Age and area; a study in geographical distribution and origin of species

NORTH CAROLINA STATE UNIVERSITY LIBRARIES

S02842957

This book is due on the date indicated unless
recalled by the Libraries. Books not returned on
time are subject to replacement charges.
Borrowers may access their library accounts at:
http://www.lib.ncsu.edu/ads/borrow.html

AGE AND AREA

CAMBJIIDGE UNIVERSITY PRESS

C. F. CLAY, Manager

LONDON : FETTER LANE, E.G. 4

LONDON : H. K. LEWIS AND CO.. Ltd.,

136, Gower Street, AV.C. I
LONDON: WHELDON AND WESLEY. Ltd..

28. Essex Street, Strand. W.C. 2
NEW YORK: THE MACMILLANCO.
BOMBAY \

CALCUTTA \ MACMILLAN ANDCO.. Ltd.
MADRAS J
TORONTO : THE MACMH,LAN CO. OF

CANADA, Ltd.
TOKYO : MARUZEN-KABUSHIKI-KAISHA

ALL BIGHTS KESER\-ED

AGE AND AREA

A STUDY IN GEOGRAPHICAL DISTRIBUTION
AND ORIGIN OF SPECIES

BY

J. C. WILLIS

M.A., Sc.D., Hon. ScD. (Harvard), F.R.S.

European Correspondent, late Director, Botanic Gardens,

Rio de Janeiro

WITH CHAPTERS BY

HUGO DE VRIES, F.M.R.S.
H. B. GUPPY, M.B., F.R.S.
Mrs E. M. REID, B.Sc, F.L.S.
JAMES SMALL, D.Sc, F.L.S.

[These authors are not committed, by writing these chapters,
to the support of all the doctrines here advanced]

CAMBRIDGE

AT THE UNIVERSITY PRESS

1922

,%

^

a

^y^h

4

PREFACE

Some thirty years ago, a pupil of the strictest school of natural
selection, and enthusiastic in my belief in its principles, I set
out upon a course of independent observation of nature. Ten
years of such work convinced me that a simpler explanation of
phenomena was always to be found, and one that seemed more
in accordance with the facts; and I endeavoured — with what
success this book will show — to free myself from the trammels of
the natural selection theory, and to work as if I had found myself
in another planet where scientific investigation was just begin-
ning. Stationed in one of the best centres in the tropics (where
the phenomena of distribution are more impressive than in
Europe), badly handicapped in laboratory work by a serious
accident, and finding my chief pleasure in travelling about the
world to see its vegetation — I took up the study of distribution,
in which I had always taken much interest.

Here, as elsewhere, it was soon evident that the current
theories pro^'ided an explanation that was not only unnecessarily
complex, but one that did not explain. As one of my critics
words it, "for some reason the plant has advantages which
enable it to spread"; and beyond that point we cannot go.
Gradually it became clear to me that plants spread very slowly,
but at an average rate determined by the various causes acting
upon them, so that age forms a measure of dispersal when one
is dealing with allied and similar forms.

Age as an explanation of spread is enormously simpler than
natural selection, and that it is probably valid is shown by the
way in which it can be used for prediction. An opponent re-
marks that "it is too simple to be true," but this very simplicity
seems to me a strong reason in favour of its adoption, at any
rate as a preliminary hypothesis. Of two explanations take the
simpler, is an old rule, and as Hooker has said, "no speculation
is idle or friiitless, that is not opposed to truth or to probability,
and which, while it coordinates a body of well-established facts,
does so without violence to nature, and with a due regard to the

vi PREFACE

possible results of future discoveries." To find explanation of the
facts of distribution under the current theories has always
seemed a very hopeless task, and any hypothesis that offers a
way out should at least receive attention. No hypothesis can,
after all, alter the facts, though it may show ways in which to
accumulate new ones.

In the second part of the book, I have pushed my hypothesis
to what seem to me its logical conclusions, conclusions which
are sometimes subversive of received opinions. To be compelled
to re-examine the bases upon which those opinions are founded
will do science no injury, however.

While the defects of the book are my own. I owe what is good
in it very largely to the constant help, advice, and criticism of
many friends, among whom I would specially mention Dr Hugo
de Vries, Dr H. B. Guppy, Mrs E. M. Reid, and Prof. James
Small, all of whom have also contributed chapters to the work.
To these four I must add my friend Mr G. Udny Yule, to whose
trained mathematical skill I owe much useful help and criticism.
Prof. J. Stanley Gardiner has helped me very greatly in the work
upon animals. In particular he was so kind as to obtain for me
the help of Dr Hugh Scott, who spent hours with me in counting
beetles, Mr E. Meyrick, F.R.S., who gave me figures for dis-
tribution of Micro-Iepidoptera, Mr G. C. Robson, and Dr W. T.
Caiman, F.R.S. To the criticism of Prof. E. S. Prior, A.R.A., I
largely owe the present simplified form of the book, and its
freedom from technical terms; he was also so kind as to obtain
for me the aid of Dr W. D, Lang. References to literature, and
other valuable help, I owe to Sir David Prain and Mr S. A. Skan
at Kew, Mr G. Goode, ]\I.A,, at the University Librarj^ IVIiss
Taylor at the Balfour Library, and others, whilst I am also
deeply indebted for help and criticism to (the late) Dr E. A. N.
Arber, Mrs Agnes Arber, Prof. Margaret Benson, ^Ir E. Breakwell,
Dr W. B. Brierley, Dr N. L. Britton, Dr J. Brownlee, Mr J. Burtt-
Davy, Dr L. Cockayne, (the late) Mr R. \V. Davie, Mr C. E.
Foweraker, Mr E. G. Gallop, Prof. R. Ruggles Gates, Dr B.
Daydon Jackson, (the late) Dr A. Lofgren, Dr D. T. MacDougal,
Dr J. H. Maiden, Miss E. R. Saunders, Dr D. H. Scott, Prof.
A. C. Seward, Mr A. M. Smith, Dr Norman Taylor, Dr R. J.

PREFACE vii

TiUyard, Prof. A. Wall, Dr J. E. B. Wanning, Prof. D M S
Watson, and many others. That I have been able to carry out
this work at all I owe to the labours of generations of systematists,
botanical and zoological, foremost among whom, inasmuch as
the hypothesis of Age and Area was originally founded upon
their work, I must place my predecessors in Ceylon, G. H. K.
Thwaites and Henry Trimen. I must also specially mention
Sir Joseph Hooker, as this work forms a continuation of his
labours of the fifties. Last, but not least, I am deeply grateful
to my wife, and to my relatives, Mrs and Miss Steel, Vo^'r much
help ungrudgingly given.

For illustrations I am much indebted for loan of blocks to the
Royal Society, and to the Editors of the Annals of Botany, Nature,
and Nezv Plnjtologist: also to my daughter Margaret, who made
the drawings from which all, except those on pp. 125, 153, 173,
241 and 242, were jDrepared.

J. C. WILLIS.

Cambridge,

4 April, 1922.

CONTENTS

PART I. THE PRESENT POSITION
OF AGE AND AREA

CHAP. PAGE

I. Introductory 1

II. The Dispersal of Plants into New Areas . 10

III. The Introduction and Spread of Foreign

Species 21

IV. Acclimatisation 29

V. Causes which favour or hinder the Dis-
persal OF Species 32

VI. Age and Area 54

VII. Age and Area {contd.). Confirmation by

Prediction 66

VIII. Age and Area {contd.). Invasions ... 70

IX. Objections to the Hypothesis ... 84

PART II. THE APPLICATION OF AGE AND AREA

TO THE FLORA OF THE WORLD,

AND ITS liSIPLICATIONS

X. The Position of the Age and Area Theory 101

By H. B. GUPPY, M.B., F.R.S.

XI. The Further Extension of the Application

of Age and Area 107

XII. Size and Space 113

XIII. Age and Area, and Size and Space, in the

Compositae 119

By Jajies Small, D.Sc, F.L.S.

CONTENTS

CHAP.

XIV.

XV.

XVI.

XVII.

XVIII.

XIX.

XX.

XXI.

XXII.

Age and Area from a Palaeobotanical
Standpoint

By Mrs E. M. Reid, B.Sc, F.L.S.

Endemism and Distribution: Species .

Endemism and Distribution : Genera .

The Monotypic Genera, and Genera of
Larger Size ......

The Hollow Curve of Distribution

Applicability of Age and Area to Animals

The Origin of Species ....

Age and Area and the Mutation Theory
By Hugo de Vries, F.M.R.S.

Geographical Distribution: General .
List of Literature ....
Index

137

148
169

185
195
200
204
222

228
247
253

The illustrations on pp. 56, 66, 76, 78, 79, 80, 153, I owe to the courtesy
of the Editor, Annals of Botany.

PART I

THE PRESENT POSITION
OF AGE AND AREA

CHAPTER I
INTRODUCTORY

1 HE existing distribution of a plant (or animal) upon the surface
of the globe, which is often a very complex phenomenon, is due
to the interaction of very many factors. Sometimes they are
inherent to the plant itself, sometimes they are incidental to its
surroundings, sometimes they partake of both qualities. At
times they may be active, at others ver}^ active, and at some
periods, or in some places, they may be more or less quiescent.
One pulls in one direction, another in another. As a plant spreads
from the place in which it originally commenced, therefore, it
comes under an ever-varying pull, causing it to spread more or
less rapidly, or at times not at all, according to the different and
ever-altering combinations of these factors — different climates,
different soils, different groups of plants that occupy the soil,
presence or absence of such barriers as are offered l)y mountains,
seas, changes of climate, and many other things. To all this it is
obvious that age must be added — the older the species is, the
more area Avill it have had time to cover.

But mere spreading is not all; a species may at one time be
common in a certain region, and at a subsequent time may be
very rare, or even non-existent there. This may be due to many
things, for example, the arrival of a disease-organism to which
the plant may be very subject, and to which it falls an easy prey,
or which so reduces its vigour that it falls a prey to something
else. Or some new competitor may appear, which is so much
better suited to the local conditions that the first plnnt is reduced
to rarity or perhaps even to extinction. In this connection by
far the most important factors are those introduced by geological
and other changes. In times which, geologically speaking, were
but yesterday, Britain was united to the continent of Europe,
and the way was open for the passage of any species that grew

2 INTRODUCTORY [pt. i

upon the latter; now it is closed, or closed to all but a few whose
seeds may be carried, by wind, birds, or man, across the dividing
seas. In Tertiary times, Europe was covered with forest in which
grew many things not now found there ; the onset of the cold of
the glacial period, and the secular changes of climate, have so
altered the conditions that the Tertiary forest has disappeared.

So complicated is the interaction of all these many factors,
and their continual changes, that in general it has been con-
sidered impossible to say why a given plant should be found to
occupy a given area, while another species of the same genus
occupies one much larger or much smaller, though it may look
almost exactly like the first, and may differ from it only in
characters to which we cannot, without great stretch of the
imagination, attach any serious importance for life or success.
We have been luiable to say why, for example, Coleus barbatus
should be found almost over tropical Asia and Africa, while
C. elongatus, which differs chiefly in the form of the calyx and
of the inflorescence, is confined to the summit of one mountain.

For sixty years we have been under the wonderful fascination
of the theory of evolution by means of infinitesimal variations,
or minute changes of character from individual to individual.
At first, and for a long period, this theory seemed to be capable
of explaining almost everything, and to it we owe what could
perhaps have come in no other way, the establishment of the
doctrine of evolution, now universally adopted, but which until
the latter part of the last century, though 2000 years old, had
met with no acceptance. To quote Huxley (22 in List of Litera-
ture, II, pp. 180, 197), "To any one who studies the signs of the
times, the emergence of the ])hilosophy of Evolution, in the
attitude of claimant to the throne of the world of thought, from
the limbo of hated and, as many hoped, forgotten things, is the
most portentous event of the nineteenth century." "...the pub-
lication... had the effect... of the flash of light, w^hich to a man
who has lost himself in a dark night, suddenly reveals a road
which, whether it takes him straight home or not, certainly goes
his way."

Under the glamour of this theory, the tendency naturally was
to lay the greatest stress upon the vital factors in distribution,
for these were the only ones which could differ from individual to
individual, or from species to species. The means of dispersal
open to plants, their reactions to the climate, etc., and their
adaptations to various ends, were therefore studied with re-

CM- iJ INTRODUCTORY 3

newed and extraordhiary vigour, whilst the mechanical factors,
except perhaps the purely negative influences of barriers, were
left comparatively neglected. For many years there was re-
markable progress in our knowledge of geographical distribution,
but this has now all but ceased, except in regard to the study of
the purely local distribution of species in reference to the purely
local changes of the different factors of climate, water-supply,
associations of plants covering the ground, and the like, in
which direction much work of extreme value is being carried' on.
But in regard to the wider general distribution of plants about
the globe, we seem to have arrived at a period when the limiting
factor, to use Blackman's words, has become the lack of a satis*^
factory theoretical background, which will provide efficient
working hypotheses for the conduct of investigations that shall
lead to real advances in our knowledge of the fascinating sub-
ject of geographical distribution. I have myself heard a leading
authority upon this subject say that he thought that it was
almost beyond the range of human capacity.

In this emphasising of the effects of the vital factors, the
action of mere age, which must evidently be of some importance,
has been more and more lost to view. And yet in 1853 Lyell
(69, p. 702) wrote

As a general rule, however, species common to many distant
provinces, or those noAv found to inhabit very distant parts of
the globe, are to be regarded as the most ancient. Numericallv
speaking, they may not perhaps be largelv represented, but their
wide diffusion shows that they have had a long time to spread
themselves, and have been able to survive many important
revolutions in physical geography.

Again he says

Nor do I doubt that if very considerable periods of equal
duration could be compared with one another, the rate of change
in the living... world might be nearly uniform.

And yet again

Every local revolution... tends to circumscribe the range of
some species, while it enlarges that of others ; and if we are led
to infer that new species originate in one spot only, each must
require time to diffuse itself^ over a wide area. It will follow,
therefore, from the adoption of this hypothesis, that the recent
origin of some species, and the high antiquity of others, are
equally consistent with the general fact of their limited dis-
tribution ; some being local, because they have not existed long
enough to admit of their wide dissemination; others, because

1—2

4 INTRODUCTORY [pt. i

circumstances in the animate or the inanimate world have
occurred to restrict the range which they may once have
obtained.

