Impatiens 15/ 63 4-2 4/ 9 2-2 2/ 3 D5 Ochna 1/ 6 6-0 1/ 3 3-0 1/ 4 4-0 Yitis ... ... . .
5/ 21 4-2 7/ 21 3-0 7/ 16 2*2 Sapindus 2/ 10 5-0 1/ 4 4-0 2/ 5 2-5 Crotalaria 1/ 3 3*0 7/ 35 5-0 15/ 50 3-3 Desmodium 2/ 11 5-5 2/ 7 3-5 17/ 42 2-4 Derris 2/ 11 5-5 1/ 5 5-0 3/ 7 2-3 Pithecolobium 1/ 2 2-0 1/ 2 2-0 2/ 7 3-5 Eugenia 29/ 128 4-4 6/ 17 2*8 8/ 20 2-5 Osbeckia 5/ 20 4-0 3/ 4 1*3 1/ 4 4-0 Sonerila 8/ 41 5-1 3/ 15 5-0 1/ 6 6*0 Memecylon 21/ 97 4-6 3/ 13 4-3 3/ 7 2-3 Hedyotis 16/ 76 4-7 4/ 9 2-2 i/ 2 2-0 Ophiorrhiza 3/ 14 4-6 1/ 1 1*0 2/ 4 2*0 Randia 1/ 4 4-0 2/ 6 3*0 2/ 5 2*5 Knoxia .
2/ 2 1-0 1/ 5 5-0 1/ 3 3-0 Canthium 4/ 20 5-0 2/ 5 2-5 1/ 2 2-0 Pavetta 3/ 10 3-3 1/ 4 4-0 1/ 2 2-0 Vernonia 9/ 30 3-3 1/ 4 4-0 3/ 6 2-0 Blumea .
2/ 11 5-5 3/ 11 3-6 5/ 15 3-0 \gt ; Anaphalis 4/ 20 5-0 3/ 5 D6 1/ 2 2-0 Senecio 1/ 6 6-0 3/ 9 3-0 2/ 5 2*5 Ardisia 3/ 15 5-0 2/ 6 3-0 3/ 7 2-3 Diospyros 8/ 38 4-7 8/ 36 4-5 4/ 9 2-2 Symplocos 17/ 78 4-5 2/ 4 2-0 1/ 1 DO Linociera 1/ 1 1-0 1/ 5 5-0 1/ 3 3-0 Tylophora 3/ 13 4-3 2/ 8 4-0 3/ 12 4-0 Strychnos 3/ 13 4-3 2/ 8 4-0 2/ 6 3-0 Argyreia 1/ 1 1*0 2/ 11 5-5 2/ 7 3-5 Lettsomia 1/ 6 6-0 1/ 2 2-0 1/ 1 DO Ipomoea 1/ 6 6*0 2/ 10 5-0 8/ 27 3-3 Barleria 2/ 10 5-0 3/ 11 3-6 3/ 11 3-6 Justicia 3/ 11 3-6 2/ 7 3-5 4/ 9 2-2 Premna 2/ 11 5*5 1/ 4 4-0 4/ 12 3-0 Coleus 2/ 8 4-0 1/ 2 2-0 1/ 3 3-0 Anisochilus 1/ 4 4-0 1/ 6 6-0 1/ 5 5-0 Piper 3/ 9 3-0 1/ 2 2-0 5/ 11 2-2 Peperomia 2/ 7 3-5 2/ 9 4*5 1/ 2 2-0 Myristica 2/ 7 3-5 1/ 2 2-0 1/ 3 3-0 Litsea 9/ 35 3-8 1/ 2 2-0 2/ 2 DO Loranthus , 8/ 40 5-0 5/ 17 3-4 4/ 11 2-7 VOL. covi.\#151 ; b. 2 u Table IX\#151 ; continued .
Genus .
1 Species and marks .
Ceylon .
Rarity .
Ceylon and P. India .
Rarity .
Wider .
Rarity .
Phyllanthus 9/ 46 5-1 2/ 6 3*0 9/ 25 2*7 Glochidion 8/ 31 3-8 1/ 6 6-0 1/ 2 2'0 Croton 2/ 10 5-0 3/ 9 3-0 2/ 10 5-0 Agrostistachys 1/ 6 6-0 m 4-0 1/ 2 2-0 Mallotus 3/ 9 3-0 2/ 8 4-0 3/ 8 2-6 Macaranga 1/ 5 5-0 1/ 1 1-0 !
/2 2*0 Ileus 3/ 12 4*0 4/ 14 3*5 14/ 40 2-8 Pouzolzia 1/ 3 3-0 2/ 6 3-0 2/ 5 2-5 Microstylis 3/ 15 5-0 2/ 10 5-0 1/ 3 3-0 Liparis 5/ 27 5-4 3/ 12 4*0 4/ 12 3-0 Dendrobium 3/ 12 4-0 3/ 10 3-3 3/ 9 3-0 Vanda 1/ 6 6-0 1/ 3 3-0 2/ 6 3-0 Saccolabium 5/ 17 3-4 2/ 8 4-0 2/ 9 4*5 Habenaria 8/ 37 4-6 7/ 26 3-7 2/ 6 3-0 Dioscorea 1/ 3 3*0 1/ 2 2-0 4/ 5 1*2 Asparagus 1/ 2 2-0 1/ 2 2-0 2/ 4 2-0 Commelina 1/ 6 6-0 1/ 4 4-0 8/ 31 3-8 Cyanotis 2/ 9 4-5 4/ 12 3*0 2/ 4 2-0 Eriocaulon 7/ 38 5-4 3/ 14 4-6 8/ 17 2-1 Fimbristylis ... . .
3/ 16 5-3 2/ 7 3-5 24/ 63 2-6 Carex 5/ 20 4-0 3/ 8 2-6 13/ 47 3-5 Isachne 2/ 9 4-5 2/ 6 3-0 3/ 8 2*6 Panicum 1/ 6 6-0 1/ 4 4-0 40/ 109 2-7 Arundinella 3/ 13 4-3 1/ 4 4-0 2/ 7 3-5 Pollinia 1/ 5 5-0 1/ 4 4-0 2/ 7 3-5 Cymbopogon 1/ 6 6-0 1/ 4 4-0 7/ 27 3-8 Eragrostis | 2/ 9 4-5 1/ 5 5-0 10/ 26 2-6 Total 321/ 1367 \#151 ; 166/ 569 315/ 895 \#151 ; Rarity 4 .2 3- 4 2 .8 Of the 74 genera enumerated in this list 29 only are exceptions to the rule that the order of rarity is Ceylon , Ceylon and India , Wider distribution , but of these 29 only one ( Saccolabium ) goes in the exact reverse direction all through .
Most of them show Ceylon commoner than Ceylon-Peninsular India , or the latter than Wide .
If we add the whole 29 together we get\#151 ; Table X. Ceylon .
Ceylon-P .
India .
Wider .
Number .
Marks .
Number .
Marks .
Number .
Marks .
Rarity . . .
91 3 349 .8 52 3 178 .4 80 3- 260 2 DE .
J. C. WILLIS ON THE ENDEMIC FLOEA OF CEYLON .
So that even the exceptions give a result in the right direction when added together .
If we add these genera together into their families , we find 33 represented .
Of these Table XI .
Genus .
Species and marks .
Ceylon .
Earity .
Ceylon and P. India .
Earity .
Wider .
Earity .
Anonaceae 6/ 25 4*1 6/ 11 1*8 4/ 10 2-5 Guttiferae 9/ 37 4-1 3/ 6 2*0 4/ 15 3-7 Oleaceae 1/ 1 1*0 1/ 5 5-0 1/ 3 3-0 Convolvulaceae ... .