Hooker (55a, p. xxv), in the same year, 1853, quotes the first
passage from Lyell, and goes on

If this be true, it follows that consistently with the theory of
the antiquity of the alpine flora of New Zealand, we should find
amongst the plants common to New Zealand and the Antarctic
Islands some of the most cosmopolitan, and we do so in Montia
fontana, Callitriche verna, Cardamijie hirsida, Ejnlohium tetra-
gonum and many others.... On the other hand, it must be recol-
lected that there are other causes besides antiquity and facility
for migration, that determine the distribution of plants, -these
are their power... of invading and effecting a settlement in a
country preoccupied with its own species, and their power of
adaptability to various climates... though we may safely pro-
nounce most species of ubiquitous plants to have outlived many
geological changes, we may not reverse the position, and assume
focal species to be among the most recently created, for species,
like individuals, die out in the course of time; whether following
some inscrutable law whose operations we have not yet traced,
or whether (as in some instances we know to be the case) they
are destroyed by natural causes (geological or other) they must
in either case become scarce and local while they are in process
of disappearance.

It is thus clear that the subject of Age and Area is by no
means new. Until comparatively recently I was not aware of
the above very striking quotations, and it is interesting to find
that my experience of actual distribution in many lands has led
me, as it has led Guppy and many more, to much the same
conclusions as those reached by two authorities so great as Lyell
and Hooker. Had it not been for the appearance and rapid rise
of the great theory of Darwin, with its ine\'itable diversion of
effort into other and at the time much more profitable lines, it
is evident that Hooker or some other worker of an earlier time
would have discovered not only the principle which I have
termed Age and Area, but also the many and remarkable con-
clusions to which it leads.

During the last twenty years, since finishing my monograph
of the Indian Podostemaceae (116), I have devoted my spare
time to the study of geographical distribution. My studies of
that family had convinced me that the vital factors were not,
to any great extent, responsible for the existing dispersal of the
species, and in May, 1907, I published the first sketch of the

CH. I] INTRODUCTORY 5

theory that was growing up in my mind in respect to it. Almost
simviltaneously Copeland (18) presented evidence for the same
view, practically enunciating the hypothesis itself, though not
in definite arithmetical terms.

In various subsequent papers I published further suggestions
in regard to Age and Area, and the other hypotheses that I had
associated with it, but it was not until in 1912 I actually worked
over the complete flora of Ceylon with respect to local distribu-
tion that I discovered that the effects of mere age upon dispersal
were so clear and unmistakable that they could be expressed in
figures. My paper embodying these results was published in
1915, and has been followed by many others upon the same
subject.

While the distribution of any single species is due, as has been
said, to the complex interaction of man}^ factors and barriers,
it must be remembered that only in the case of a group of allied
species will these be likely to act with some uniformit^^ Age,
on the other hand, pulls all alike, so that if one deal with groups
of allied species, and call the various factors a, b, c, d, e, etc.,
while some Avill probably pull different ways on different species,
and so cancel one another, others will pull the same way upon
all, so that the dispersal of one group of ten may be due to
{a + b + e + f) X age = 10, and of another (alhed to these)
{a + b + e + g) X age = 20. The latter will evidently be of
much greater age than the former, as it occupies twice the
area, and the factors other than age are much the same. But if
one take two groups of unallied types, e.g. one of Leguminosae
and one of Gramineae, or one of trees and one of herbs, one may
have in one case (a + c + <Z + e) x age = 10 and in the other
(b + d + f + g) X age = 20, and a comparison as regards age
alone will evidently be impossible.

A very excellent illustration of the principle here involved is
given by the tables of expectation of life published by the in-
surance companies. In no single case does "age" alone deter-
mine the period to which a man will live, yet by taking averages
of men of the same race it is possible to say with perfect accuracy
how long an average man of 45 will have to live, or a man of
46, etc.

If one be deahng with one species only (or one life only), then
the interaction of many factors, including age, will be so com-
plex that one cannot say to which the distribution (or length of
life) is actually due. It must always be remembered that the

6 INTRODUCTORY [tt. i

effects of age only show clearly when one deals with mani/
species, and those allied (and therefore more or less similar, both
in structure and reactions).

I must consider myself very fortunate in having finally dis-
covered that the effects of age were sufficiently clear to be
arithmetically expressed. It is consequently possible now to
disentangle them to some extent from the effects of the other
factors acting upon distribution, and this should tend to make
the study of these other factors and their results an easier matter.
It seems to me by no means impossible that they too may prove
amenable to statistical treatment. Many biologists have a feeling
of dislike to the introduction into biology of the more exact
methods of arithmetic; as Hooker wrote, many years ago, "all
seem to dread the making botanical geography too exact a
science," But we have become accustomed to their use in the
study of genetics, and we may hope that their employment in
geographical work may not ultimately prove too repugnant.

What has really surprised me in my Avork upon Age and Area
more than anything else, and what seems at the same time to
rouse some antagonism, is that the figures that have been given
in many papers, by myself and others, show such clear and un-
mistakable results that it is evident that mere age of species is
a much more important factor in geographical distribution than
we had been inclined to suppose. By the use of my hypothesis
that area occupied is largely dependent upon age. one can make
so many predictions about the geographical distribution of
plants, especially within comparatively small areas, and find
them correct Avithin such small limits, that it is evident that
mere age is a Axry important factor indeed, and consequently
that distribution, when one works with groups of species, and
OA'er enormous periods of time, is a much more mechanical pheno-
menon than we had been inclined to think.

Of course age in itself cannot effect anything; what is really
meant is that the resultant effect of all the active factors, like
dispersal methods, etc., is so uniform, when one considers long
periods of time and takes an average of several allied species,
that these species spread indefinitely at a fairly steady average
rate. This rate, as I have pointed out in most of my papers,
will probably not be the same for any two species, but for allied
forms Mill not usually differ very much, so that by taking groups
of ten allies, and comparing with other groups allied to the first,
the rate of expansion of area will be a fair measure of age.

CH. I] INTRODUCTORY 7

Argument of Part I. In the next eight chapters I have en-
deavoured to set forth the hypothesis as thus far developed, and
shall follow this in Chapter xi (Part II) with a general statement
ot the argument of the remainder of the book, in which the
hypothesis, which now stands upon a good basis of facts is
pushed to some of the conclusions to which it appears to me to
lead, and which are so wide-ranging that they cover much of
the ground occupied by all the biological sciences.

In Chapter ii the dispersal of plants is considered. Only by
such dispersal, accepting the views of the present day about
ongm, could they have covered the large areas that so many
now occupy. It is shown that while the possession of a good
mechanism for the purpose is of great advantage to a plant,
especially in reaching areas that are a little distance away, it is
by no means necessary for world- wide distribution. The examples
quoted about the actual dispersal of plants into new areas are
practically always cases in which there was virgin soil available
for their reception, and in actual life one very rarely sees such
distribution. Most places are occupied by societies of plants,
into which a newcomer will find it very difficult to enter, and it
may have to wait a very long time until the changes that are
always going on allow it to get a foothold. Barriers to dispersal,
even though quite small, may produce very large effects, and
as a rule dispersal appears to be extremely slow.

The questions of Introduction and spread of foreign species
and of Acclimatisation are then dealt with, and it is shown that
the popular interpretation of the rapid spread of introductions
—that they spread, and especially in islands, because they have
come from continental areas or from the north, where the
struggle for existence is keener, and has made them more effi-
cient— rests upon very insufficient evidence, and that the real
explanation, in all but a very few doubtful cases, is that their
spread is due to change of conditions. This has usually been
effected by man, who has often altered, or even destroyed, the
conditions under which many societies of plants formerly
flourished, thus giving a fair field to those newcomers that were
suited to the new circumstances. Aeelimatisation is very briefly
considered, the general conclusion indicated being that as a rule
it must be very slow and gradual, as in fact is the case with
most of nature's work.

In Chapter v it is pointed out that only in rare cases will a
seed be carried more than a few yards to survive and grow, and

8 INTRODUCTORY [pt. i

also that in view of the time available there is no need for rapid
dispersal. The various causes are then considered that may help,
or far more often hinder, dispersal, e.g. purely physical barriers
like seas or mountains, barriers partly physical, partly dependent
upon the constitution of the plant, like changes of soil or of
climatic factors, or barriers (or aids to spread) dependent wholly
upon the latter, like the fact that herbs may spread much more
rapidly than trees, that parasites can only spread with their
hosts, that a plant may or may not spread quickly according to
the particular society of plants with which it meets, and so on.
The general impression is that dispersal in nature, except in a
few (probably very few) cases, must be an exceedingly slow pro-
cess. Only in cases where man has interfered is there much
evidence of rapid spread, and the popular impression that this
is general cannot be justified by any of the facts at our disposal
as to plants in unchanged natural conditions.

Passing on to the consideration of Age and Area itself, in
Chapter vi, it is pointed out that when I began to investigate
the flora of Ceylon, I soon noticed the extraordinary differences
in area occupied that were to be found in species of the same
genus, where there were no characters of difference that could,
by any stretch of imagination, be regarded as fitting or unfitting
them for the struggle for existence. Endemic or purely local
species very rarely occupied the whole island, and must evi-
dently be adapted, if adapted at all, to local conditions within
its area. This led to a careful study of areas, and it was found,
for Ceylon, New Zealand, and elsewhere, that those species were
the most widely distributed in a country which had the widest
distribution outside, while the local or endemic species showed
the smallest areas of distribution; in both cases working always
with averages of ten allied species.

Dividing the species of a country into classes according to the
amount of area occupied, it was found that the endemics were
most numerous in the lowest class (smallest areas), the numbers
decreasing steadily upwards, while the widely distributed species
were arranged in the exact reverse direction. Such facts were
much opposed to the supposition that endemics Avere adapta-
tions to local conditions, and equally so to the other supposition
that they were relics. The facts call for a mechanical explanation,
and the most reasonable seems to be that area occupied on the
average increases with age, independently of the origin of the
species. Endemic species are usually young beginners.

CH. ll

INTRODUCTORY

The next chapter gives a few ilhistrations of the successful
manner in which Age and Area has been applied to the making
of predictions about local distribution. For example, the floras
of the outlying islands of New Zealand being in general derived
from the same sources as that of the main islands, must be com-
posed of species that were among the earliest arrivals, in their
own affinity groups, in New Zealand, and should therefore by
hypothesis, be very widespread there. This proved to be the
case, in a very striking manner, the species of the islands ranging
on the average nearly 300 miles farther in New Zealand than the
species that did not reach the islands. Further, the endemic
species that reached the islands ranged much farther in New
Zealand than the widely distributed species of New Zealand that
did not reach them. This result seems explicable only by aid of
Age and Area. Other predictions that were equally successful
are also instanced, and it will suffice to say that as Age and
Area has been applied in this manner in over ninety cases with-
out a failure, the hypothesis now stands upon a very firm basis.
A further chapter is then given to the consideration of the
way in which it may be applied to the study of the invasions of
plants that may have reached a country, New Zealand being
taken as an example. By a consideration of an imaginary case
in which a single widely distributed species enters New Zealand
and gives rise to endemics in a casual way, it is shown that the
endemics in a country will in general show numbers decreasing
from the centre where the parent entered down to the two ends.
On examining the facts it was found that all the genera of the
New Zealand flora gave such curNxs. A study of the position of
the maxima shows that they are concentrated in three chief
regions— north, south, and central— and one infers that these
must have been the centres of corresponding invasions. Careful
study of the curves given by the single invasions goes to show
that the northern was much older than the southern, and this
is confirmed by the fact that the latter is mainly composed of
the more mobile group of herbs, while the former is chiefly trees.
Lastly, Chapter ix is devoted to a detailed consideration of
the many objections that have been brought >ip against Age
and Area, and many or most of them seem to be satisfactorily
met, very many of them depending simply on misunderstanding
of the work upon which it is based.

CHAPTER II

THE DISPERSAL OF PLANTS INTO
NEW AREAS

A \TERY large number of species are to be found at more or less
frequent intervals over enormous areas of territory, often in
regions separated by large stretches of water, or sometimes of .
land. Never, since the days of the hypothesis of special creation,
has it been maintained that a species originally arose over the
whole of the area upon which it now occurs. This would be a
difficult proposition to uphold, as it is usually found that when
a species occupies a large territory, it has different varieties in
different parts. Various views, however, have at times held sway
as to the probable extent of the land surface upon which a
species began, Darwin (22, in, 109), for example, had at one time
the idea that it might arise under Natural Selection from one or
a few individuals varying in the desired direction, but Fleeming
Jenkin brought up a criticism of this position so incisive that
he was forced to abandon it, and postulate for a much more
numerous original ancestry, of course occupying a much larger
amount of ground. It is perhaps from this latter position taken
up by him that the current view has arisen, according to which
species that now occupy very small areas of country owe the
smallness of that area to the supposed fact that they are really
in process of dying out, for they could not have arisen by aid
of the Darwinian mechanism of Natural Selection upon so small
a space.

At the present time, however, when this mechanism of in-
finitesimal variation with natural selection (or survival of the
fittest) is not commonly accepted as being the principal factor
in the production of new species, it is probable that comparatively
few people would be found to demand more than a relatively
limited area for the purpose. Not many, perhaps, have any
exact idea of how much would be needed, but possibly the
majority would require either a little more than just a few
square yards, or the repeated origin of the same species upon
the same area. A good many writers, both of former times and
of the present, have adopted the view that it is not absolutely

PT.i.cH. iij THE DISPERSAL OF PLANTS ii

necessary that a species, genus, or tribe should arise upon one
spot only, or even in one region only. They consider that the
same thing may arise independently in different places, very
rarely indeed the species, more often the genus, tribe, or family,
either from the same species by the same road (as would probably
be the case with the origin of a species in this wav), or from
different species, which all made the necessary changes to place
them in the same genus or tribe (cf. 116, p. 446). This sup-
position would unquestionably get rid of some of the difficulties
of explaining many cases of discontinuous distribution, where
the same species, genus or tribe appears in widely separated
regions.

Whatever view has been held as to origin, however, it seems
to have been generally taken for granted that except in so far
as they have been prevented by actual barriers, such as seas,
ranges of mountains, sudden changes of climate from one dis-
trict to the next, and the like, species have spread over the
whole area to which they are suited, i.e. where they can grow
and reproduce in spite of any adverse conditions to which they
may be subject. In other words, it seems to have been assumed
that the distribution about the world of the species now existing
therein is largely a closed chapter, except in so far as man by
his various activities may alter it. Why this idea of finality
should have sjirung up is not quite so easy to decide, unless it
has been that people take for granted that in nature dispersal
of plants is rapid^, and it is one of the objects of the present work
to show that we are still dealing here with open questions.

It is clear, however, that the large areas now occupied by
many sj^ecies must almost always, if not always, be due to
spreading from others originally much smaller, and a careful
study of the ways in which this dispersal may be effected must
form a necessary preliminary to the study of geographical dis-
tribution in general. It is of course obvious that, as a rule, a
plant once established will not move again, but its seeds, or
detached portions of itself (or sometimes, as in the case of runners,
connected portions), maj'^ in A^arious ways be carried to a distance

^ People see a dandelion scattering seed over a large area, or notic-e tin-
rapid spread of a new weed in the garden, and are apt to reason that this
sort of thing is always going on with all species, while at the same time
they forget that most, if not almost all, seeds dropped upon ground already
fully occupied by plants, fail to grow, even if they germinate. One may see
the same clump of traveller's-joy, for example, occupy the same place
without spreading, for a whole lifetime.

12 THE DISPERSAL OF PLANTS [pt. i

from the parent. This dispersal impHes the concurrence of various
circumstances, and when all of these are external to the plant
it is spoken of as occasional or accidental, while when some are
inherent in the nature of the plant itself, it is said to take place
by aid of the regular "mechanisms." As instances of "irregular"
dispersal, we have such cases as the carriage of heavy seeds by
a hurricane, or their casual attachment to a log which is acci-
dentally floated across the sea to a new country; whilst it is
"regular" in the case of seeds so hght that they will always be
carried by wind to some little distance, or fleshy fruits which
are eaten by birds and the seeds subsequently dropped. It may
prove of more interest if an account be given of some actual
researches carried out upon this subject, rather than a mere
enumeration of the various mechanisms, etc. (54, 71).

My chief pleasure in life being travel, I have always been
interested in the movement of plants, and in 1893, with Mr I. H.
Burkill, published (137) a study of the flora foimd in the bowl-
like tops of the pollard willows that line the banks of the Cam,
especially from Cambridge to Ely. We examined about 4000 of
these trees, and counting each occurrence of one species in one
tree, whether represented by few or many individuals, as 1, and
only as 1, we obtained 3951 records. The tops of the trees being
about six feet above the ground, it is clear that without some
assistance seeds would be quite unable to reach them, though
Avhen once reached, a Millow top presents a virgin area of soil,
with no other species groAving there. There were some 200 to 240
species in the neighbourhood wliich if planted in the willows
would probably have been able to grow there, but of these we
found that only 80, or about a third, actually occurred, showing
that the presence of a barrier even so trifling as the height of a
willow Avas sufficient to exclude very many. Most of the plants
with well-marked "regular" mechanisms Avere among the 80,
though one missed Cornus (dogAvood), Salix, the avUIoav itself
(possibly it Avould not groAV in its oAvn humus), Pojmlus, the
poplar (jjossibly for the same reason, it belonging to the same
family), and a fcAV Compositae and the orchids. The commonest
plant in the tops AA-as Galium Aparine, the goose-grass, found in
644 trees, or over 16 per cent, of the total records. The fruit of
this plant has little hooks, so that it may easily chng to an
animal or a bird for time enough to be carried to a Avillow. But
it Avas also found to be largely used by birds in nest-making,
and probably the bulk of the records are due to this, for ripe

CH- "] INTO NEW AREAS 13

fruit would often be present upon the pieces carried to the trees
for this purpose.- The next most common plant was Samhucus
nigra, the elder, with 550 records; this has a fleshy fruit which
is eaten by birds, and the seeds subsequently dropped. The
third plant was Rosa canina, the dog-rose (410 records), also with
a fleshy fruit; the fourth Urtica dioica, the nettle (306 records)
with very light seeds that are easily carried by wind, but also
largely used in nest-making. These mechanisms were repeated
in the next two or three plants on the list, and then followed the
ash, h-raxi7ius excelsior, with 100 records. This has a winged
fruit, which when falling from a tree of some height during a
fairly strong wind may be carried to some distance; and as there
were many ash trees close to the river, this accounts for the
frequency of the occurrence of this species in the willow-tops
Next after this came the dandelion. Taraxacum officmale (82
records), with a fruit which in a breeze is easily carried upwards
by means of its parachute of fine hairs. By the time that we
come down the list to plants with 40 records, or 1 per cent, of
the total, 21 species have appeared there. All but one of these
have well-marked "regular" mechanisms, but the remaining 59
include a considerable number whose arrival in the tree-tops
must have been due to some "irregular" aid, for they have
neither light, winged, burred, nor fleshy fruits or seeds. Nineteen
of tliem showed only one record each, and their appearance must
be due to some such accident as having been carried in a ball of
eartii attached to a bird's foot, dri^-en by an unusually strono-
wind, or some other irregular transjDort. °

Classifying the records according to mechanism, we find:

Per cent.

Species Records of records

Fleshy fruit (animals) 19 1763 44-6
Winged or feathered

fruit or seed (wind) 33 995 25-1

Burred fruit (animaJs) 3 631 16-4

Light seed (Avind) ... 9 425 10-7

Doubtful methods ... 16 117 2-9

Thus quite an appreciable number of species are sometimes
transported, though in no great numbers. Of the 117 records,
Anthriscus sylvestris, which is used in nest-making, accounts for
63.

Three important facts appear in this result: (1) that even a
slight barrier may produce a large effect; (2) that the bulk of
the individual plants (not species) travel by aid of the "regular"

14 THE DISPERSAL OF PLANTS [pt. i

mechanisms, especially by help of birds; but also that (3) a large
number of species, even if few plants, travel by aid of "irregu-
lar" or accidental methods. If one could follow up the entire
history of distribution of plants about the globe, one would be
quite likely to find that all species sometimes travel in this way,
even though only very rarely. One would hardly expect to find
the buttercup, Ranunculus bulbosus, or Lathyrus pratensis, in the
willow-tops, yet both occur, though one docs not find such com-
mon plants as clover or daisy (44, p. 277).

Two other important results also appeared: (4) that in only
two instances did a plant occur of which there was not a repre-
sentative actually growing on the soil within 200 yards. Even
in these cases it was quite possible that at the time of reaching
the willows the distance to be traversed did not exceed that
figure, for one of the two, Lactuca muralis, was recorded for the
same tree in Babington's Flora of thirty-five years earlier. In
any case, it was clear that as a rule transport was only over
short distances; and (5), a result which appeared on comparison
with similar work done elsewhere in Europe, that the proportions
of species distributed by the various mechanisms were much the
same (10, p. 120), so that one might be able to predict to some
extent the probable composition of such a flora.

Another type of distribution was studied in working out, with
Prof. J. Stanley Gardiner, the flora of the Maldive Islands (138),
a group of coral atolls about 400 miles south-west of Ceylon, far
removed from other land. There is no reason to suppose that
any of their flora survives from the far-distant period when there
was probably a land bridge from India to Africa, so that they
probably formed a A'irgin area for the arrival of species from
elsewhere. Of their IGO species, 66 proved to be suited to carriage
by sea currents, possessing easily floated seeds or fruits, im-
pervious to salt water; 17 were bird-carried, with fleshy fruit,
4 were wind-carried, and there remained 73, probably mostly
due to unintentional carriage by man, but some doubtless
brought upon floating logs or in other ways. Again a large per-
centage of the species had thus arrived "irregularly."

Another piece of work of this kind was done upon the flora of
Ritigala (117), a solitary precipitous peak, rising to 2506 feet in
the low-lying "dry" north country of Ceylon, about 40 miles
from the main mountain mass to the south, which forms part
of the "wet" zone. The dry zone receives practically no rain
during the six months of the south-west monsoon, and has thus

CH. II] INTO NEW AREAS 15

a long period of drought, but Ritigala is high enough and steep
enough to condense the moisture of this wind, and its upper part
therefore forms an outher of the wet zone. Upon the summit is
a wet-zone flora, which must in general have reached it by over-
steppmg the whole 40 miles of separation, for the configuration
of the country, and the course of the monsoons, render it very
improbable that the intermediate country can ever have been
"wet," i.e. have received rain in the south-west monsoon also
which alone would render life possible for these species. Of the
103 wet-zone plants at the summit, 24 had fruits suited to bird
carriage, 49 had light fruits, seeds, or spores suited to wind, and
30 may be classed as doubtful, being entirely unsuited to any
of these methods, and yet equally so to growth in the inter-
mediate "dry " country. Here, therefore, was carriage by doubt-
ful methods over a good 40 miles, most probably bv the aid of
birds in some way, as the species were largely mountain species.
Of the actual Avind-carried species, 24 were ferns and lycopods
with dust-like spores, 20 were orchids with very light seeds, and
the other 5 were Compositae, Apocynaceae, and Asclepiadaceae,
with parachute-like fruit or seed.

It is noteworthy that the peak of Ritigala, a mere small area
projecting out of a sea of dry-zone plants, was probably not a
virgin area, though suitable to wet-zone forms. It was probably
covered with plants of "dry-zone" type, M-hich have only gradu-
ally been ousted by "wet-zone" arrivals, and in the whole of
the enormous period since it became suitable to the latter it has
only received 103 of them, and also bears a great number of
plants which are the same as those of the dry-zone areas below.
The Maldive Islands, which were probably a virgin area, have
received 160 species, in probably much less time, and Krakatau,
which we shall next consider, received 137 in thirty years.

Krakatau. the classical instance of the distribution of plants
to new ground, is an island in the strait between Java and
Sumatra, about 25 miles from each, and about Uh from the
nearest island with vegetation. In 1883 it was "absolutely
sterilised by the famous eruption. In 1886 Dr Treub of Buiten-
zorg vi.silud it to see to what extent it had been re-coloniscd (109);
he found many blue-green Algae, 11 ferns (spores easily carried
by wind), 9 flowering plants on the beach (carried by currents,
or drifted over by wind), and 8 inland, two of these the same as
on the beach. These eight were a Wedelia, two Comjzas, and a
Senecio, all Compositae, with dandelion-like fruits, easily carried

16 THE DISPERSAL OF PLANTS [pt. i

by wind, Phragmites and Pennisetum (grasses, ditto), Tourne-
fortia and Scaevola (fleshy fruit, bird-carried). In 1897 (34) a
further examination showed that there were 50 flowering plants,
of which about 30 were due to sea carriage, and 16 to wind. In
1905 the number had increased to 137, and the island was be-
ginning to show thick forest growth. But again the effect of a
barrier should be noted, for the flora of Java alone is over 5000
species.

Thus at first only the regular mechanisms produced any result ;
but sooner or later the irregular begin to show, for in the 137
are a few species as to whose method of reaching Krakatau it is
impossible to do more than guess. On Ritigala, where there are
30 species of doubtful method of transport, the time allowed has
been enormous, while on Krakatau it was less than thirty years.
Yet in those thirty it had, thanks to virgin soil, and somewhat
greater nearness to the sources of supply, received many more
species than Ritigala.

Another case of this kind was the re-vegetation of the Taal
volcano (38), in the middle of a lake in the Philippine Islands.
Here, again, the wind-carried plants arrived very early, and in
larger numbers of species, but the bird-carried tended to be
numerous in individuals. Both upon Krakatau and upon Taal
the vegetation began before very long to settle down into asso-
ciations of plants. While at first chiefly herbaceous plants, these
were soon followed, as happens in damp regions when sufficient
time is allowed, and no other agency, such as man, interferes, by
shrub and forest.

Incidentally, a method of dispersal which has not been men-
tioned above must receive a word of notice. This is the explosive
mechanism, as it is sometimes called, where, owing to tensions
set up in the fruit by turgidity, as in Impatiens, or by drying,
as in Claytonia, Montia, Hevea, Hura, etc., the seeds when ripe
are jerked away from the plant. The distance is commonly quite
small, but when, as in Hura or Hevea, the fruits are at the top
of a tall tree may be slightly increased.

In many respects, the last regular mechanism which has to
be mentioned, that of vegetative reproduction by portions of
the plant itself, like runners, suckers, bulbils, etc., is the most
efficient of all, as witness the profusion of daisies in most lawns,
or the difficulty of eradicating Jerusalem artichokes once estab-
lished; while anyone who has had the misfortune to have his
garden infested with goatweed, enchanter's nightshade, celan-

CH. II] INTO NEW AREAS 17

dine, or couch-grass, will need no information as to the efficiency
of this method. Tithonia diversifolia (Compositae), which has no
pappus, and is dispersed almost entirely by vegetative methods,
has spread in Ceylon as widely and almost as rapidly as Lantana,
which is bird-carried. Elodea in the waters of western Europe
was a similar case, for only the female plant is known there.
Vegetative reproduction cannot carry a plant very far at one
operation, but it is probable that to travel far, unless into virgin
soil, is really rather a handicap; and the young plant has the
enormous advantage of connection with the parent, or in any
case of a good supply of food with which to commence life.

Several other researches have been carried out in recent years
upon the actual transport of seeds and fruits. Of these by far
the most important are those of Guppy upon the stocking with
plants of islands of the Pacific and Atlantic Oceans (44, 47).
He discusses in detail the agencies that can effect distribution,
pointing out that the currents only take a comparatively minor
part in it. About 90 per cent, or more of the plants in the islands
have fruit which is not buoyant, and could only be carried by
some accidental concurrence of circumstances. After talking
about the lists of sea-carried plants given by Schimper and
Hemsley, and including in each case about 120 species, he says:
"De Candolle was quite right in minimising the effect of currents
on the distribution of plants," and again, "one can scarcely
controvert Kerner's opinion that the dispersal of plants, as a
M'hole, is not appreciably affected by this process." Leguminosae
as a family are conspicuous among sea-borne plants.

He considers that as an agency in stocking far outlying islands
birds take the first place, though there are many difficulties in
explaining the distribution. ^Vhy, for example, should Fiji have
about 200 genera not found in Hawaii or Tahiti, and yet many
of them just as well suited for bird carriage as those that actu-
ally occur there? He considers, however, that the age of bird-
dispersal is now practically over in the Pacific, and that just
like the plants the birds have tended to become local species
confined to islands or groups of islands. This phenomenon of
ciidvnmin or local species is shown most markedly in tlic case
of both plants and birds in the far outlying islands of the Pacific,
while in islands where none of the plants are peculiar, endemic
birds are few or wanting.

He goes on to point out that the development of local species
is largely correlated with degree of isolation, not only as regards

W.A. 2

18 THE DISPERSAL OF PLANTS [pt. i

distance from the mainland, but as regards frequency of arrival
of species from elsewhere. There are few local island species
among the beach plants, which are continually arriving with the
ocean currents, more among the mountain-top plants, where
probably birds most commonly alight on arrival, and most
among those of intermediate elevation.

He regards as the oldest, on the whole, those groups with
actual genera confined to the island or group of islands, then
those with genera all of whose species are endemic, followed by
those having genera with some species endemic and some widely
distributed, and as the youngest, on the whole, those having
only genera with no species endemic. He regards the develop-
ment of endemic species as due to what he calls the principle of
differentiation. They are most often allied to some common
widely ranging and polymorphous species which he regards as
the parent. To this very important conclusion he returns in
other papers (45-6), and in his later book upon the Atlantic
Ocean (47), where he comes to much the same general conclu-
sions upon distribution as in the case of the Pacific.

In fact, as wc shall see in more detail in the course of this
book, Guppy arrived at, and published a year sooner, the same
general conclusions to which I also have been driven by a life-
time spent, like his, in travel and botanical investigation, chiefly
in the tropics.