3/ 13 4-3 5/ 23 4-6 11/ 35 3-1 Piperaceae 5/ 16 3-2 3/ 11 3-6 6/ 13 2-1 Myristicaceae 2/ 7 3*5 1/ 2 2-0 1/ 3 3-0 Commelinaceae ... .
3/ 15 5-0 5/ 16 3-2 10/ 35 3-5 29/ 114 \#151 ; 24/ 74 \#151 ; 37/ 114 Earity ... .
3- 9 3 .0 3 .o Table XII .
Genus .
Endemic species .
Non- endemic species .
Genus .
Endemic species .
Non- endemic species .
Eanunculus . .
C C Equal .
Cyathula ... .
VE VC Dillenia ... .
EC C Psilotrichum . . .
E E Equal .
Miliusa ... .
C EC * Cryptocarya . . .
E EC Pittosporum . . .
EC EC Equal .
Balanophora . . .
E C Salomonia . . .
EE EC Putranjiva . .
VE E Mesua EC C Trigonostemon . .
VE E Ternstroemia . .
EE c Claoxylon . . .
EE C Hugonia ... .
E EC Artocarpus . . .
C EE * Glycosmis . . .
YE YC Debregeasia .
EE c Aglaia EC C Calanthe ... .
EC EE * Walsura ... .
EE VC Hetseria ... .
E E Equal .
Microtropis . .
EC EC Equal .
Cheirostylis . . .
E E Equal .
Gymnosporia . .
YE EE Vanilla ... .
EE C Ehamnus ... .
EE C Phrynium . . .
E C Nephelium . .
YE C Areca . . . . .
EE YC Connarus ... .
EE YC Phoenix ... .
YC VC Equal .
Pygeum ... .
EC C Amorphophallus E c Carallia ... .
C EC * Aponogeton . . .
C c Equal .
Alangium . . .
EE EC Mapania . . . .
YE EE Mastixia ... .
EE EE Equal .
Leptaspis . . . .
EE EC Emilia C YC Zenkeria ... ' . !
EE EC Caralluma YE E Coelachne . . . .
EE EC Tournefortia . . .
E E Equal .
Lophatherum . .
E C Klugia ... .
C C Equal .
1 2 u 2 24 show the expected order of rarity , two have Ceylon-India and Wide equal , though less than Ceylon , and there are seven exceptions which , however , when added together give the result indicated in Table XI .
Another interesting comparison is between the species in those genera which contain one endemic and one non-endemic species each .
In this way we obtain the result shown in Table XII .
Here we find that the endemic species is rarer in 32 cases , the non-endemic rarer in four cases , which are marked with an # , whilst the records are equal in 11 cases .
In 15 cases the non-endemic species is confined to India and Ceylon , and in these cases the endemic species shows an average rarity of 3 while in the 32 cases where the non-endemic species is of wider range the endemic shows a rarity of 4*2 .
Adding up all these species we get the following result:\#151 ; Table XIII .
Class .
Endemic species .
Marks .
Non-endemic species .
Marks .
YC 1 1 7 7 C 7 14 17 34 RC 7 21 11 33 RR 14 56 5 20 R 10 50 7 35 VR 8 48 0 0 Rarity 47 4* 190 0 47 2- 129 7 Even yet the possibilities of comparison are not exhausted .
If we take the seven genera in which there are three species , one of each type of distribution , we get the following Table :\#151 ; Table XIV .
Genus .
Ceylon species .
Ceylon and P. India .
Wider .
1 .
Alphonsea YR C VR 2 .
Ochna YR RC RR 3 .
Linociera VC R RC 4 .
Lettsomia YR C VC 5 .
Anisochilus RR VR R 6 .
Agrostistachys . . . .
- .
VR RR C 7 .
Macaranga R VC C Marks 34 23 23 Rarity 4-8 3-2 3-2 DR. J. C. WILLIS ON THE ENDEMIC FLORA OF CEYLON .
The last two figures are equal , but we are dealing here only with seven species , five of which are exceptions to the rule .
It is thus clearly evident that the rule that we enunciated above , that the order of rarity is Ceylon , Ceylon and Peninsular India , Wider dispersal , holds throughout the comparison with the most extraordinary regularity .
Though there are a number of individual exceptions in the genera , as would be expected , for all species cannot be at the same exact degree of adaptation , nor have had the same favourable or unfavourable chances , the rule in any group of more than about 15 species comes out with the regularity of Mendels law , especially when we consider that we are dealing with estimates , and not directly with actual facts of observation .
Incidentally , a comparison may be applied to the numbers of species of each type of distribution found in the different climatic zones of the island , whether the wet zone , with the well distributed rain , or the dry , with its long dry season .
In this way we obtain the following figures :\#151 ; Table XV .
Ceylon endemics .
Ceylo n and P. India .
Wider .
Dry .
Dry and wet .
Wet .
Dry .
Dry and wet .
Wet .
Dry .
Dry and wet .
Wet .
VC ... .
1 1 17 9 12 24 25 145 51 c. . . .
6 8 76 22 27 69 53 231 178 RC ... .
5 3 131 17 19 67 69 102 142 RR . . . .
8 0 128 21 6 57 55 46 108 R. . . .
16 4 172 18 6 40 49 25 85 YR . . . .
12 2 219 33 1 44 53 7 84 Total . .
48 18 743 120 71 301 304 556 648 Marks . .
212 58 3248 476 183 1055 1121 1264 2194 Rarity . .
4*4 3-2 4-3 3*9 2-5 3-5 3-6 2-2 3-3 Adding up all the species of the dry zone , we find 472 confined to it with 1809 marks , or a rarity of 3*8 ; those of the wet zone only are 1692 with 6497 marks , or also a rarity of 3*8 .
But the species that occur in both zones , 645 with 1505 marks , are much commoner in both , and show a rarity of 2'3 , are fairly near to the level of " Common .
" How this result is to be interpreted it is difficult to say ; whether it is that these species are really specially adapted to both zones , and have become common in both , or whether it is that they are on the whole the first arrivals in the country , and that most species can really live in both zones , or what , we do not at present know ; but the facts are very striking .
It is also very noteworthy that in the dry zone there are only 48 endemic species against 743 in the wet , though the species of wide distribution are only in the proportions of 304 to 648 , and the dry zone has twice the area of the wet .
Invalidity of the Theory of Natural Selection .
Now , when one conies to look into all this , it is seen to be a somewhat important result .
One may , so it seems to me , draw from it two or more conclusions , the first of which is that the local endemic species have not been\#151 ; as I have already shown in other ways in other papers\#151 ; developed in any kind of advantageous response to local conditions , as must have occurred did natural selection obtain .
Had this been the case , one might reasonably have expected them to be on the whole , at least , as common as the species of wide distribution .
But by no conceivable argument can this be maintained .
Trimen doubtless made errors in his estimates , but there is no reason to suppose that he always gave the widely distributed species the preference to the others in the direction of commonness.* The mere fact that the figures come out in such remarkable arithmetical progression is sufficient to show this , and the fact that the local endemics increase in number down to Very Bare , instead of up to Very Common ( as do the species of wide distribution ) , is enough to show that the endemics were not developed in advantageous response to local conditions .
A second great argument in favour of this contention is that the species common to Ceylon and Southern India , i.e. , species which must , on the whole , have been developed before the separation of those two countries , and which must therefore , on the whole , be older than those endemic to Ceylon only , are more common than the latter .
If we imagine that they were all developed in Ceylon , and spread subsequently to South India\#151 ; a purely gratuitous assumption\#151 ; we might then say that they had been developed to suit the Ceylon conditions , and , being older , had had more time to spread , and thus become commoner .