Interesting facts in regard to the distribution of the Compositae
have been worked out by Small (103). The fruits of these plants
are usually carried by aid of a parachute-like tuft of hairs, as
may be well seen in the dandelion. The general evidence that he
marshals goes to show that the fruits may frequently be dis-
persed to a distance of from four to twenty miles, and even at
times over one hundred (cf. Ritigala and Krakatau above). His
experimental observations show that so long as the relative
humidity of the air remains at a figure that keeps the pappus
open, a wind of two miles an hour (barely perceptible) is enough
to keep the fruit floating in the air for an indefinite period, but
if the moistness increases, the pappus closes, and the fruit soon
falls to the ground. Thus the dispersal of these j^lants on land,
where the air in general is drier, may at times be to great dis-
tances, but over the sea such conditions of dryness will com-
paratively rarely occur.

Important papers have also been published by Ridley on the
actual facts of spreading observed by him (91-2). For example.

<^"- "] INTO NEW AREAS 19

he studied the Dipterocarpaceae in the Botanic Gardens at
Singapore. These are tall trees with rather large fruit, upon
Avhich two or more of the persistent sepals grow out into large
wuigs. Falling as they do from a considerable height, and re-
volvmg as they fall, these fruits may be carried to some distance
before they drop, if there be a wind blowing. A Shorea, 100 feet
high, was found to scatter its fruits freely up to 40 yards dis-
tance, but not beyond 100. As it fruits at thirty years old a
httle calculation will show that in the most favourable circum-
stances conceivable, with the ground clear of other vegetation
It would take about 60,000 years to migrate 100 miles. DiiJtero-
carpus grandifolius, another of this family, ranges from the
Malay Penmsula to the Philippines, and Ridley estimates that
at least If million years would be needed to traverse this dis-
tance. He considers that liglit powder-like seed affords the most
rapid transit, plumed fruit or seed, hke the dandelion and other
Compositae, next, and winged fruit or seed, like the ash or the
Dipterocarps, the slowest (of the "regular" mechanisms for
wind-dispersal). In another paper he gives interesting points
about the dispersal of seed by mammals, calling especial
attention to the small distances usually travelled in such
cases.

What has been said so far might be read to mean that dis-
persal of plants was always a comparati^•eIy simple and rapid
process, only interfered %vith to some extent"by actual barriers;
and it is necessary now to make clear that in nature this is far
from being the case. The desirability, under the Darwinian
theory, of finding as many, and as effectual, "adaptations" as
possible, has led to those for seed-dispersal receiving much greater
credit than is their due. In all the cases (except Ritigala) that
we have so far considered, the dispersal of the plants has been
into areas of ground that could be easily occupied, on account
of the lack of competition ; and the same is the case with the
introductions described in the next chapter. But suppose that,
instead of the 4000 willow-tops, one thought of 4000 areas of a
square yard each (or of a single acre) upon a moor or in a forest,
it is at once obvious, from ordinary observation, that in 100
years they would not receive 80 new species of plants, even
though these might be growing within 200 yards. It is doubtful
if they would even receive one or two. Nor would an area equal
to that of the island of Krakatau, but upon a tropical savannah.

20 THE DISPERSAL OF PLANTS [pt. i

receive 137 new species in less than 40 years. As Lyell stated
in 1853:

Every naturalist is familiar with the fact, that although in a
particular country, such as Great Britain, there may be more
than 3000 species of plants, 10,000 insects, and a great variety
in each of the other classes ; yet there will not be more than a
hundred, perhaps not half that number, inhabiting any given
locality. There may be no want of space in the supposed tract ;
it may be a large mountain, or an extensive moor, or a great
river plain, containing room enough for individuals of every
species in our island; yet the spot will be occupied by few to
the exclusion of many^ and these few are enabled, throughout
long periods, to maintain their ground successfully against every
intruder, notwithstanding the facilities which species enjoy, by
virtue of their power of diffusion, of invading adjacent terri-
tories (69, p. 670).

This fixity of the vegetation in any given neighbourhood,
though familiar enough to everyday observation, tended to be
ignored during the period of the hunt for adaptations; but with _
the rise of the study of ecology it has once more come into
prominence, and the tendency at present is perhaps to regard
it as too permanent. A given area of ground is occupied by a
society or association of plants, made up in a fairly definite way.
This association may be open, leaving room for possible new-
comers, but tends always to become dosed, by taking in the
maximum number which can mutually adjust themselves to the
conditions there prevailing, and as altered to some extent by
each new arrival. It is a matter of extraordinary difficulty for
a newcomer to obtain a foothold in a closed association, which
mav thus form an almost complete barrier to passage. But with
the changes brought about in the soil, etc., by the vegetation
itself, and for other reasons, an association sooner or later passes
its climax, and tends to be succeeded by others. As Clements
says (16), "the most stable association is never in complete
equilibrium"; and again, "local migration is primarily respon-
sible for the population of new areas... most of the evidence
available shows that effective invasion in quantity is always
local." It is clear that to think of plants in general as travelling
rapidly about the world by aid of their dispersal mechanisms is
to take a completely incorrect view of the situation.

In fact, it is clear, and will be made clearer in the chapter upon
barriers, that in nature dispersal will be an extremely slow pro-
cess. The majority of plants have no special "mechanism" for

CH. II] INTO NEW AREAS 21

the purpose, and depend on a small transport due to wind or
animals, often only of a few inches. Ritigala, which Avas probably
covered with a "dry-zone" flora, but which has apparently
existed in its present place since the Tertiary period, has only
received 103 "wet-zone" plants in all that time, though the
conditions are favourable to them, while Krakatau, with virgin
soil, has received 137 in less than thirty years. All the work,
whether upon dispersal or upon plant-associations, that has been
quoted, goes to show the enormous influence of barriers ; but as
the floras of most countries, even of most islands, do not show
any such influences of the barriers that cut them off, the natural
inference is that in general they received the bulk of their floras
when the barriers were not there.

Looking at the dispersal mechanisms in a general way, one
gathers a broad impression that they are really of much less
importance to plants than one has been inclined to imagine. This
is confirmed by the fact that one finds many genera with little
or no mechanism for dispersal just as widely spread and cosmo-
politan as others with the most perfect arrangements. For
example, among the former we find Callitriche, Ceratophyllum,
Carex, Cocculus, Desmodium, Evphorbia, Hij^puris, Jwicus,
Lemna, Piper, Pistia, Polygonum, Salvia, Utricularia, etc. Al-
together more than half the cosmopolitan genera have no good
dispersal mechanism. (Cf. Lantana and Tithouia mentioned
above, p. 17.)

Of genera occurring in both Old and New Worlds, the family
Avith most (97) is Gramineae, whose fruits are to some degree
suited to wind dispersal, but it is followed by Leguminosae (79)
which are ill-suited to rapid spread, except to some extent by
currents. These families are followed by Compositae, Orchida-
ceae, Rosaceae, Rubiaceae, Scrophulariaceae, Liliaceae, Umbel-
liferae, Cyperaceae, Cruciferae, Caryophyllaceae, Ericaceae,
Euphorbiaceae, Ranunculaceae, Acanthaceae, Convolvulaceae,
Coniferae, Labiatae, and Malvaceae, in the order named. The
general impression is not that of the predominance of plants
with good dispersal mechanisms.

The first ten largest families in the world (judged by number
of genera) — the Compositae, Orchidaceae, Leguminosae, Rubi-
aceae, Gramineae, Asclepiadaceae, Euphorbiaceae, Umbelliferac,
Cruciferae, and Acanthaceae — are not remarkable for the pos-
session of extra good methods of dispersal, excepting the first
two. Yet not only have they the largest number of genera in

22 THE DISPERSAL OF PLANTS [pt. i

the world, but they have also tlie largest number in most large
sections of it, e.g. the Tropics, or the islands of the world, taken
together. This fact goes to show that dispersal has not altogether
depended upon the possession of a good "adaptation" for
the purpose, and also that when one takes large numbers and
long periods, it is to a marked degree mechanical. Attention was
first called to this striking fact by Hooker in 1888 (56, p. Ixiv),
in these words "the conditions which have resulted in Mono-
cotyledons retaining their numerical position of 1 to 4 or there-
al^outs of Dicotyledons, in the globe, and in all large areas
thereof, are, in the present state of science, inscrutable."

If the methods of dispersal be compared throughout a family,
it Avill be found that they are often attached only to a genus or
group of genera, and thus are probably comparatively modern.
Even in Compositae, Avhich as a whole have the same mechanism,
there are a good many widely dispersed forms with no pappus.
"Of the Compositae common to Lord Auckland's group, Fuegia,
and Kerguelen's Land, none have any pappus at all ! Of the
many species xvith pappus, none are common to tw^o of these
islands" (55a, p. xxi, note). " Phyllanthiis shows by its distribu-
tion in the Pacific that dry-fruited Euphorbiaceae are as widely
distributed and as much at home as the fleshy-fruited ones"
(44, p. 325). And of. 7, p. 573.

Summary
It being generally agreed that plants dispersed over large
areas began upon smaller, a study of the methods of dispersal
must form an introduction to that of distribution in general,
and a number of cases of such inA'cstigation, ffom the flora found
in the pollard-willow trees near Cambridge to the new flora of
the island of Krakatau, are given. The general results that seem
to come out of all such work are (1) that barriers to spreading
produce very important results; (2) that most individual plants
travel (to anything more than the very smallest distance) by
aid of the "regular" mechanisms for dispersal by wind, water,
or animals (or vegetative reproduction); but (3) that a great
many species are sometimes, even if very rarely, carried by
various "irregular" methods — mud on birds' feet, hurricanes,
floating logs, etc.; (4) that the distance covered is usually very
small; but (5) that dealing with large numbers and long periods,
the general result tends to be much the same in all cases under
somewhat similar conditions. On the other hand, the fixity of

CH. II] INTO NEW AREAS 23

the vegetative covering of any given area shows that the pos-
session of a good dispersal mechanism is only rarely of much
value. A general comparison of the flora of the world shows that
more than half of the most cosmopolitan genera have little or
no mechanism for dispersal, nor is such Avell marked, on the
whole, in the largest and most widely distributed families. It
also goes to show that in most cases where there are now wide
separations by seas, etc., the floras are too large to have been
able to arrive across such formidable barriers.

CHAPTER III

THE INTRODUCTION AND SPREAD OF
FOREIGN SPECIES

One of the most commonly misunderstood or misinterpreted
phenomena in connection with the distribution of plants is that
exhibited by many species that have been introduced, whether
intentionally or not, into countries to which they were not really
native. Often they have spread rapidly, and are now among the
most common plants. The casual traveller in Ceylon, for example,
will notice everywhere by the roadside the sensitive plant
(Mimosa), the Mexican sunflower {Tithonia), Lantana, Mikania.
various Cassias, guavas, Turnera, Vinca rosea, etc., not one of
which is really native. Higher in the hills he will see abundance
of clover, dandelion, gorse, shepherd's purse, spurrey, etc., also
introduced in recent times.

When Europeans first settled in tropical and other countries
to which they were newcomers, the places in which they located
themselves were not determined by mere chance, but were
places to and from which transjiort was most easily and cheaply
obtainable (114, p. 36). They had not come to these countries for
the benefit of the inhabitants, but to begin trade with Europe
in those products that they only could supply. Accordingly the
white men settled at the mouths of the great rivers like the
Ganges, Yang-tze-kiang, Amazon, de la Plata, etc., where ports
existed or could be easily made, and goods could be easily
brought down from inland. E\'en more frequently they settled
upon the islands, beginning with the smaller ones. Here there
was less risk of invasion by the natives in great force, and trans-
port from the interior was usually easy, by reason of the com-
paratively small size of the country, though of course river-
mouths were utilised, for purposes of port accommodation, and
of transport from the interior, \vhenever possible.

In these places introduced plants were soon foimd spreading
about, especially when the country, prior to occupation — as was
very often the case, especially on the islands — Avas in its natural
state of forest. No notice was taken of this spread until the rise
of the theory of Natural Selection, when it was found that these
introductions apparently gave good evidence in its support. This

PT. I, CH. Ill] SPREAD OF FOREIGN SPECIES 25

evidence was accepted without being always subjected to proper
sifting, and it was for a long time believed that these introduc-
tions spread with this rapidity and success because they came
from large continental areas (chiefly Europe and tropical
America) where they had, so to speak, become highly efficient
and "up-to-date" by competing in the struggle for existence
amongst a large crowd of other species, and were in consequence
exterminating the native productions because these had not
had such advantages. In a comparatively short time the fact
that introductions also occurred on continental areas was almost
lost sight of, and the argument was applied almost entirely to
islands. Darwin, for example, states (23, p. 340) that " in many
islands the native productions are nearly equalled, or even out-
numbered, by those which have become naturalised; and this is
the first stage towards their extinction." Wallace (111, p. 527)
makes very similar statements.

Now the fact of rapid spread in many cases is undeniable, and
also that it has been largely, if not mainly, recorded from islands.
But no proper analysis of the evidence has been made. One
soon finds that introductions are just as common on continental
areas, especially where these (as was nearly always the case upon
islands) were untouched forest at the time of settlement. Thus
141 species have been recorded as spontaneous in the Transvaal,
S68 in South AustraHa, 364 in Victoria; 800 introductions, of
which 107 have become naturalised, occur near Montpellier, and
548 in the Tweed valley (14, 11, 35, 107, 143). One also finds
that the cases of rapid spread without alteration of the conditions
are very few indeed ; in most cases man has removed the forest,
made great clearances, introduced grazing animals, or in other
ways completely altered the circumstances, thus enabling those
introductions to survive and prosper which were suited to the
new conditions. And one further finds that when introductions
have spread, it has been just as much, if at all, at the expense
of species in the native flora that are of wide distribution as of
species of the most strictly local kind. The great bulk of cases
of spread of introductions are due, not to the fact of their having
come from Europe or America, but to the fact of their suiting
the new conditions created by clearance of the forest, or cultiva-
tion of the ground, to which there were few or no native species
suitable, and to the fact that man has thus broken up the old
associations of plants that covered the ground, and made it
possible for new plants easily to gain a foothold. In Ceylon, for

26 THE INTRODUCTION AND [pt. i

example, which has been much quoted in this connection, all
but 11 of the 387 naturalised species (115) are either weeds about
houses, due to cultivation, or weeds of open ground, which was
all but unknown in the old days of forest. Seven of these 11 are
only a clump or two of planted trees, and there are really only
tAvo cases of natural or nearly natural spread, and then only to
a distance of a few hundred yards, downstream, in a very steep
valley.

The only instances of rapid spread in Ceylon, on land already
occupied by a growth of plants suited to the conditions, have
been in the case of a few such weeds as Tithonia and Mikania^
M'hich have spread rapidly over the open ground already occu-
pied by weeds previously introduced. One is inclined to think
that this is due to the fact that such areas have not as yet
elaborated the best plant societies suited to their conditions,
and that room is still left for newcomers. On the other hand,
there is no evidence for rapid spread in forest, where the adjust-
ment of species to environment has probably been carried to
great lengths.

In a few cases, new species have spread rapidly over ground
already occupied by herbaceous plants, like Elodea in the waters
of western Europe, or Spartina in the low coast lands on the
south of England. Here, again, one may suppose that there has
still been room for newcomers, especially in the case of the
Spartina, which is largely found on land that was submerged
not so very long ago.

The enormous majority of cases of rapid spread of introduced
weeds are due to cutting of forest, or other serious alterations
of conditions ; in North America and Argentina often to cultiva-
tion of the soil (even if only once), leaving conditions different
from what they were. In St Helena, which has been much used
as an argument for nat\n-al selection, man introduced goats,
which are most destructive to vegetation, with the result that
there is left only a flora practically "goat-proof." Even the
largest trees are not safe, for the goats may destroy the smaller
trees, and expose them to the action of the sun and wind. [Cf.
also the effects of the exclusion of rabbits from a heath (36,
1917, p. 1).]

Cockayne has devoted much attention to New Zealand, an
island in which over 550 introductions have become more or
less naturalised, and which has often been quoted as evidence
for the great superiority of introductions from the crowded flora

CH. Ill] SPREAD OF FOREIGN SPECIES 27

of Europe. In actual fact, however, man, by felling the forest
ntroduemg cattle, and in other .vays, has completei? altered the

InTv Te r^H ^•"•''''" ^''' P- ''^ ''^* introduction:
only give the characteristic stamp to the vegetation "where

tusso"/' -Itivation, constant burning of fofest, scrub, and
mX' h"l . r ^""^5 °f-^-^ltitude of domestic animals have
tW t;^^''''^^ r: I^P^^^^ conditions, which approximate to
those of Europe." And farther on in the same piper he says,
T3lan'ts".'f.f "h"' vegetation is still virgin, and the introduced
plants altogether absent, where grazing animals have no access,
and where fires have never been."

Bolle regards the Canarian endemic flora as "everlasting" and
indestructible," and writing of the same flora Christ views the

intrideTs""''' ^' """ "' ^''''"''' °^ '^' "^'^''^ P^^"*" ^"^ ^^^^^^^
There are very few cases of rapid spread of introductions that
cannot be accounted for by changed conditions, and in many of
these It IS probable, as in the cases of Elodsa or Eichhornia
water-hyacinth) in the water, or cacti or Cynara (cardoon) on
the land, that they have proved suitable to joining a plant
society which as yet was incomplete (open) and allowed room
for newcomers The few cases remaining, that are quite in-
capable of explanation as yet, are so very limited in number
that to base any argument upon them would be in a very high
degree dangerous. ^

The spread of introductions is often so rapid and striking that
one IS tempted to lay too much stress upon it, and to think that
the ongmal rate of spread of most species was something of the
same kind But there is no evidence to support this view, and
the natural rate of spread is often so slow that one may even
think that nothing is happening at all, and that a species has
reached its limit of distribution, whereas if things could be left
quite untouched for several centuries, one might find an ap-
preciable change at the end of that time.

Summary
An erideavour is made to show that in the great majority of
cases the rapid spread of plants introduced into new countries
IS due to the changes of conditions that liave been made bv
man, and not to the fact that these plants have usually come
from more complex and "eflicient" floras. Introductions are
just as common upon continental areas as upon islands (to

28 SPREAD OF FOREIGN SPECIES [pt. i, ch. hi

which they are sometimes supposed to be largely confined); and
their spread is practically always due to some change of con-
ditions that has to a greater or less extent interfered with the
success of, or has even destroyed, the society of plants formerly
growing upon the ground which they now occupy.

CHAPTER IV

ACCLIMATISATION

A.CCLIMATISATION may be described as the accustoming of
plants to new conditions and climates till they are not only
capable of growing there, but also of reproducing themselves
freely. Thus, though the cherry and apple will grow readily
enough in the hills of Ceylon, they are not really acclimatised,
for they do not produce fertile seed, and if left to themselves
would inevitably die out. Lantana, on the other hand, is com-
pletely acclimatised, and seeds freely.

As practised by man, acclimatisation is chiefly modern, but
in nature it has been going on for ages. Hers is much more
gradual, but there is no nursing of a delicate plant till it can
survive and reproduce; if in any way unsuitable to the altered
conditions, it will die out. Man used to try to make enormous
changes, as from Europe to the Tropics, but has slowly learnt
that this is usually impracticable, and has even begun gradual
acclimatisation, as for example in the way in which he has
treated Liberian coffee in Java, taking the seed of successive
generations a few score yards higher up each time, till he has
persuaded the tree to do well at a much higher elevation than
that to which it is naturally suited.

In the Ceylon Botanic Gardens we were very anxious to
acclimatise the beautiful Cyperus Papyrus; so long as we tried
seed from Europe we failed, but seed from Saharanpur in India
succeeded at once. Sometimes the difficulty is M'ith change of
climate in regard to periodicity, as when one tries to acclimatise
plants of the southern hemisphere in Europe. Sometimes the
plant requires a mycorhiza (or fimgus in association with the
roots) for its successful growth, and it may not be possible to
persuade this to grow, as with heather in Ceylon, which has
never succeeded there. But it would lead too far to discuss all
the many and complex i^hcnonienu uf neelimatisiilloii !ifi prac-
tised by man, and we must return to that carried on by nature,
which almost never attempts to make great changes at once,
except when, for example, the Gulf Stream carries to Europe
seeds which refuse to grow there except in hothouses.

As a rule, nature's acclimatisation is simply to the slightly

30 ACCLIMATISATION [pt.

different conditions that may be experienced in a transit of a
few score of yards or less, but small journeys like this, added
up over many centuries, ultimately result in enormous differences
of conditions, as when one finds Hydrocotijle asiatica growing in
the low country of Ceylon, with a steady mean temperature of
80°, and in Stewart Island, New Zealand, with Avinter snow and
frost.

Acclimatisation may also take place in nature without the
plant changing its position, by the secular changes of climate
which are usually going on. New Zealand had probably at one
time a more or less tropical chmate, and now has a temperate
one, yet the tropical species are still to be found there, and quite
probably may have originally arrived when the climate was
warmer, and then become gradually acclimatised, themselves
and their descendants, to climates steadily becoming colder.
The rise of a mountain chain may gradually acclimatise plants
to a colder climate, by carrying them upwards.

This gradual acclimatisation that is carried on by nature has
often been so successful, as illustrated by Hydrocotijle asiatica
above, and by scores of other "tropical" species which are found
far south in cold but still damp climates (to the northwards the
change to dry is more sudden) that it makes it very difficult to
say when a species has really reached its climatic limit, beyond
which no amount of acclimatisation would be of any use. People
are apt to say that laboratory experiments show that such or
such a temperature is the lowest that a plant will stand, for-
getting nature's very gradual acclimatisation. Hydrocotyle from
Stewart Island would almost certainly gi^e reactions in the
laboratory different from those of the same plant from the
plains of Ceylon.

It is possible, again, that in nature's acclimatisation by
gradual change of climate, plants may become slower in the
performance of their functions, or in groAvth, so that the genera-
tions may be farther apart.

Yet another factor that has probably an important influence
is the increasing number of species upon any given piece of
country. As the species increase in number, they probably begin
to form more or less complete or "closed" associations of plants,
into which intrusion of a newcomer becomes increasingly diffi-
cult, so that probably both rate of travel and acclimatisation
are rendered slower and more troublesome.

^^' IV] ACCLBIATISATION 31

Summary
Acclimatisation in the hands of man, who is impatient of
results, has been largely a matter of trial and error, with nume-
rous failures, but there is reason to suppose that this is not so
much the case in the hands of nature, working as she does over
vast periods of time, with very small steps. Species have thus
been acclimatised to conditions wonderfully different from those
in which they began.

CHAPTER V

CAUSES WHICH FAVOUR OR HINDER
THE DISPERSAL OF SPECIES

It being generally considered that a species commences upon a
comparatively small area (or areas), it is clear that it has to do
much travelling to cover the large territory which is now occu-
pied by so many forms. In general it will be dispersed by aid
of one of the methods already described, whether regular or
not, and will be aided, or far more often hindered, in its
journeyings by various factors which we have now to consider.

Whatever one's view^s may be as to the efficacy of transport
to a distance, it is unquestionable that as a rule new plants of
a given species grow up fairly near to pre-existing specimens of
the same kind. For one thing, though it is often overlooked, it
is much more difficult for a plant carried to a distance to estab-
lish itself, under the different conditions of climate, soil, and
especially of plant societies, etc., that it will then meet with,
than if it were simply transported a few yards, just as it would
be more difficult for an emigi-ant from England to establish
liimself in a foreign country, rather than in a colony or the
United States.

If near to a solitary tree of a given kind there exist, at dis-
tances of ten, a hundred, and a thousand yards, spaces where
its seeds if sown w^ould stand a reasonable chance of growing
and flourishing, then it is clear that to put a seed on every
square yard^ up to a distance of ten (supposing the impossible
case of uniform distribution), the tree would have to disperse
314 seeds; up to 100 yards 31,400; and up to 1000 yards it
would need no less than 3,140,000 seeds, a number probably
far beyond the capacity of most trees. In actual fact, the seeds
are notoriously carried in such vastly greater numbers to the
smaller distances, that this figure would probably have to be
multiplied by 100, or even 1000 or more, to allow of the plant
placing a seed on every square yard up to a radius of 1000 yards.
But even this is not enough, for a seed placed in one part of a
square yard, while the suitable spot for its growth is in another,
will have no better chance of success than if a dozen yards away.
1 Area of circle = Trr", e.g. 314 x 10 x 10.

PT. I, CH. vj THE DISPERSAL OF SPECIES 33

One must again multiply by say 150, to allow a seed for every
three mches square, or probably by even more than this. Unless
therefore, a seed just happens to fall on the exact spot where it
can grow the chance that the plant .will ever travel more than
a few yards from its parent is but a small one^; and the majority
ot plants cannot m any case travel more than a few yards
except by irregular aid, for want of a suitable mechanism

Of course, m cases where a wall of uniform vegetation, like
the edge of a pme forest, is advancing, the number of seed re-
quired per tree to reach to a considerable distance will be much
reduced, or even where the plant, as is more often the case is
thinly scattered along a given front, but in any case, to reach
a favourable spot at some distance away, a vast number of seed
will be required. In the temperate zone, where seed may survive
for a long time, the chance of such success is greater, but in the
tropics, where they rarely remain ^•iable for long, is but slight.
Not only so, but the vegetation of the wetter tropics is usually
forest, and so thick that a seed dropped near the top of the tree
canopy will be unlikely to reach the soil if not very heavy,
unless by mere chance.

There is not the least need, when one has regard to the vast
periods of time that are available for the purpose, for rapid
dispersal. Few people, perhaps, have fullv grasped the fact that
while some species occupy very large areas, the bulk of them do
not, and the average area is but comparativelv small. Upon
50 milhon square miles of land there are about "100,000 species
of flowering plants, so that if each occupied its own area, and
alone, the average would be about 300 square miles. But in
fact, at a rough and fairly hberal estimate, there are say 3000-
4000 in any given country, which would make the average about
a million square miles, probably an overestimate. But taking it
at a million for convenience, this area could be covered in a
million years (a mere detail in geological time) by an annual
plant Avhich merely moved forward a yard a year, and which
started on an open plain of the necessary size, with a uniform
chmate.

While the radius of the area occupied increased 1, 2, 3, 4, 5,
etc., the area {nr^) would increase 3, 12, 27, 48, 75, 108, the
differences being 9, 15, 21, 27, 33, or an annual increase of G.

1 The rapid spread of weeds does not affect this argument, for tJiey are
spreading upon cultivated ground, and owe their rapid dispersal to changed
or unnatural conditions, as do the introductions considered in Chapter in.

W. A. o

34 CAUSES WHICH FAVOUR OR [pt. i

The area would thus grow with increasing speed\ and though
for a long time it would be A^ery small, at the end of a million
years the radius Avould be 1 million yards, or roughly 600 miles
(London to the Shetlands, Dresden, and the Pyrenees), and the
area over a million square miles.

Thus, in a period of time which is almost insignificant from
a geological point of view, a species without competition or
interference might cover an area which is probably larger than
the average area of a species to-day. At the same rate of travel,
in 12 million years it would cover an area of 50 million square
miles, equal to the whole available land-surface of the globe,
and in 24 million years might cover the entire surface of the
earth, supposed land with uniform conditions. All these periods
are probably small compared even to the Tertiary period of the
earth's history, for Lord Rayleigh has estimated the time since
the Eocene alone at 30 millions.

These figures are of course the merest rough approximations,
and are given simply to show how little actual forward move-
ment is required to do, in a comparatively short space of time,
what has actually been done by even the most widely distributed
species. No special mechanism for dispersal would be imperative
in such a case. It is clearly obvious that in nature what actually
happens must be delay of spread rather than acceleration.

Another important point that one must not allow to be for-
gotten, and which may perhaps be dealt with best in this place,
is the simple arithmetical ratio in which an early species will
gain upon one that appears at a later period, both in the area
occupied, and in the chance of giving rise to new species. Let
us suppose that both of these are purely mechanical processes,
and that the species spread uniformh^ in every direction, as
before, without let or hindrance. Then if two species A and B
start at different periods, spreading at the same rate, B will
never catch up to A, but Avill always fall behind. The areas
occupied will be (cf. above):

^ 3 12 27 48 75 108 147 192
B — — — 3 12 27 48 75

Difference... 45 63 81 99 117

1 The dispersal would of course tend to become less and less dense, but
as for Age and Area purposes area is estimated by drawing a circle round
the outermost localities, this matters little.

CH. vj HINDER THE DISPERSAL OF SPECIES 35

The differences in area occupied will continually increase (by 18
every time) though in radius of area B will always be only 3
behind ^. And in the same way, if each give rise to new species
in proportion to the area occupied {i.e. number of individuals)
A will continually gain upon B. In actual practice, of course
the result will not be so mechanical, but 07i the average the earlier
formed species will gain upon the later, both in area and in
number of progeny of new species, unless the later formed ones
are superior to the parents. This gain is incidentally shown to
be the case by looking over the geological record. The o-enera
that are found in the earliest horizons are in general large genera
of the present day. Twenty of them from one horizon, though
one or two are now extinct, include over 2000 species now living,
so that their average size is at present over eight times the
average of twelve species per genus.

It is clear that in nature the usual case will be transport to a
small distance only. But when this has been accompHshed, the
seed has still to become a plant capable of reproducing itself,
and to do this it has to overcome manv difficulties, the chief
perhaps being the fact that, as a rule, the ground is already all
but completely occupied by plants more or less fully grown, so
that even a vacant space left by the death of one of them will
be full of roots, and overshadowed by the neighbouring plants.
Not only so, but the plants that grow upon any given piece of
ground in its natural state generally form Avhat is called an
association or society, into which a stranger, i.e. a plant of a
species not usually occurring in that association, will find entrv
very difficult. ^Ve shall return to this subject below.