But , unfortunately for this hypothesis , these very species are still much less common than the species of wide distribution , and this fact is very difficult to explain on any theory of adaptation .
If they have been able to spread over Ceylon and Peninsular India , they have had ample time to become common in Ceylon ( i.e. , as common as the species of wide distribution ) if they were specially suited to the local conditions .
This leads on to another argument .
If these Ceylon endemics had been specially adapted by natural selection to the local conditions , there has unquestionably been ample time for them to become very common .
Trimen accepts as Very Common a species which is very common in the district which is suitable to it , whether wet or dry , warm or cold .
For instance , he labels leptopoda ( endemic ) as Very Common , though it does not occur in the warm or dry zones , but is only to be found in the wet , cool zone of the higher mountains , where it may be seen in nearly every * To equalise the endemics and the " wides " there would have to be made 1040 corrections in direction ; i.e. an inconceivable amount of error must have taken place in the estimates .
DR. J. C. WILLIS ON THE ENDEMIC FLORA OF CEYLON .
damp corner .
Out of the whole number of 809 endemics only 19 have spread to the stage indicated by Very Common .
And yet several introduced species have spread to this stage in a comparatively few years , though they belong to Ceylon families or even , as in the case of the famous Lantana , to Ceylon genera .
Tithonia diversifolia ( Composite ) , which was introduced in 1851 , began to spread about fifteen years later , and in 1900 was all over the island in damp enough spots , though its seeds have no pappus , and it spread almost entirely by vegetative reproduction .
Mikania scandens , which only began to spread about ten years ago , is already common over a large area near Peradeniya , reaches about 30 miles along the railway to Colombo , and has reached Trincomali down the river , and established itself near that place .
There has been ample time for any of the endemic species , had they been really well adapted to the local conditions , to have spread to the stage of Very Common , and the fact that they have not done so goes to show that they are no better adapted , on the average , than their forerunners .
Natural selection , again , to be effective , requires that many forms shall modify in the same direction , as Darwin himself admitted after the criticisms of Fleeming Jenkin .
Now , this is exactly what cannot be allowed in the case of the numerous endemic species of Ceylon , which very commonly are confined to one spot , where they cannot ever have been very numerous , and where yet , as I have already pointed out , they must have been evolved .
Coleus elongatus must have evolved on the summit of Ritigala , where it exists as about a dozen individuals , and cannot ever have been much more numerous .
The most numerous group of the Ceylon endemics are these Very Pares , and the numbers decrease steadily up to Very Common .
They must obviously have begun at one or other end of the scale .
They could not begin at Very Pare ( on the theory of natural selection ) , because the numbers are insufficient .
But if they began at Very Common , and were developed to suit the local conditions , it is a very remarkable thing that they increase steadily in number down to Very Pare , and are therefore , so far as one can see , in process of dying out .
Yet another argument against natural selection , though it is not a very strong one , may be derived from the fact that the families with many endemics , i.e. , the families which on the theory of natural selection one would suppose very well suited to local conditions , have their endemics rarer than they are in families that possess only few .
If we arrange the families in the order of the number of endemics contained in them , as in Table VI above , and add them up in approximately equal groups , we get the result indicated in Table XVI .
There is thus quite a perceptible difference in the degree of rarity between families that have many , and families that have few , endemic species , and the figures progress arithmetically , without a break .
These conclusions can be further confirmed by looking at the endemic genera .
If local species had been evolved to suit local conditions , one would , at any rate , expect Table XYI .
Endemic species .
Marks .
Rarity .
3 families 196 882 4-50 6 " .
204 909 4-39 14 " 201 870 4*31 68 " 208 857 4-11 the genera to be common , but this is by no means the case .
The degrees of commonness of the 52 species of the 23 endemic genera are shown in the following Table :\#151 ; Table XVII .
Schumacheria C , PC , Trichadenia . .
PR .
Doona . . . .
C , YP , Stemonoporus .
PR , P , YP , Monoporandra .
P , YP .
Julostylis .
PC .
Pityranthe .
c. Pseudocarapa pp. Gleniea .
c. Pericopsis .
c. Leucocodon . .
p. Schizostigma . .
PC .
Nargedia .
YP .
Scyphostachys .
YP , P. Champioriia . .
P. Ptyssiglottis .
P. ITortonia .
PC , P. Podadenia .
, YP .
Adrorhizon .
PC .
Alvisia . . . .
PR .
Octarrhena .
PC .
Cyphostigma .
PC .
Loxococcus PC .
Classifying these into their headings , we find : DR. J. C. WILLIS ON THE ENDEMIC FLORA OF CEYLON .
Table XVIII .
Marks .
VC .
... .
0 with 0 c .
... .
5 10 RC .
... .
9 27 RR .
... .
7 28 R .
... .
15 75 VR .
... .
16 96 52 with 23G marks , or a rarity of 4*5 .
Or , in other words , the species of these endemic genera are even rarer than the endemic species as a whole , and are grouped in the table in a similar way , with numbers increasing down to Very Rare .
But we may go further yet , and take the two genera , Doona and Stemonoporus , which have 11 and 15 species respectively , and on the theory of natural selection would therefore be supposed to be especially suitable to the local conditions .
These show\#151 ; Table XIX .
Doona .
Stemonoporus .
VC .
... 0 0 c .
... 1 0 RC .
... 1 0 RR .
... 2 2 R .
... 4 4 VR .
... 3 9 11 Marks 51 15 Marks 82 Rarity ... 4*6 5'4 Or , in other words , both these genera are rarer than the average of all the endemic genera , and Stemonoporus , which has more species than Doona , is the rarer of the two .
One gathers the impression that these two genera have perhaps been formed by what one may almost term explosive methods , a number of new species being formed comparatively rapidly , and in Stemonoporus later than in Doona .
It will bring out these points more clearly if we give them in tabular form , thus :\#151 ; 2 x VOL. CCVI.\#151 ; B. Table XX .
Number .
Marks .
Rarity .
Species of wide distribution 1508 4579 3-0 Species confined to Ceylon and Peninsular India ... .
492 1714 3-48 Mean rarity of all \#151 ; \#151 ; 3-49 Species endemic to Ceylon 809 3518 4-3 Species of all 23 endemic genera 52 236 4-5 Species of Doona 11 51 4-6 Species of Stemonoporus 15 82 5-4 Such figures as these no ingenuity can torture into any kind of support for a theory of the development of endemics to suit the local conditions .
There is not even a break in the continuity of the .
figures , which go in the exact reverse direction to what they should follow if the theory of natural selection could be relied upon .
Taking these arguments , together with those which I have already given elsewhere , * to show that endemic species have not in any way been developed to suit local conditions , there can , I think , be no hesitation in rejecting the latter hypothesis .
I may perhaps quote , as bearing upon the present paper , two short paragraphs from the paper referred to : " How can the existence of numerous endemic species , all closely allied to one common species , and all separated by some distance from one another , be explained on any but the parent and child theory ?
Take the Indian flora as an example .
To begin again at the beginning , in the genus Clematis S Cheiropsis , we have C. montana common all along the Himalaya , while C. napaulensis , C. barbellata , and C. acutangula are confined to particular sections .
We have Anemone ricularis common throughout the Himalaya , and various local allied species " ; and " It is possible , if not probable , that a group of allied species represents so many more or less stable positions of equilibrium in cell division , and it is at least entirely doubtful if any given species is specially adapted for the circumstances in which it is found .
" If all the species that now exist in Ceylon had been introduced at the same time by mixing their seeds in equal ( and excessive ) quantity , and sowing them broadcast over the island , their relative degree of abundance , and their particular geographical distribution , would undoubtedly , as these figures clearly show , have been entirely different from what they now are , and they would to some extent have afforded a test of the relative degree of adaptation of the different species to the local conditions .