When a species is just commencing its life as such, and con-
sists possibly of a very fcM^ individuals, there is no doubt that
its chance of spreading, by seizing upon spots more or less vacant,
will be much less than when it becomes more common, as indi-
cated by the very few plants that make up many endemic
species (below, p. 55). A species, unless it start upon an un-
occupied piece of ground, will probably take a very long time
to spread from the condition of half-a-dozen plants on a few
square yards to reasonable frequency on a square mile. Once
established with commonness more or less equal to that of its
neighbours, it will probably spread with a rapidity much the
same as that of other species of the same genus living in the
same country and in the same type of vegetation, inasmuch as
all will probably have much the same type of mechanism for

a— 2

36 CAUSES WHICH FAVOUR OR [pt. i

dispersal, and will react to their surroundings in much the same
way. But as yet we have no means of comparing the rate of
spread of species that are separated in systematic relationship,
and which may differ in many ways. Some may have powder-
like seed, easily carried bj^ wind, others fleshy fruits dispersed
by birds; some may be herbs, with a generation every year or
two, and a corresponding chance of frequent dispersal, others
may be trees with as much as twenty to thirty years between
generations; and so on.

If the dispersal of plants depended simply upon their
"mechanism" to that end, it is evident that (working with
groups of species, and long periods) it would be almost a purely
mechanical process, the area occupied enlarging steadily with
the increasing age of the species; and of course each species
would probably progress at a different rate, those with good
mechanisms, or in good environment, or flowering while still
young, travelling more rapidly. After a certain period of time
the areas occupied by a set of different plants, say a Dipterocarp
tree (p. 19), a Leguminous tree, a Cruciferous herb, and a Com-
posite herb, all starting simultaneously on area represented by
1, might at a guess be, say, 2, 5, 10, and 100. But in actual life
many other causes come in to facilitate or delay the spread of
species, and it seems probable that delay, rather than accelera-
tion, is the usual result. This is chiefly the case, for instance, with
the actual physical features of the world, which we shall
consider first.

Ojjen seas, for example, and even comparatively narrow arms
of the sea, like the English Channel, may offer practically in-
superable barriers to migration, only to be occasionally passed
by a few species, unless M^ith the assistance of man. An im-
portant point to remember is that such seas, or arms of the sea,
may be comparatively recent, or of very ancient standing in.
geological history, so that their total effect upon distribution
may be relatively small, or of very great importance indeed.
Once formed, however shallow or deep, a sea will offer much
the same obstacle, and the degree to which it obstructs passage
of species will to some extent depend upon the direction of any
currents that may traverse it. Further, even when it has become
wide enough, in the process of formation,' to stop some species
completely, others, by virtue of good dispersal mechanisms, may
be able to cross.

Mountains, again, are of great importance. Considered merely

CH. V] HINDER THE DISPERSAL OF SPECIES 37

as elevations of the ground, they would probably make com-
paratively little difference to the existence or to the migrations
of plants, unless very high or very steep, but their presence
usually mvolves change of climate from one side to the other
and trom bottom to top, so that they mav produce cxreat effects
upon the composition of the vegetation, whether as seen in
simply ascendmg them, or in crossing to the other side The
climate usually becomes cooler and damper in ascendino- until
the cloud belt is passed at high elevations; and if the ra'nge be
transverse to a damp air-current, as so often happens owing to
the fact that ranges are frequently parallel to the sea, much rain
will be precipitated on the nearer side, and the farther side will
have a much drier climate. This effect can be well seen in the
mountains of Scandinavia, of Portugal, of New Zealand, in the
VVestern Ghats of India, the northern Rocky Mountains the
Cascades, etc. If the change is very great, the flora mav be
almost totally different on the two sides of a range.

Mountains may also serve as agencies facilitating migration
of species, inasmuch as they may enable the passage into or
through a country, otherwise unsuitable in whole or in part, of
the plants of cooler or moister climates, or of herbs of open
gTound. They are also fa^^ourable to rapid migration because
the frequently occurring landslips may open appreciable areas
ot new soil not covered by vegetation, upon which plants may
at once take hold, without having to wait to secure a spot
temporarily free, or struggling to effect an entrance into a closed
association of plants. Such plants will probablv be mostly herbs
or small shrubs, inasmuch as landslips will be more common at
the higher elevations, which are above the tree line in many
cases. Owing to the fact that changes of climate have often
taken place in a north and south direction, mountain chains
running east and west have been of especial importance.

As a general rule, a river hardly seems to be of sufficient
width to offer a very formidable obstacle to migration, though
It will doubtless delay it considerably. The only river that really
seems large enough to be, possibly, an actual boundary to
migrntinn in some eoscs is the Amazon in the lower half of its
course, from Manaos, where it is joined by the Rio Negro, to
the sea, and where it may be several miles wide. Owing to the
density and enormous size of the forests, howe\xr, we do not
yet know enough of the local distribution of the plants of that
region to be able to say whether or not any species really meet

38 CAUSES WHICH FAVOUR OR [pt. i

the river with a long frontage to it, and are not found at all on
the other side.

Soil may be considered as a geographical factor in migration^
inasmuch as it depends upon the geology of the country, or may
be considered under the next heading, of ecological factors.
Nothing has been a subject of greater controversy than the
effects of its composition upon the vegetation which it carries.
There is no doubt that one may observe quite different floras
upon, say, a chalk soil and a sihceous soil in England, and quite
another again upon a soil impregnated with salt. Exactly to
determine, however, what part of this effect is due to the
chemical composition of the soil, and what part to its physical
constitution, is a very difficult problem. My own experience
with tropical agriculture, extending over nearly tAventy years,
inclines me to lay more stress upon the physical constitution,
for crops will succeed almost equally well upon soils of very
different chemical composition, if only they be, for example, of
such physical consistency as to retain water Avell. Chalk soils in
the natural condition arc dry, and little retentive of water, sandy
soils even more so, while clays may retain water very well indeed.

It is comparatively rare for any plant to be confined in its
growth to one kind of soil only. Feshica ovina is so abundant
and successful upon the chalk downs that one is tempted to
think it a chalk plant till one finds it almost as common upon
a bilberry moor in Derbyshire, or a grass moor in Scotland,
with peaty soil. Both chalk and peat demand in the plants that
grow upon them some capacity of resistance to insufiflcicncy of
Avater, and it may be the physical rather than the chemical
constitution that matters most.

There is no doubt that if in the same climatic and other general
conditions there exist two belts of different soils, these will be
covered with floras that will be differently constituted in detail,
but it is comparatively rarely that a species will not occur on
both, though it may be common on the one and very rare on
the other. The only chemical constituents present in the soil
that really seem to have a determining effect in allowing some
species and excluding others are calcium carbonate (chalk or
limestone) and sodium chloride (salt). A good case is mentioned
by Drude (33) of a line of chalk-loving shrubs found running
through a forest on siliceous soil in France; on investigation it
was found that they occupied the track of an old Roman road,
for which chalk had been used.

CH. v] HINDER THE DISPERSAL OF SPECIES 39

As a general rule, a change of soil does not cover a breadth of
country sufficiently wide to form an absolute barrier to the
passage of some species, or a special assistance to that of others.
If it is broad one way, it may be narrow in the direction per-
pendicular to that. There can be no doubt, however, and this
is all that matters to our present discussion, that it may readily
hinder or delay the passage of some species, and assist that of
others; and that it may distort as well as delay some species in
their distribution, by compelling them to go round.

We come now to those hindrances interposed b}^ change of
conditions (to which plants react in different manners) either
from one place to another, or from one time to another, which
in a general Avay may be classed as ecological. The change may
be very sudden, as from forest to dry grassland (seen very
strikingly at the edge of the patanas of Ceylon; cf. 81), or from
a wet to a dry climate, as on the two sides of man)- mountain
chains; and in this case one comparativel}'^ seldom finds the
same individual species growing on both sides of the barrier
thus formed. But if the change be more gi-adual, as from warm
to cold in ascending a mountain, one often finds this to occur.
To what extent the barrier is effective, therefore, will depend
largely upon its sharpness of definition, as well as its width and
depth, and upon whether a genus on reaching it is able to form
new species capable of living upon the other side. This is a
phenomenon Avhich is very often seen, and it is in fact b\'- no
means certain that an ecological barrier will interrupt com-
pletely the progress of a genus, though it may stop a species.
When a genus is found confined to wet or dry, high or low, it is
most probably, as we shall see, because it is still comparatively
young in that country, and has not j'ct had time to spread
widely; quite possibly it has not yet even reached the actual
boundary. Widely distribvited genera, if they have many species
in the country, more usually have species on both sides of the
boundary. In the first hundred genera of the Ceylon flora, for
example, the genera which have species in both wet and dry
zones (which have a very different climate, cf. p. 14) are 32
with 141 species, or an average of 4-4 per genus, while those
confined to one zone are 68 with 135 species, or an average of
2 only. [The average for the whole flora is 2-7.] Of genera in
the entire flora that have over 10 species, seven only, with 21,
20, 13, 12, 12, 12, and 11 (average 14), are confined to one zone;
eight, the largest with 27 species, have one or more species

40 CAUSES WHICH FAVOUR OR [pt. i

occurring in both zones (average 16), and 21 genera with 484
species (largest 43, 42, 40, 38, average 23) have separate species
in each zone.

There are many ecological changes which may be summed up
as climatic, and which, if they occur over a sufficient depth and
width of country, may offer very formidable checks or barriers
to dispersal. Such, for instance, are change of rainfall, of dis-
tribution of rainfall, of temperature, of dampness of air, of light,
of wind, etc. The combined effects of these form what may be
termed the cHmate of a place. In the existing conditions of the
world the climate is determined in broad outline chiefly by lati-
tude, position with regard to the sea, to prevailing winds, and
to mountain chains which are at no very great distance. The
lower the latitude, the Avarmer the climate; the nearer the sea,
and the more wind blows from it, the damper; the nearer the
lee side of a range crossing the ])revailing wind, the drier.

Further, during at any rate the later periods of the world's
history, great ranges of mountains have sprung up in different
directions, especially from east to west in the Old AVorld, from
north to south in the New. These ranges are so lofty that apart
from the changes of climate due to them, they have acted as
very formidable barriers. And when to this is added the enor-
mous difference of climate on the two sides, it is clear that they
must have completely altered the distribution of species, and in
general rendered it more difficult for the greater number, though
on the other hand, species, chiefly herbaceous, which can live
at high levels in the mountains, have been enabled to travel
through and into regions otherwise impassable (cf. p. 37). It
is in this way, probably, that many herbaceous and shrubby
types of vegetation, including such genera as Caltha, Lignsticum,
and Veronica, characteristic of the north temperate regions, but
now also found in New Zealand, South America, etc., have been
enabled to reach those countries; and that the comparatively
young Compositae have spread so widely over the world.

The effects of the mountain ranges on the two chief continents
may be seen by comparing the climates of North America and
of Europe, both in the zone of pre\ailing westerly winds from
the ocean. The west coast of the former is very wet, in latitudes
equal to those of northern Europe, and was originally covered
with forest; but as one comes to the east of the Cascades and
Rocky Mountains, which lie across the path of the westerly
winds, one reaches the land of prairie, which is especially dry

CH. V] HINDER THE DISPERSAL OF SPECIES 41

in the near neighbourhood of the mountains. In north-central
Europe, on the other hand, the cHmate slowly becomes drier
Avith fair regularity in passing from England to the Urals and
then becomes suddenly much drier. In the Scandinavian
peninsula, the mountains he more across the wind, and Sweden
is much drier than Norway.

Farther south, there is a great belt of more or less dry and
desert country, almost round the world in the northern, much
less marked in the same latitudes of the southern, hemisphere
and between these two drier regions, which oppose all but im-
passable barriers, lies the wet zone of the equatorial tropics
where the climate is usually damp, and often very rainy through
a great part of the year, though there are alternations of drier
and wetter periods.

If a country be flat, or nearly so, as, for instance. North Europe
from England to the Urals, the rai7ifall gradually falls off as
one goes inland from the sea, but only in averages over a number
of years. If, for example, at a series of stations, Avorking inland
from the sea, the rainfall average 50, 45, 40, 35 inches, it^'is quite
possible, if not even probable, that in some years the fall at the
station farthest inland may be 50, or in others that the fall at
the station nearest the sea may be only 35. Unless, therefore,
plants are suited to a great range in the amount of rainfall, they
cannot hope to succeed in most stations, and it also becomes
doubtful when and where the rainfall reaches an absolute
maximum or minimum which causes it to be an ecological
barrier. It is also highlj^ unlikely that this point will be the
same for any two species. That there is such a barrier seems not
improbable when we consider the difference in flora between
the steppes of Russia and the British Islands, but where it
exists for any single species we are unable to state.

If, however, as very often happens, a mountain chain stand
athwart the prevailing or most frequent winds, there may be a
sudden change in the rainfall. The damp air from the sea, striking
the mountains, is forced upwards and cooled, parting with much
of its moisture; then as it descends upon the other side, it be-
comes warmed, and thereby much drier. In Ceylon, for example,
the south-west monsoon blo^vs for about six months over a vast
expanse of ocean, and reaches the island a saturated wind.
Meeting the mountains, it deposits an enormous rainfall (over
100 inches at the foot of Adam's Peak), and upon the eastern
side (they reach 8000 feet) becomes a dry scorching wind, deposit-

42 CAUSES WHICH FAVOUR OR [pt. i

ing less than 10 inches of rain at Batticaloa on the east coast.
The change at the summit-level is so sudden that one may some-
times find a wet climate at one end, and a dry and sunny one at
the other end, of the short summit tunnel on the railway. The
position is largely reversed during the other monsoon, so that
very many species can grow on both sides, though usually with
different periodicity, Para rubber, for example, ripening its
seeds on one side of the mountains in February, on the other
in August.

In South India the chain of the Western Ghats causes a heavy
fall of rain in the south-west monsoon on the western side, while
the north-east monsoon is comparatively dry, so that there is
a great difference in the climate of the two sides, and many
species are confined to one or the other. This contrast in climate
and vegetation between lee and weather*sides is also well shown
in the trade belts in the tropical Pacific islands, large and small,
and is very marked in the Andes, in the section from 10° to SO""
south of the equator. The wind striking them is usually the
easterly trade wind, and their western side is almost completely
dry. Farther south the eastern side is comparatively dry, be-
cause of the westerly winds from the Pacific Ocean. Chains
that run north and south are of greater importance in this con-
nection than chains that run cast and west, regarded simply
as mountain chains causing differences in rainfall, for the question
is less complicated Avith change of temperature following lati-
tude. But from the general historical point of view of geo-
graphical distribution, the east and west chains, by forming
barriers to the plants spreading south or north with the ad-
vancing or retreating cold of a glacial period, have been, in all
probability, of enormously greater importance than the chains
that run north and south. An immense number of species, and
even genera, have probably perished against the chain of moun-
tains that runs east and west with few gaps from Spain to
eastern Asia.