At present , when we have no idea of the relative age of different species , to talk of them as if one were well adapted , and another not , is purely an effort of the imagination .
We have been in the habit of looking upon endemic species with very small range * " Some Evidence against ... .
Natural Selection ... .
, " 'Ann .
Perad .
, ' vol. 4 , p. 1 ( 1907 ) .
And cf. other papers in same journal , 1906-11 .
DR. J. C. WILLIS ON THE ENDEMIC FLORA OF CEYLON .
as if they were suited to very special local conditions , and yet some of these , when planted elsewhere , grow with remarkable success .
A very notable example is that of Cupressus macrocarpa , confined in nature to a small area in California , and yet proving to be one of the most successful Conifers in all the sub-tropical regions of the world .
If a species happen to get into a locality where its local adaptation is good , as , for instance , any of the weeds which have at different times become common in different parts of the world , it may spread rapidly to the stage of Very Common .
But this will only rarely happen .
Out of the many hundreds of species introduced into Ceylon at different times , not more than about 60 at most have become common .
Commonness chiefly Dependent on Age .
The second conclusion that we may , I think , justly draw from the remarkable Tables which have been set out is that , on the average , the commonness of a species depends upon the time that has elapsed from the period when it was first evolved in , or arrived in , the country .
No other explanation appears to me to fit the striking facts that the species of wide distribution are commoner than the species confined to Ceylon and India ( even within one family ) , and these commoner than the species confined to Ceylon , while at the same time the first show numbers increasing towards VC and the latter to VB .
The widely distributed species must , on the whole , be the oldest , the Ceylon-Indian the next , and the Ceylon endemics the youngest .
There is no evidence that any are dying out .
In other words , species are , on the whole , developed quite indifferently to local conditions , but , having evolved , they spread , if their degree of local adaptation is sufficient , and as they must have descended from others already growing in the neighbourhood , this should , in general , be the case .
They will in some cases , doubtless , from having bad local adaptation or meeting with unfavourable chance , die out , but most will spread to one of the stages VB , B , BB , BC , C , or VC , according to the time allowed , their degree of local adaptation , and the incidence upon them of favourable or unfavourable chance .
The same species may perhaps be ultimately VC in one country , say , Ceylon , and VB in another similar one , say , South India , owing to the action of one of these factors .
But on the average the commonness will mainly depend upon the age .
And to judge from the figures set forth , for example , in Table VI , no very large number of species need be taken to get an average ; apparently from 15 to 20 is enough .
All the figures given in this paper go to show that in all reasonable probability there is little to choose between species in their degree of adaptation to local conditions .
This comes out especially clearly in Table VI , where the whole of the endemics , taken in groups of not less than 14 , show rarity greater than the average of the whole flora , and yet agree wonderfully closely among themselves in the degree of rarity , which only ranges from 3'9 to 4'9 .
When one remembers that the figures are not taken directly from facts , but from estimates , of observation , this regularity 2x2 is wonderful , and the regular order and progression of the figures goes to show that it cannot be accidental .
From these and other facts one may reasonably , it seems to me , conclude that in the peopling of Ceylon with flowering plants the first species to appear , whether coming from elsewhere or developed for the first time within the island\#151 ; then probably still attached to Southern India , if not also to Africa\#151 ; were those which are now of wider distribution than Ceylon or Ceylon and South India .
They would find the island presumably more or less covered with cryptogamic vegetation\#151 ; -just as Krakatoa was at a certain period after the eruption\#151 ; but they would begin to spread , and probably , as their demands for food would , on the whole , be slightly different , they would gradually cover large areas , and in time a great number of them would reach the stage of commonness now represented by the expression YC , while those of them , which from lack of adaptation , late arrival , or ill chance , only reached the stage then represented by VR , would perhaps be at the stage of commonness now represented by R or RR .
Later on , and probably long before all the widely distributed species had arrived , there would begin to arrive , or to evolve locally , the species now confined to Ceylon and Peninsular India , which in turn would begin to force their way up in the scale .
These , having to compete with other flowering plants of the same genera , would probably be slower in ascending , but gradually they have done so , and in so doing have forced some of the " wides " to climb down a bit .
Still later , on the whole , would appear the species endemic to Ceylon , and begin forcing their way up .
As yet they have by no means succeeded in arriving at anything like an average dispersal , and a corresponding number of " wides " are similarly in their wrong positions .
But if time enough could be given , one might expect ultimately to find all three classes distributed fairly evenly over the scale .
We may I think reasonably draw the deduction that on the average of say 15 to 20 a species develops indifferently .
But each species in itself starts with the capacity of ultimately reaching ( under given circumstances ) one of the stages YC , C. RC , RR , R , YR , or extinction , according to how well it is adapted to the local conditions .
We are using the word adaptation here to include all the relationships of the species to its environment , e.g. , its liability to any diseases that may exist there .
This last is a very important point .
Some experiments which I have been carrying on upon the struggle for existence among closely crowded plants go to show that it is disease which kills the bulk of the losers , and that these are not necessarily the weakest but often well-grown plants .
The species then , will tend , on the whole , to spread to the local degree of commonness to which it is entitled by its degree of adaptation to local conditions .
But it must not be forgotten that this may be completely changed as time goes on , by the introduction of some new factor in the situation , such for instance as the arrival in the district of some new disease to which it falls an easy prey .
There is no doubt that on the whole , natural selection , given enough time , will finally ensure that each DR. J. C. WILLIS ON THE ENDEMIC FLORA OF CEYLON .
species takes up the local distribution and degree of* commonness to which it is entitled , but there must be an enormous element of chance .
The process must be extraordinarily slow ; we have no evidence that it is complete in any given case , and good indication in the figures that we have given above that it may be far from complete Ultimately , one would expect , the figures under the different degrees of commonness or rarity would equalise themselves for the various Ceylon species , and in so doing would force many of the species of wide dispersal to climb to some extent down the ladder , so as also to equalise them among the different headings .
On the average , or in some cases , possibly , the species endemic to Ceylon which have already reached the stages indicated by YC or C are species which are really better adapted to the local conditions than the average of all or of the endemic species , but we do not know that they are not simply the oldest endemic species , which were very early developed from species of wide distribution which arrived very early .
There are 19 of these species , belonging to the families Geraniacese ( Balsaminacese ) , Myrtacese , Iiubiaceae , Composite , Oleacese , Apocynaceae , Gentianaceae , Convolvulaceae , Labiatae , Palmaceae , Pandanaceae , Araceae , and Gramineae , all of which are widely distributed orders , and often orders which we have other reasons to suppose are old .
The rarity of these orders taken together is 3*36 , against 3'5 , so that on the whole they are probably old.* The rarity of the Ceylon and Indian species of those genera which contain the endemics is 3'2 , or less than the rarity of these species as a whole ( 3*5 ) , and the rarity of the widely dispersed .species ( 20 in number ) contained in these genera is 2'3 , or less than the rarity of the widely dispersed species as a whole , which is 3 .
All these facts go together to show that perhaps the commonness of these 19 endemic forms is really only a question of age .
As we have already said , it perhaps becomes harder for every new species that comes along to reach the Very Common stage , as the ground will be more closely taken up by a greater variety of species .
The wonderful regularity and uniformity with which these figures come out goes to show the great accuracy of the estimates made by Trimen in his Flora , and at the same time shows that there must be an important general law underlying them , which , as I have endeavoured to show , must be that on the average the commonness of species depends on their age , and is independent of their local adaptation .
The latter will in each individual case have a determining effect in the degree of commonness ultimately reached , but we have no ground whatever for making statements to the effect that a given species is common and is therefore well adapted .