The effect of the drier climate on one side of a chain of moun-
tains M'ill generally be to encoiu-age a more herbaceous type of
vegetation. So long as there is a reasonable amount of rainfall,
not too much concentrated into one period of the year, the usual
type of covering of the soil, in countries that have not been dis-
turbed by ice periods, or by man, is forest. But below a certain
amount of rain, forest does not seem readily to survive, nor to
occupy new ground, even if it survive upon ground that was

CH. V] HINDER THE DISPERSAL OF SPECIES 43

forest in days of greater moisture. The general tendency, there-
tore, of the change of climate brought about by a chain of
mountains transverse to the prevailing damp ^nnd, is to en-
courage the growth upon the lee side of herbaceous and shrubby
plants which can stand greater extremes of drought and to
make It very difficult, if not impossible, for the forest species
whether trees or undergro^vth-herbs and shrubs, to travel into
the drier country. A complete barrier may thus be offered to
the passage of some species, while others, that would have been
quite unable to pass the level forest, may be enabled to pass
easily by the development of a mountain chain at a later period
like the development of the Andes in Cretaceous times.

Distribution of rainfall and moisture of the air is of even
greater importance to a plant than total rainfall. The largest
rainfall in the world is at Waialeale, in the mountains of "the
Hawaiian Islands; it is also well distributed throughout the
year so that the place is always wet, with no dry season at all
As the result, it has a flora of a very moisture-loving kind
Uierrapunji, in Assam, which has almost as great a rainfall, but
badly distributed through the year (April 29 inches, Mav 50
inches, June 110 inches, July 120 inches, August 78 inches
September 57 inches, October 13 inches, and the other five
months only 14 inches amongst them), does not show this, but
has a vegetation which almost suggests a dry climate.

Kandy in Ceylon has a very steady mean temperature just
over 75° F., and a rainfall Avell distributed through the year (the
twelve months have approximately 5, 2, 3; 7, 6. 9; 7, 6, 6; 11
10, and 9 inches, total about 82), and though there is a ''dry
season" in February and March, the flora is distinctly forest of
the ordinary rain-forest type. In the dry zone of northern and
eastern Ceylon lies Anuradhapura, with a total rainfall of
55 inches, distributed mainly in the north-east monsoon from
October to April (3, 1, 2; 7, 3, 1; 1, 2, 3; 8, 10, 9). In a hot
climate like Ceylon, a fall of less than 4 inches in a month is
practically negligible, so that there is really a long drought from
January to September, broken only by the April rains, and the
flora is of the dry-forest type, with comparati^'ely few species
in common with Kandy, only about 90 miles away. Calcutta,
on the edge of the tropics, with a hot sun, and a rainfall of
66 inches, is equally a "dry" chmate (rain 0-4, 1, 1-3; 2-3, oQ,
11-8; 13, 13-9, 10; 5-4, 0-6, 0-3). Going to the other and damper
hemisphere, at Rio de Janeiro, also on the edge of the tropics.

44 CAUSES WHICH FAVOUR OR [pt. i

one finds a place with only 40 inches of rainfall (5, 4, 5; 4, 4, 2;
2, 2, 2; 3, 4, 5) which shows as much of the character of rain-
forest and a wet climate as does Kandy with 82 inches. Evi-
dently the distribution of the rainfall, and of the humidity of
the air, which largely goes with it, is of much greater importance
than the actual total. Rio, with 40 inches, is better suited to
plants needing a moist climate than Cherrapunji with 470; its
svm is not so hot as that of tropical Asia, and its season of less
rainfall coincides with the weaker sim of June-September.

Change of distribution of rainfall, if at all sudden, usually
coincides with the presence of a mountain chain. The presence
of the mountains may alter the periodicity of the rain, as in
Ceylon (abo^•e), when the only plants that can cross the boundary
will be those that can alter their periodicity; or it may completely
alter the rainfall, as in the case of the Andes, where the flora is
very markedly different on the two sides. Gradual change, on
the other hand, will usually accompany gradual change of rain-
fall. Change of dampness of air, again, if permanent between
one place and another, will involve differences in the plants in
their reactions to moistiu-e, and some will be more drought-
resistant than others.

Change of temperature is usually of a more permanent, or
regularly recurring nature, especially in the tropics. At Colombo
in Ceylon, for example, the maximum is usually about 88° F.,
the minimum about 75°, all the year round, except for a small
increase from PVbruary to May. At Rio, on the edge of the
tropics, there is more range, from say 98° absolute maximum
in summer to 52° absolute minimum in winter, and at Nuwara
Eliya (ele^'ation 6000 feet) in Ceylon the absolute maxima and
minima are about 81° and 28°, with much greater daily ranges
in dry than in wet weather. The farther one goes from the
equator, or the higher in the moimtains, the greater the range
on the whole, whether annual or diurnal, and the range is
also greater the farther one goes inland from the sea. The
extreme variation of all is reached b}- going both north and
inland, to the centre of northern Siberia, where it may touch
80° in summer, and — 60° in winter.

More rapid change of temperature is experienced in ascending
a mountain, the mean falling about 3-4° F. for every 1000 feet
of ascent. Correlated with this is the rapid change of the com-
position of the flora, as compared with the change experienced
in going north or south at the same level and under the same

CH. V] HINDER THE DISPERSAL OF SPECIES 45

conditions. The plants in the mountain garden in Ceylon are
very different from those in the gardens in the "low" country,
not from any special wish to keep the collections distinct, but
from the great permanent difference of 20° (F.) in the mean
temperature, though the highest and lowest of both stations may
easily be reached in the same day at the same place in Europe
or North America.

Every function in every plant has a temperature (the mini-
mum) below which it will not go on, a temperature (optimum)
at which it will best go on, and a third (maximum) above which
it ceases. As these differ for every species, one kind of climate
will suit one, and not another, though there is no doubt that
species may become acclimatised (cf. Chapter iv). If the ex-
tremes of temperature come at a season when the functions
concerned are not being performed, they may be easily with-
stood, as for instance the great cold of winter in North Siberia,
which does not kill the conifers there. Extreme cold, when un-
seasonable, does at times kill out species, but the loss is usually
recoverable, especially as it is only necessary for the plant to
regain a foothold in societies of plants of Avhich it has already
been a member.

Light, again, changes too gradually from place to place for it
to be supposed that it has any appreciable effect in opposing
a barrier to any species. Species from one part of the equatorial
tropics do just as well in another part Avith much less intense
light, or vice versa. It is in general only in descending into deep
water that there is any great change in light over a large area,
and even there some plants are found below the limits of darkness.

Wiiid is chiefly of importance in an indirect manner, according
to whether it is wet or dry, and according to its direction in
reference to that of the mountains, but if very strong, it may
alter or prevent the growth of some species. On the west coast
of Britain one may often see trees blown into a one-sided type
of growth, and a little more wind would prevent their growth
altogether. A cyclone may uproot so many trees that it may
render passage through a country possible for herbs which can
quickly seize upon the vacant spots before the growth of the
forest once more suppresses them.

Though climatic differences are thus of such enormous im-
portance one must be careful not to say of any species that it
has certainly reached its climatic limit, when one has regard to
the very slow and gradual acclimatisation that is practised by

46 CAUSES WHICH FAVOUR OR [pt. i

nature. If one carried seeds of Hydrocotyle asiatica (p. 30) from
Ceylon to the south of New Zealand, and planted them there,
they would probably refuse to grow, yet nature has gradually
acchmatised the species to both regions. Many species range
4000-8000 feet vertically in the Himalaya; seeds from the higher
levels produce plants much more at home in Europe than seeds
from the lower levels, and might spread much more rapidly in
cooler climates than the latter. Travel may be much slower in
a vertical direction, where conditions change comparatively
rapidly, than in a horizontal.

Finally, we must go on to consider what are probably the
most important positive causes favouring or hindering species
in their dispersal; barriers are obviously negative. These causes
may also be classed in general as ecological, depending on
some peculiarity inherent in the plant itself, often described as
being an "adaptation" to something or other. We have already
considered in Chapter ii one of the most important of these
—the method of dispersal of the plant— and must now go on
to deal briefly with the others. In my published papers I have
perhaps not allowed enough for ecological barriers, but I am
not sure that they are sufficiently permanent to do more than
delay spread, rarely to completely stop it.

Take, for example, the wide differences seen between trees,
shrubs, and herbs. The flora of the wetter tropical and southern
regions of the globe, and of large portions of the north, prior to
the great clearances made by man in recent times, consisted
mainly of trees. These had, it is true, more or less of herbaceous
undergrowth, but there was comparatively little open country
covered with herbs suited to a life exposing them to the sun and
the wind. Even in much of Europe, Asia, and North America,
that is now covered with herbaceous or shrubby vegetation,
there appears to have been forest over a great part of the country
during Tertiary times.

It used to be generally supposed that the Angiosperms com-
menced as herbs and that trees were a later de\'elopment, but
this view is now usually reversed, and the herbaceous form is
looked upon as the younger. The change of view dates largely
from a paper by Sinnott and Bailey (99) in Avhich they marshal
the evidence from paleobotany, anatomy, phylogeny, and geo-
graphical distribution, etc., showing that it all points in the
same direction, to the conclusion that herbs on the whole are
the younger form of vegetation.

CH. V] HINDER THE DISPERSAL OF SPECIES 47

Now this conclusion, taken just as it stands, is open to exactly
the same objections to which, as we shall presently see Ao-e and
Area is subject. One must not say that all trees are older than
all herbs, or that such or such a tree is older {i.e. as a species)
than such or such a herb. One must work with averages of
species, and keep to the same circle of affinity. One may with
reasonable safety say that ten allied herbs, belonging say to the
family Leguminosae, are on the whole probably younger {i e as
independent species) than ten alhed trees belonging to" the same
family, but one cannot say with any approach to certainty if
even of probability, that ten herbaceous species of Piperaceae
are younger than ten woody Proteaceae.

But, in general, there is little doubt that the bulk of the
chiefly herbaceous families, like Compositae or Cruciferae, has
developed in comparatively recent times, while the bulk of the
chiefly woody families, like Euphorbiaceae or Rubiaceae is
probably very old. It must be clearly understood, however
that this is not saying that the families Compositae and Cru-
ciferae are younger than the Euphorbiaceae or Rubiaceae, but
that the great development of the herbaceous type has probably
talven place since the glacial period, the gradual desiccation of
climate, and other causes, have rendered vast spaces of country
which were formerly largely covered with forest, available for
the growth of herbs of open ground.

So long as a region is covered M-ith forest, no herbaceous veae-
tation can succeed that cannot live in the shade, or (in the case
of deciduous forest) vegetate before the leaves of the trees have
grown so much as to make the shade too deep. There is little
evidence to show that herbaceous vegetation can actually invade
and replace forest without assistance from desiccation of the
climate, or from man or animals, but a good deal to show that
the reverse may happen, and that forest may overwhelm and
replace herbaceous vegetation.

Another point that must not be forgotten is that "trees" as
a whole have not descended from a single tree ancestor. The
group is extremely polyphyletic, i.e. its members have arisen
mdepcndcntly from many different and often unrcIaU-d an-
cestors. Within the same genus one often finds trees or shrubs,
and herbs, e.g. in Solanum, Hypericum, Euphorbia, Senecio,
Phyllanthus, Ficm, Urtica, etc. It is evident that for nature to
form a tree from a herb or shrub, or vice versa, is not a specially
difficult or unusual feat.

48 CAUSES WHICH FAVOUR OR [pt. i

But we must go on to consider the advantages or disadvan-
tages in the matter of spreading about the world that arise from
herbaceous or M-oody nature. It is clear that a herb will in general
go through its generations more rapidly than a shrub, and still
more quickly than a tree. A herb producing seed in its first
year may get three or more generations, and as many chances
of dispersal, whilst a shrub, starting at the same time, is getting
one, and may get from ten to thirty for the single opportunity
offered to a tree. It is thus evident in the first place that the
chance of rapid dispersal to a distance is much greater for the
herb, and in the second that the chance of forming a new species,
by whatever method it may be evoh^ed, is also much greater in
a given time.

It must, however, be clearly understood that dispersal is
chiefly conditioned by the barriers which have already been dis-
cussed. Though the Compositae, for example, developed into a
herbaceous type, and though they developed a firstrate mechan-
ism for dispersal, they would not be so widespread and abundant
to-day were it not that the north temperate regions of the world
were largely cleared of forest by the ice in the glacial period, that
large areas became more open on account of desiccation of
climate, and that they were enabled to spread widely by the
development, often in comparatively recent periods, of the great
mountain chains which form an almost continuous track leading
over a very great proportion of the world, upon which they were
able to move above the limit of the forest, and often aided by
the formation of landslips (p. 37). One can clearly see that had
the world remained comparatively flat, and covered by forest,
to the present time, the Compositae to-day might be little more
widespread and abundant than say the Dijisacaceae.

One may thus point to the development of herbaceous habit,
with the capability of living in open ground exposed to the sun,
as an ecological featiu'e which has made possible the compara-
tivel}' rapid and extensive spread of certain families, the spread
being accompanied by a correspondingly rapid development of
new forms, whether species or genera. But that rapid and wide
spread was only rendered possil^le by the incoming of certain
physical conditions to which these plants proved suited. It is
quite possible, if not probable, that these families, and even the
herbaceous type suited to open ground, are really very ancient,
but were confined to small localities, and never able to spread
widely, till the new conditions rendered it possible. There is

CH. v] HINDER THE DISPERSAL OF SPECIES 49

reason to believe that given sufficient time, and no interference
by man, forest would once more replace the open herbaceous
vegetation of the damper parts of the globe.

Other types of habit may have entirely different effects upon
spread. Water plants can obviously only spread so long and so
far as there is water available (leaving out of account in this
place all negative factors like barriers of temperature, etc.,
already considered above), parasites can only spread with their
hosts, saprophytes only with the presence of the necessary pro-
ducts of decay in which they live, epiphytes with the presence
of sufficient moisture, etc. Halophytes can spread wherever the
ground is sufficiently salt, mangroves where it is muddy and
covered by a quiet sea at high water. Climbers as a rule can
only go where there are plants sufficiently tall upon which to
climb. Xerophytes or plants of dry climates, once formed, will
be able to advance into dry country until the drought becomes
too great for them to survive, and so on.