Probably mere chance has also a great deal to do with the commonness of a given species in any particular place .
Relative Age of the Monocotyledons Ceylon .
But now , if we accept these estimates of Trimen as reliable , as we are obviously bound to do , and if we accept the deduction that I have made , that commonness on * They contain 29 .
8 per cent , of endemics against an average of 28 .
8 per cent. , so that their rarity should be a fraction above the average .
the whole depends on age , we may test many other points .
Some of these will be dealt with in subsequent papers , but we may give two instances here .
For example , if there be any truth in the current theory that the Monocotyledons are younger than the Dicotyledons , then , unless the former group was developed in or near Ceylon , so as to start there as soon as the latter , it should be the rarer group .
By comparing the plants of these two groups , the rarity of the Dicotyledons as compared with the Monocotyledons proves to be Table XXI .
Dicots .
Monocots .
In the Ceylon endemics as ... ... ... ... ... ..4*31 to 4'46 Ceylon and P. India species . . . .
3'44 3*65 Widely distributed species . . . .
3'01 3'08 The Monocotyledons are the rarer , and therefore perhaps the younger , in each case .
But when we add up the totals of each kind , such is the greater proportion of endemic Dicotyledons that the Dicotyledons as a whole have a rarity of 3*49 to 3 '47.On the other hand , this greater proportion of endemic Dicotyledons may be used as an argument that they are the older group , and have therefore had more time to give rise to endemics .
We find in fact : Table XXII .
Species of wide distribution .
Ceylon and P. India .
Ceylon .
Total of first two .
Total of last two .
Dicotyledons . .
1004 396 623 1400 1019 Monocotyledons .
504 96 186 600 282 The widespread Dicotyledons which have reached Ceylon are accompanied in Ceylon and Peninsular India by more endemic species than their own number , the Monocotyledons by merely 65 per cent , of their own number .
Or we may better put it that the Dicotyledons of Wide and of Indian distribution have given 44 per cent. , the Monocotyledons 31 per cent , of their own number , of Ceylon endemic forms .
On the whole , then , the evidence is in favour of the greater age of the Dicotyledons in Ceylon , though of course no great stress can be laid upon it .
At least , however , it does not go against that hypothesis .
The Value of " In cases where we get two large and well-defined groups in a family , we may compare their degrees of rarity , when the difference between them is what is usually DR. J. C. WILLIS ON THE ENDEMIC FLORA OF CEYLON .
looked upon as an adaptation .
For example , in the Rubiaceae it is usually supposed that the sections with fleshy fruits are more recent than those with dry .
But on the other hand the former is supposed to be an adaptation to enable the seeds to be easily dispersed , which should tend to make them more common , and one would expect them to be on the whole at least as common as the dry fruited species , whereas on the theory here enunciated , if the dry fruit be really the older , it should be the commoner .
In actual fact the figures show : Table XXIII .
Number .
Marks .
Rarity .
Dry fruit 62 218 3-5 Fleshy 76 285 3-7 Or , taking only the Ceylon endemic forms\#151 ; Table XXIV .
; Number .
Marks .
Rarity .
Dry fruit 30 124 4'1 Fleshy 41 182 4-4 It is evident that the fleshy fruit has not proved of any special value to its possessors , and is probably a more recent development .
There are many other so-called adaptations in many other families which can be dealt with in the same way .
Always the figures go to show that evolution of forms is on the average indifferent , and that these so-called adaptations are of no special advantage to their possessors .
Possible Size of Mutations .
Many of the advocates of mutation hold to the view that mutations must of necessity be what they call " small , " or such differences as are , for example , exhibited by the small species into which Jordan divides the Linnean species verna .
This seems to me to be an unnecessary handicap to the theory of mutation , and I propose that it should be replaced by the hypothesis that no specific change is too great to appear in one mutation .
In the first place , we have no criterion to go by , by which to affirm that a certain specific difference is " small " and another " large .
" We have no right to say , for example , that if a leaf of one species is simple and of another compound , this is a larger difference than if one is pinnatifid , the other pinnatipartite .
We have not the least idea whether the changes in internal construction of the nucleus necessary to form Jordanian species are in any way different from , or smaller or larger than , those necessary to give Linnean species .
The differences between many of the very rare endemic Ceylon species are large , and these species are accepted as well-marked Linnean species by both Trimen and Sir Joseph Hooker .
Even in an endemic genus this is the case , as , for instance , in Doona or Stemonoporus , where a great number of species are very rare , and only represented by a few plants each .
When we reflect upon the facts that have been brought forward in this paper it is clear that we have no right to invoke extensive killing out of intermediate forms .
It is a great deal more difficult to define the difference between a Linnean and a Jordanian species than one is at first inclined to imagine .
' In such a plant as D. verna , where the Linnean species is made by drawing a line round a number of forms at the point of their greatest separation from allied forms , it is comparatively simple ; but all the Linnean species are not like this , as we have just pointed out in the cases of Doona and Stemonoporus .
The species in which Jordanian splitting can be applied are just such species as D. verna , common species of wide distribution .
This we have just seen to imply considerable age , and it is just as probable that there was originally a single well-separated form to which the name D. verna could be applied , which subsequently broke up into many forms , as not .
We must simply take account of all definite and hereditary differences , whether we consider them large or small .
Every one appears to imply a mutation , but whether some mutations are large and others small , we have no idea , for we do not know in what a mutation really consists .
It seems to me better to adopt the hypothesis that any specific difference may appear at one step , whether it be large or small .
But we may go further than this , and claim that even " larger " differences than any we have as yet discussed may also arise at one step .
For instance , the endemic Coleus elongatus on the top of Ritigala differs so much from all other Colei in its equally toothed calyx , and raceme-like inflorescence , as well as in other points , that it must probably be regarded as almost , if not quite , subgenerically distinct .
Yet the whole species is confined to the summit of this one mountain and exists there as about a dozen individuals , a number which can never have been much exceeded , if at all ; and it must in all reasonable probability have arisen there at one step .
But even with the formation of a sub-genus the possibilities of single mutations do not cease .
Of the Ceylon endemic genera no less than 17 out of 23 have only one species each , and these have a rarity of 3*7 , showing that on the whole they are probably older than the endemic species as a whole .
But as there is only one species in the genus , this species must presumably have arisen at one step , forming the genus at the same time .
We are thus led to suppose that the distinction between genus DR. J. C. WILLIS ON THE ENDEMIC FLORA OF CEYLON .
and species is really more or less artificial , depending upon our ideas as to what are large and what are small changes/ " The distinction is extremely useful , and so long as further mutations , or turns of the kaleidoscope , do not interfere with those characters which we regard as generic , or impose upon them new " generic " characters , the descendants of the first species will continue to belong to its genus .
In the case of the Ceylon endemics we seem to have caught most of the genera still in the stage of one species each , but a few have two or three , and two , Doona with 11 species , and Stemonoporus with 15 , have gone a stage beyond this .
The fact that in widely separated islands , such as the Hawaiian , the proportion of endemic genera is larger , though the number of endemic species is not so large , would suggest that perhaps " generic " differences sometimes arise by the superposition of specific differences affecting the same character .
The fact that in Ceylon there are six families , five with one species each , and one with two species , with all species endemic , goes to show that it is possible for all representatives of a family to become endemic .
When we turn to the genera in the other families we find quite a number in which all the species are endemic , which give the following Table :\#151 ; Table XXV .
VC . . .
3 HR . . .
33 C ... 24 R ... 34 RC . . .
31 VR . . .
44 Total 169 species , with mean rarity 4'20 , or slightly more common than the endemic species as a whole .
The number of genera with one species only is 63 , with two 10 , with three 6 , with five 5 , with seven 3 , with nine 1 , and with thirteen 1 .
There is a marked difference in rarity between those genera with few , and those with many , in these cases with all the species in a genus endemic .
When the genus contains one or two species only , the rarity is 4 , when it contains more than two the rarity is 4'3 .
This is exactly what we found in the endemic genera , and would also suggest that genera may perhaps tend , after a certain period , to break up into many species .
It would also appear to follow that there is no evidence that the same mutation must go on appearing in order to establish itself , f Many of the Ceylon endemics in the VR class , e.g. , Coleus elongatus , on the summit of Ritigala , are represented by only a very few individuals , and cannot so far as one can conceive ever have had more .
Whether a species begins as one solitary individual one cannot say ; more probably , perhaps , it begins as the seeds of one individual , which may come to nothing , or may give rise to one or many individuals of the new species .
But there appears no reason to suppose that the same mutation must appear in several * Or , more correctly , what are more , and what less , permanent changes , t Cf.de Vhies , ' Mutation Theory , ' English edition , p. 208 .
VOL. CCV1.--B .
2 successive generations in order to become established , though , of course , in that case its chance of becoming established will he greatly increased .
Evolutionary History of a Species .
We may now go on to consider the light which these observations throw on the probable evolutionary history of a .
species .
Mutations being granted as the only available method of origin , the species must have sprung directly from some other species , and natural selection being reduced to very secondary importance , there is no reason why this other species should be killed out .
As the numbers of endemic species increase from VC to VR , this would indicate that they began at one or other end of this scale , and not somewhere in the middle .
Now it is highly improbable that they began at the VC end , for this would indicate that they are now dying out , and it would seem a most remarkable thing if species actually developed in Ceylon should not even have enough local adaptation to be able to survive there .
It is far more probable that they began at the VR end of the scale , and this is a view strongly supported by the fact that there are amongst them a great number of species confined to one mountain top or to some very small area .
No fewer than 108 species# are thus confined to one mountain top or to small areas in the mountains .
Many of these only occur , as , for instance , those on the top of Ritigala , f as a very few individuals ; they must have been evolved on the mountain top where they are found , as I have shown elsewhere ; and they can never have existed as many more individuals than at present .
Each species then , we may suppose , began as VR , as one or a few individuals in some very limited area .
In the case of C. elongatus , which is confined to Ritigala summit , the whole plant , so far as I could discover , is represented by a few individuals ( not more than a dozen or so ) , and as there are several other endemic forms on this mountain top , whilst the area of ground suited to the Coleus is very limited , there cannot ever have been many more .
Beginning thus , the species , we may take it , either spread , just held its own , or died out .
We may represent what occurs , perhaps , by such a diagram as that on p. 333 .
If the species appears first as several individuals ( a dozen or more ) , then the line through VR should be the horizontal one .
Some of the new species developed will probably be so much worse adapted to the local conditions ( which are the only ones that matter for a long time ) , or will meet with such unfavourable chances , that they will die out , and their progress will be indicated by the straight line A\#163 ; .
Others will be able to creep up in the direction of VC .
How far they will get on this line will depend not only on the degree of local adaptation with which they set out , but also on numerous such chances as , for * " Floras of Hilltops in Ceylon , " 'Ann .
Perad .
, ' vol. 4 , p. 131 ( 1908 ) .
t " Flora of Ritigala : a Study in Endemism , " loc. cit. , vol. 3 , p. 271 ( 1906 ) .
DR. J. C. WILLIS ON THE ENDEMIC FLORA OF CEYLON .
example , the arrival of a disease to which they may be more than usually subject .
The first species of a group starting in a new country will perhaps mostly be able to reach a greater degree of commonness than they would now reach , whilst the least .common of them , which would then be described as VR , would perhaps be about as common as those now called R. Each new arrival would perhaps find it harder to reach VC , and take longer to do it .
vc ( Appearance of new species ) ( Extinction of old species ) Now , of the species that thus start to climb , it cannot be said that they begin with the certainty of reaching any definite point , such as RC , for as they climb they may come under the influence of new factors which may make great differences .
A species that looked as if it was going to reach C may be compelled to stop at RR .
The element of chance must be enormous , though natural selection is probably the ultimate deciding factor in the position finally reached .
Ultimately , however , each species will reach a maximum height on the curve .
It may stay there a long time , or may go up or down with the appearance on the scene of new factors affecting it .
But as a good many of the newer species which come into being and climb the curve will tend to be a little better adapted* than those that preceded them , we may take for granted that in the great majority of cases the species , however high it may have climbed , will ultimately begin to descend .
Now , if the zenith reached has only been YR , which , we may take it , will be the case with about a sixth of the species\#151 ; whether owing to poor degree of adaptation , or simply to ill chance , or late arrival on the scene\#151 ; the species will always be in a precarious position .
For example , a serious fire on the summit of Ritigala , or a forest clearance for planting on the summit of Hinidunkanda , in Ceylon , would in all likelihood exterminate four or five species .
How many have thus been * On the average perhaps half will be better , half worse , adapted than the average of their forerunners .
exterminated we have no means of knowing , but it does not seem a priori probable that it must have been many , as we may infer from the figures in this paper that species increase or decrease in commonness at a very slow rate under natural conditions , and we have also seen that there is probably little difference in adaptation to local conditions amongst them .
But the point is , that while at YB or below it , either on the upward or downward journey , the species is liable to extinction .
Species that have reached the stage B will be much safer from this liability to extinction , and those that have got to BB safer still , and so on in increasing certainty up the series of stages .
But in all likelihood a number of good species , well enough adapted to have reached above the average , will be exterminated at the stage YB on their upward path .
Beaching its zenith on the curve , then , and remaining there for a longer or shorter time , a species will sooner or later begin to descend on the other side , and ultimately will reach YB.* Now , if there has been nowhere any serious lack of local adaptation , or serious local mischance , the species will , as it descends the scale , tend to more or less shrink its area in such a way as not to leave unoccupied gaps therein .
But if it is , for example , YC in several countries , and B or YB in one or two countries lying among these , then , if the same causes of descent are operating in all , it may tend to disappear sooner in some than in others , and so give a disjointed area .
Once the species reaches YB or below that point , it runs the risk of being any day exterminated , just as it did when passing that point on the upward curve , and sooner or later , in the case of a descending species , it will die out .
The killing out of the species at the stage of YB or below it , with the death of those which are so badly suited to the local conditions that they go along the line AZ , will tend to widen the gaps between the survivors , so that at last not only will genera ( in some cases ) be formed , but also families , and larger groups .
But as we have already pointed out , these , or , at any rate , genera , may also be formed by single mutations on the upward curve .
Instead of the old view that the ancestral forms of the now existing species are all extinct , we must now take the view that there is no very particular reason why they should not exist on the earth at this present moment .
As to which species is the ancestor of which we can only sometimes hope to be able to give a definite well-founded opinion , but there can be little doubt that we shall someday find out the principles on which evolution works , sufficiently well to be able to commence breeding new species .
The adoption of this view renders the time necessary to evolve some of the families of plants at present existing distinctly less than on the old theory , and the new view has this additional point to recommend it .
As most of the families of plants now existing occur in both hemispheres , they are probably very old .
But they were not * In a subsequent paper it will be shown that there is little evidence that any species are dying out .
DR. J. C. WILLIS ON THE ENDEMIC FLORA OF CEYLON .
then necessarily families ; that has perhaps come about by superposition of mutations , and dying out of types .
Geographical Distribution in General .
We may now go on to point out what is one of the most important bearings of the facts that have been brought forward , and which are so striking and consistent that they demand some explanation , even if that which I have given be not accepted .
In a paper published over seven years ago , * I pointed out the effect upon the current theories of geographical distribution which an acceptance of mutation without natural selection would have .
I then chose the Dilleniaceae for no other reason than that it was the first order in Hooker 's Indian Flora with distribution less than world-wide .
In the present place I take the Menispermaceae because the monograph in Engler 's ' Pflanzenreich ' happens to be lying on my table .
Nearly all families have the same general type of distribution .
In the paper referred to , I pointed out that there was no particular reason why the whole tree of descent of a family should not exist on the earth at the present moment , and I suggested that all the Dilleniaceae might be descended from Tetracera , the most widely distributed genus in the family .
This idea that the most widely distributed genera are the most primitive , rather than necessarily the best adapted , receives great support from the facts which have been brought forward in the present paper .
It is true that these facts have chiefly been used to show that local commonness , on the whole , increases with age , but as the commonest species are those of wide distribution , the next commonest those developed somewhere in Ceylon and Peninsular India , and the least common those confined to Ceylon , we may take it as extremely probable that wide distribution also goes with commonness , and therefore with age .
It hardly needs pointing out that quite possibly some of the species that are now of wide distribution originated in Ceylon in very early times .
The Ceylon endemics have not yet nearly reached the stage of average distribution ; only , perhaps , when they have become so old that the difference in age between them and the other species has become negligible , will they do so .
The facts of distribution would seem to indicate that the time still required to reach average dispersal would be at least , as long as the average time between the arrival of the widely distributed species and those also found in Peninsular India , whatever time that may have been .
As the Ceylon endemics climb the scale some of the widely distributed species will be forced to descend .
One may , perhaps , imagine that the various species that occur in Ceylon will all the time be becoming more and more accustomed to the local conditions , by natural selection acting on fluctuating variation , but in any case this cannot amount to much , or cause any structural differences between them .
* " The Geographical Distribution of the Dilleniaceae , as Illustrating the Treatment of this Subject on the Theory of Mutation , " 'Ann .
Perad .
, ' vol. 4 , p. 69 ( 1907 ) .
It being , then , assumed that not only , on the whole , do age , and commonness go together , but also age and wide distribution , we may now proceed to follow out the probable course of development of the different systematic groups .
According to the hypothesis proposed in my former paper , taken together with the interesting facts which have been set out in the present contribution , a species probably starts on a small area as one or more individuals , and from that area spreads by aid of its dispersal mechanism .
Under the influence of natural selection , modified very much at first ( i.e. in the earlier stages ) by chance , but less in proportion later , the species will spread , gradually occupying a larger and larger area of the earth 's surface , so long as there remain places suited to it within reach .
Sooner or later , in perhaps a large proportion of the cases , it will give rise somewhere or another to another species , which again will commence as a single one or as a few individuals in a definite spot .
If this spot were always near the centre from which the original species started , we should on the average , after some time , get a series of concentric circles , representing the various species developed , the largest circle representing the oldest species .
But in actual fact we know that new species may .develop , as they must have developed in Ceylon , near to the margin of the area occupied by the parent species .
Further , while each new species on the average will be at approximately the same degree of adaptation as its forerunners , this will rarely happen exactly in any individual case , nor will the effects of chance be the same , so that the size of the area occupied will , in comparatively few cases , indicate the exact age of the species as compared with others of the same genus or family.# It will generally happen that no two species are exactly alike , and consequently that while the " circle " of species 2 will for a long time be enclosed in that of species 1 , it may sometimes exceed it all round , or overlap it at one place , where the conditions of life suit 2 but not 1 .
Also , obviously , as the circles spread , the mere effect of climate will tend to make them spread more to east and west than to north and south .
We may thus presently get a distribution diagram something like the following :\#151 ; We have seen above that the origin of genera appears to be similar to that of species , and in the gradual dispersal of a family it is obvious that that of the * And , of course , a species will some day occupy a maximum area beyond which it will not go .
DR. J. C. WILLIS ON THE ENDEMIC FLORA OF CEYLON .
genera* will follow the same rule as that of the species , so that , after a time , the family , now supposed to contain three genera , may show such a distribution as the following:\#151 ; Of course , in actual fact , the dispersal will not be so simple .
Even in one country , the climate and other conditions will vary from place to place , so that the actual distribution can very rarely be a circle or even an ellipse , unless just at the start in such a flat country as Russia .
As the forms spread , their dispersal will be altered by geographical obstacles such as chains of mountains , arms of the sea , etc. , and geological changes will also affect it .
But the underlying principle will remain the same .
Now , when one comes to examine the distribution of almost any order , one finds that it may be very well explained on such principles .
I have here taken the Menispermaceae , but the distributions already given of the Dilleniaceae and Podos-temaceae , etc. , in other papers agree perfectly well with it , and , in fact , so do most orders .
The following rough map gives sufficient to show the general outline of the dispersal of this family .
Geographical Distribution of Menispermacese in outline .
At X ( West Africa ) 12 genera .
At Z ( Madagascar ) 5 genera .
M = Menispermum .
* Genera will obviously tend to follow the age and area rule more closely than species .
The map here given shows that two genera , Cocculus and Cissampelos , have a distribution which is almost exactly that of the order , though each overlaps the other to some slight extent in some places .
One or other of these genera , on my hypothesis , is to be regarded as the original ancestor of the existing Menispermaceoe ; it is noteworthy that they both belong to the same tribe , the Cocculeae , and are chiefly distinguished in their respective sub-tribes by a mere difference in the number of the carpels , a matter which might easily be the subject of an early and simple mutation .
These genera , and probably Menispermum , which occurs on the eastern sides of Asia and North America , as do so many genera , must have commenced when there was still a possibility of reaching both the great land masses .
Later , or at least not in time to spread from one of the great continents to the other , came the genera Stephania and Tinospora in the eastern , and Hyperbsena in the western , continent , and , subsequent to these again , the groups of genera which are roughly indicated on the map , e.g. , three genera occurring from West Africa to Madagascar , five genera in the Malay Archipelago , five in Brazil , and yet later the groups of genera that are confined to one locality , e.# .
, 12 in West Africa , 5 in Madagascar , 1 in North East India , and so on .
There is nothing surprisingly new about these views , but in 1907 was the first time , to my knowledge , that they were put forward in so many words , and a proper understanding of them will make a great difference in the handling of problems of geographical distribution .
Hitherto , we have assumed that the distribution of a genus was largely determined by its degree of adaptation , and , still more important , we have assumed that the appearance of two or more new genera has meant the disappearance of ancestral forms , so that it became a matter of extraordinary difficulty to trace the history and distribution of any family .
But if we admit that one species or genus may arise directly from another by means of mutation , and also admit that natural selection has but little value , so that there is no reason why the parent form should be killed out ( almost a necessity with the old theory ) , we simplify enormously the dealing with the problems of geographical distribution , and get a standpoint for treatment which will greatly help us in determining the relationships of species , genera and other groups .
I hope to follow this paper with others , in which I shall deal at greater length with various problems which have only been touched upon here , and especially with adaptation .
This paper , it seems to me , makes a step upon the road to a sound common-sense non-teleological theory of evolution .
It is very difficult to escape from teleology and even now far too much stress is laid upon adaptation .
The figures given in this paper , showing that all families are much alike , will help to check this .
When one comes to look into it , there is really very little special adaptation in plants , and once that a plant has specialised into root , shoot , leaf , inflorescence , and flower , further specialisation goes for very little , except in special cases .
Xerophytes are more or less specialised for dry climates , hydrophytes for wet places , but the vast majority of plants are mesophytes , and within the range of temperature to which they are suited , DR. J. C. WILLIS ON THE ENDEMIC FLORA OF CEYLON .
one will do about as well as another in any given place , provided both come from somewhat similar climates , etc. The present paper goes to show for the first time that evolution itself , as has already happened to breeding , can be brought within the range of arithmetic , and that the processes that accompany the dispersal of species and genera about the globe can be expressed in simple terms , without serious complications arising from different " adaptations .
" In a later paper I hope to bring forward the tentatives towards a theory of evolution which are suggested by the work in this paper and in a previous one upon lack of adaptation in Podostemacese .
Summary .
Ceylon contains 149 families , 1027 genera , and 2809 species of flowering plants ; 91 families and 324 genera ( of which 23 are endemic ) contain 809 endemic species .
Six families with 7 species have all species endemic ; 4 families with 91 species out of 102 have over 75 per cent. , and 14 with 255 out of 435 have over 50 per cent , of their number endemic .
Trimen divides these according to rarity into six classes : Very Common , Common , Rather Common , Rather Rare , Rare , Very Rare ( Table II ) .
Marking these from 1 to 6 , the mean rarity is 3'5 .
Dividing the flora into three groups : Ceylon ( 8-09 sp. ) , Ceylon-Peninsular India ( 492 ) , and of wider distribution ( 1508 ) , Table III shows that the endemics increase in number from 19 YC to 233 YR , and have rarity 4*3 , while the " wides " go in the other direction , with rarity 3 .
The Ceylon-Indian species are fairly evenly grouped , with rarity 3'5 .
Multiplying by factors to reduce to one level of commonness ( 'fable IY ) under each head of YC , etc. ( or taking percentages ) , brings this out very clearly indeed .
Thus the rarest plants in Ceylon are the local endemics , the next most rare those also common to Peninsular India , and the commonest those of wider distribution .
The rarity of the endemics as a whole is not due to special rarity of any one family .
Taking groups of not less than 14 species , it is wonderfully uniform , ranging only from 3*9 ( above the average for the whole flora ) to 4'9 ( Table YI ) .
There is little to choose between one family and another on the average of a reasonable number of species .
This is also shown by comparing the rarities of the whole list of families containing over 20 species , which , with one exception , the Dipterocarpacese ( 47 species out of 48 endemic ) , runs only from 2'9 to 4T , or equal distances on either side of the mean .
The families with the greater rarity have most endemics , and allowing for this the range of commonness goes over a less distance .
The comparison may also go into the genera .
Adding up the species of those that contain species of all three types of distribution ( Table IX ) we find rarities 4 *2 , 3*4 , and 2'8 ; in the 47 genera that contain one endemic and one non-endemic species , the rarities are 4 and 2*7 ( fables XIE , XIII ) ; in the genera with one species of each VOL. CCVI.\#151 ; B. 2 Z distribution , of which there are seven only , the rarities are 4'8 , 3'2 , and 3*2 .
The rale comes out with the regularity of Mendel 's law .
Incidentally , a comparison of the species that occur in the dry zone , in the wet zone , or in both , shows that the two former have rarities of 3*8 , the latter of 2*3 .
From these interesting figures one may conclude that the local endemic species have not been developed in any kind of advantageous response to local conditions .
They are much rarer than those species which are also common to Peninsular India , and these a^ain than those of wider distribution .
The endemics increase in number to YE , the species of wide dispersal towards VO .
There has been ample time to become VC if really adapted to local conditions , yet only 19 out of 809 have done so .
The endemic species in families that have many are rarer than in families that have few .
Natural selection requires that many forms shall modify in the same direction , and this cannot have been the case with many of these forms , which only occur as a very few individuals in a very restricted area .
The endemic genera ( 23 in number ) have 52 species with a mean rarity of 4*5 , or greater than that of the endemic species as a whole ( Tables XVII-XIX ) .
The genera of one species among them are the commonest , while Doona with 11 species shows a rarity of 4*6 , and Stemonoporus , with 15 , one of 5*4 .
These facts cannot be explained on any theory of adaptation .
Table XX shows these facts summarised , and goes in the exact reverse direction all through to what it should follow if the theory of natural selection could be relied upon .
The second conclusion drawn is that on the average the commonness of a species depends ujjon its age from the time of its arrival in , or evolution in , the country .
The commonness of any individual species will , of course , also depend upon its degree of adaptation to local conditions , and upon many things which can only be regarded as chance , such as the sudden appearance of new factors , like diseases , etc. , in the problem .
In other words , on the average , species are developed quite indifferently to local conditions , though it is possible that they may be developed because of those conditions .
As examples of the possible use of these statistical methods , it is then shown that the evidence on the whole favours , so far as Ceylon is concerned , the theory of the greater age of the Dicotyledons than of the Monocotyledons , and that the Eubiacese with fleshy fruit are rarer than those with dry fruit , from which they are usually supposed to have been derived , and that consequently their " adaptation " to dispersal by animals has proved of no value to them except in special cases .
The possible size of mutations is then considered , and it is proposed as an hypothesis that no specific difference is too " large " to appear as a single mutation .
There is no reason to suppose that the well-separated species of the Ceylon DR. J. C. WILLIS ON THE ENDEMIC FLORA OF CEYLON .
genera with many endemics owe their separation to the destruction of intermediate types .
Sub-genera and genera may apparently arise in the same way as species by one step , and the distinctions drawn between Jordanian and Linnean species , sub-genera , genera , etc. , seem to be to a large extent artificial , though in the highest degree convenient in practice .
There seems to be no evidence that the same mutation must go on appearing in order to become established .
Many of the Ceylon endemics are represented by a very few plants , and cannot ever have had more .
It is probable that species first appear as one or a few individuals by a sudden mutation , in a very restricted area .
The diagram on p. 333 gives an indication of the subsequent history of a species .
It may progress straight to extinction , or may rise to one of the stages VR , R , etc. But ultimately better adapted forms will come into competition with it , or other chances will affect it , and it will begin to go downwards on the curve .
Finally , it will reach Vft , and then or shortly afterwards will become extinct .
Species do not , so far as we can tell , appear in any sort of advantageous response to local conditions , which are the only conditions that matter when they first appear .
Having appeared , a species will , or will not , spread , according to its suitability to local conditions .
But this will be enormously affected by chance , especially at the first .
However well a species may be locally adapted , it will be very liable to extermination until it has got beyond the degree of commonness represented by VR .
In each locality the ultimate commonness of a species will depend upon its degree of adaptation to the local conditions , and , to a large extent , on chance .
But most species can thrive in a considerable variety of conditions , and there is not the slightest reason to suppose that many Ceylon endemics would not thrive over wide areas of the earth 's surface , like the very local species macrocarpa which is now perhaps the commonest Conifer in the sub-tropics .
But the commonness of a species in a given place at any previous period will mainly depend upon its age .
There is no evidence that local species are specially adapted to local conditions .
Having reached the maximum height that it is going to reach , a species will ultimately descend , and will sooner or later be extinguished , though there is no evidence that as yet many , or any , species are on the downward road .
These disappearances of species will tend to widen the gaps between the survivors , which may also be widened by the superposition of mutations upon one another .
The bearing of the facts here brought up upon the hypothesis of geographical distribution under mutation and without natural selection that 1 proposed some years ago is then considered .
They support very strongly the hypothesis that the whole tree of descent of a family may exist on the earth at the present moment , and that the area occupied is in general an indication of the age of the species or genus , if it has not already attained its maximum .
The Menispermaceae are taken as an example , as were the Dilleniacese , Tristichacese , and Podostemaeeae in previous papers , and it is suggested that the family is descended from Cocculus or Cissampelos , the most widely distributed genera which it contains , and whose distribution is very much that of the whole order .