So long as a plant remains of average (mesophytic) type,
suited to an average damp climate and good water supply, it
may have an enormous territory possible of occupation if only
no barriers interfere, while a plant that becomes very specialised
in these respects may be limited in its capacity for spreading to
little more than the small area upon which it commenced. As
Thiselton Dyer says (94. p. 311), "The Nemesis of a high degree
of protected specialisation is the loss of adaptability,"

General evidence seems to indicate that it is not improbable
that in the Tertiary period the world as a whole was better
suited to mesophytic vegetation than at present, and hence it
is not unlikely that the earlier species not only gained in the
mechanical way described on p, 34, but also found fewer
barriers to their spread. Later formed species, on the other
hand, as they could not survive if not exactly suited to the con-
ditions in which they were evolved, would be increasingly likely
to find themselves with climatic or other ecological barriers to
further spread at no great distance away. A progressive speciali-
sation of climate and other factors seems to have been going on
in the world since the Tertiary period, the comparatively damp
and uniform climates of the latter being replaced by every
variety from very damp to very dry. Hence the more recent
species tend to become more and more specialised to match the
climates. To quote Guppy, "when one finds Salsola Kali upon
the Devonshire coast, upon a Chile beach, and upon the uplands

W.A, 4

50 CAUSES WHICH FAVOUR OR [pt. i

of Tibet, one can hardly doubt that here a very ancient type of
plant finds its still more ancient conditions of existence." On
the other hand, many species of very local range seem to be
suited to very local conditions, and more or less incapable of
further spread without further modification. Some Utricularias
in South Brazil, for example, are specialised to grow in Bromeliad
pitchers, and can only go where those exist. Copeland (18)
mentions the case of Stenochlaena areolaris, which is epiphytic
on Pandanus utilisswms only, and confined therefore to places
Avhere that grows. It seems not impossible that some Mesembry-
antheinums in South Africa are specialised to suit the exact
climate in which they grow, and are thus rigidly localised.

It is thus highly probable that at times very local species may
in reality be much older than from the area occupied one would
be inclined to think. This, however, does not affect the soundness
of the hypothesis of Age and Area to be advanced below, but
merely goes to show that ecological barriers may often be very
effectual.

An ecological factor which is of the greatest importance to
a commencing species is the type of vegetation into which
it is born. In the natural state of the vegetation of a country,
the ground in any place is covered with an assortment of plants
which is found to be fairly constant in its composition so long
as the general conditions are much the same. This assortment
is termed a plant society or association, and upon the chalk
downs, for example, or the moors of Yorkshire, one finds much
the same society, made up of much the same proportions of its
various members, in places far removed from one another. Con-
sequently, if a new species is evolved at a given place, and can-
not enter the society that exists there, it will die out again by
the simple action of natural selection. The instant that it is
produced, it will have to undergo a strenuous struggle for exist-
ence, but if it pass successfully through that, it may succeed,
and may spread with the society which it has entered, and
ultimately also enter other societies.

As the number of plants, and their variety, in any society,
increases, the entrance of a newcomer probably becomes in-
creasingly difficult. The society is said to be in progress from
an "open" condition to a "closed" one. But as Clements has
said, a society is never in a state of stable equilibrium, and
though one may regard it as perfectly closed, it may yet be able
to admit new members. The most conspicuous plant upon the

CH. V] HINDER THE DISPERSAL OF SPECIES 51

chalk downs south of Cambridge is Festuca ovina. Suppose
however that this plant in its dispersal had not yet reached th^
downs. They would, none the less, be covered by a society of
plants which might be very numerous, and which we mi^ht
thmk closed. Yet when the fescue appeared, there can be little
doubt that it would soon secure a foothold.

An association of plants ultimately passes its zenith and be-
comes gradually superseded by another, the process being known
as succession (16). "The pine... gave place at length to the oak
and the oak... yielded in its turn to the beech, the periods M-hen
these three forest trees predominated in succession taUyino-
pretty nearly with the ages of stone, bronze, and iron in Dent
mark" (68, p. 372).

The more closed an association is, probably so much the more
difficult will a newcomer find it to obtain any foothold and by
so much will its dispersal be retarded. One will expect that
most newcomers will find it quite impossible to gain a footing
at all, but that every now and then (as in the case of Elodea the
famous "American water- weed" of the last generation which
spread so rapidly through the waters of Western Europe, though
only the female plant was introduced) one will do so knd will
spread, more especially to those places which the association
concerned already reaches.

In many instances, of course, a plant in its tra^•els will come
across a type of vegetation into which it cannot spread at all
and which may thus, if broad and wide enough, form a complete
barrier. If an ordinary herb, accustomed to a good water supply,
and to hfe in the open sunshine, comes across a stretch of
country which is either a forest or a desert, it will be held up
in this manner, and whether it can cross will depend upon its
mechanism for dispersal, upon the width of the barrier, and
upon other factors. Forest trees arriving at a desert will un-
doubtedly be stopped, but when they meet a herbaceous asso-
ciation, in a country where the rainfall is sufficient, will probably
spread at the expense of the herbs, and cover previously open
ground with trees. One may see this going on at the edge of a
pine wood, or on any small clearing made by a peasant in a
tropical forest. There is hi lie evidence for the occurrence of I lie
reverse process, without the aid of desiccation of the climate or
something of the kind.

Changes of conditions Avill make great differences to the rate,
or even to the possibility, of spread, often by the effects they

4—2

52 CAUSES WHICH FAVOUR OR [pt. i

produce upon the composition of the plant societies that occupy
the ground. Farrow's Avork u])on the changes in the plant
societies upon Cavenham heath (36) made by the exclusion of
rabbits may be quoted as an example. A new disease may arrive
in a district, and a plant that was previously very common may
fall an easy prey to it; the more common it is the more likely is
it to suffer badly.

From a general distribution point of view, of course, geo-
logical changes, with the changes that they cause in climates, in
barriers of sea or mountain, and the like, are by far the most
important in this connection. They have been so fully discussed
in geological books that there is no need to enlarge upon them
in this place.

In this connection we must briefly mention the action of man,
which in recent times has become by far the greatest help or
hindrance to dispersal, though in the consideration of Age and
Area we have endeavoured to deal with the vegetation as much
as possible as it was before his interference. By clearing of
forest, opening of roads, making fires, cultivating the ground,
introducing grazing animals, carrying seeds, voluntarily or in-
voluntarily, about the world, and in many other ways, man has
made, and is making, the most enormous differences in the
vegetation of the globe, sometimes faAOuring the spread of a
species, sometimes retarding it, sometimes destroying a species
in whole or in part.

Other features, again, must be considered, which M-ould hardly
come under any of these heads, and yet which may make a great
difference in the actual spreading of species. Suppose a country,
comparatively empty of species, united to another by a broad
belt of land, which is gradually sinking. Then the first species to
arrive across it may reach almost the whole country at once,
while later ones may only reach the centre, and require to take
an immense period to spread about.

Summary
As a rule, a new plant of a given species springs up not far
from its parent, so that transport is at most a few yards. Even
if a species only travelled a yard a year, it might in a million
years (a mere detail in geological time) travel from London to
the Shetlands, Dresden, and the Pyrenees, or on an open plain
might cover a million square miles. The whole surface of the
globe might be covered in less time than it is now supposed has

CH. V] HINDER THE DISPERSAL OF SPECIES 53

elapsed since the Eocene period— a portion only of the time
during which the flowering plants have existed. It is clear that
delay, and not acceleration, of spread has been the rule.

The various barriers that species may meet with are then con ■
sidered, first those purely physical such as seas or mountains
then those due to change of climatic factors from place to place'
which are partly physical, partly depend upon the constitution
ot the plant, and lastly barriers (or at times aids to dispersal)
depending upon the type of vegetation into which a plant may
try to intrude, such as forest or open grassland, or various
associations of plants, some of which may suit it and some not.
A herb, for example, may spread ten times as rapidly as a tree.
The effects of specialisation in structure and function are also
pointed out; the more specialised a plant becomes, the more
limited its possible range.

The general impression which I have tried to convey in this
and preceding chapters is, that until man began to interfere
upon the large scale with cultivation, war, and clearing, the
dispersal of plants from one place to another must have been a
matter of the most extreme slowness.

CHAPTER VI

AGE AND AREA

The hypothesis which I have termed (123) Age and Area is not
a sudden discovery, but has grown up in my mind during a period
of about twenty years of work, in the study more especially of
the flora of Ceylon and its neighbouring countries. It will per-
haps prove of interest, therefore, to sketch this gradual develop-
ment, enlarging for the purpose a short account recently pub-
hshed (135).

Going out to Ceylon in 1896, and remaining there till 1911,
I had constant occasion to refer to the volumes of Trimen's
Flora (37). There I gradually found, someAvhat to my surprise,
that the many species which are confined to that country [en-
demic to the island) were usually confined also to small areas
within it. Now at that time I held the view, then very usual,
that these endemics were specially adapted to the local con-
ditions, and it seemed very remarkable that they should be so
rare in those very conditions. If they were specially adapted
to Ceylon, therefore, it could hardly be to the general conditions
of the island (whatever those might be), but must be to strictly
local conditions within its area. Now this was the explanation
that was usually applied to the very numerous species that were
endemic in such regions as West Australia or South Africa, and
it was therefore clear that there were no differences between the
endemics of an island and those of the mainland, and that any
explanation that fitted the one would fit the other.

Still more remarkable, therefore, did the facts appear, Mhen
I gradually began to study in greater detail the local distribu-
tion of the endemics, and found that they were not, as a rule,
confined each to one spot or small region characterised by some
special local pecuharity in conditions. Had this been the case,
they might have been supposed to have been evolved to suit
such spots, which in actual fact might be found without any
local species upon them.

Coleus elongatus, for example, was confined to the summit of
Ritigala Peak (p. 14), a minute area, and was found now^here
else in the world; but C. inflatus, another endemic species, was
common all over the high mountain regions of the island. C,

PT. I, CH. VI] AGE AND AREA 55

malaharicus also occurred there, but was found in the plains
too, and in the mountains of South India, while C. barbatus, the
remaining Ceylon species of this genus, covered the range of
a malabaricus, and also occurred almost throughout tropical
Asia and Africa. It seemed hard to believe, when one could not
see in plants like these four Colei any characters whatsoever
that one could point to as advantageous or as disadvantageous,
that there should exist internal characters so distinct and
different as would enable C. barbatus to cover so enormous an
area, and C. malabaricus a smaller but still large one, while
keeping C. inflatus confined to the Ceylon mountains, 'and C.
elongatus to a few square yards on the peak of Ritigala. No
differences in efficiency of the dispersal "mechanism" could
account for the differences in area covered by these allied species
of the same genus.

This question of areas occupied roused my interest, and a
little study soon showed that species, endemic or not, occupied
every col^ceivable area, from a few square yards to a large part
of the surface of the globe (the "area" being determined by the
outlying stations, even if the plant be absent from the area, or
part of the area, between them). On the older view that dis-
tribution was chiefly determined by degree of adaptation to
conditions, it had come to be more or less unconsciously sup-
posed that species were divided into a comparatively few' "suc-
cessful" species covering large areas, and a great number of
''unsuccessful" covering small. This view proved to be a very
inadequate explanation of the very striking facts of distribution
that have just been outlined above. We shall return to this
subject again under Endemism.

Of the 809 species of flowering plants endemic to Ceylon, less
than 200 were confined to what one might, by a stretch of the
imagination, regard as single spots, and about half of these
occurred upon the tops of single mountains or small groups of
mountains (121 ). On the summit of Nillowe-kanda, for example,
which is a mere precipitous rock, there are found, and there
only, Acrotrema lyratum, Stemonoporus reticulatus, and Ocbna
rufescens; on Ritigala (p. 14) three species, on Hinidun-kanda
(another somewhat isolated mountain) three, on Adam's Peak
ten, one of which extends into a valley 2000 feet below; and so
on. Evidently the investigation of areas occupied bid fair to
furnish interesting information, and I devoted much attention
to it. A careful study of the remaining three-quarters of the

56

AGE AND AREA

[PT.

endemics showed that they were found upon areas of various
sizes up to the full extent of the dry or the wet zone of Ceylon,
or more rarely of both, but that the numbers grew smaller as
one went up the scale toward the larger areas, Trimen in his
Flora had rendered yeoman service to the student of areas, by
attaching to every species a note to the effect that it was Very
Common (VC), Common (C), Rather Common (RC), Rather
Rare (RR), Rare (R), or Very Rare (VR). A study of the
localities in which species had been found showed that as a rule,
though with a good many exceptions, a VR species oeoirred in

ioon«tles

one place only, or two close together, R in an area about 10-30
miles across, RR in one 30-60 miles across, and RC and C in
areas larger yet, while VC referred rather to unusual common-
ness on areas represented by C. •

The three diagrams here reproduced give the ranges of a
number of the earlier endemic species in Trimen 's Flora of
Ceylon, belonging to the classes VR, R, and RR. The VR species
are, it will be seen, usually well localised, though a few (5 in the
diagram) have been recorded from two widely separated localities,
joined by a wavy line. The R and the RR species, however,
cover areas that overlap one another in every jDossible way, and
look something like the rings in a shirt of chain mail. Nowhere
do the areas occupied by two endemic species coincide, except

CH. VI] AGE AND AREA 57

(approximately) in the case of a few VR species, which occur
together on the same mountain-top. The VR species that occur
m the large forests have each their own location. Noav a little
consideration will soon show that from the point of view of
evolution to suit local conditions this is a very remarkable state
of affairs. It is of course obvious that if a species newly evolved
upon a small area does not suit the conditions that obtain upon
that area at the time in ■which it was evolved, it Avill be promptly
killed out; but while this is so there is no actual need to imagine
that it was evolved specially "adapted" to those conditions.

If two species A and B grow in overlapping areas, both must
be growing in the coincident portion; and what keeps A from
growing into the rest of B's territory, and D into ^'s? It has
usually been insisted that it was because A was adapted to its
own territory, and B to its own. But when one considers that
the conditions are never the same from one spot to the next,
nor from one year to the next, this would mean a most wonderful
adaptation if the species were not to grow into each other's
territory, especially when one remembers the many more widely
distributed species that occur in both. In reality the case is
more complex, for there are at least a dozen overlapping at any
one point, while in Ceylon the soil is essentially the same through-
out the greater part of the island, the flora was practically "all
forest before the arrival of man, and the rainfall varies very
much from year to year in quantity and distribution. It was
evident that the old ideas of particular adaptation were un-
tenable, and this view was enormously strengthened by subse-
quent discovery of the way in which species were grouped in a
country.

This conclusion was confirmed by later work on the Podo-
stemaceae (124), a family of water-plants of smooth rocks in
rushing tropical and subtropical mountain streams only. Here
there is nothing to which the many genera and species can be
adapted, for the conditions are the same for all, and could not
be equalled for uniformity in a laboratory of the temperate zone.
They grow only upon a smooth rigid substratum, from which
they take no food; all grow in water, and have no climatic
differences, no difference in circumambient medium, in light, or
in any other factor. And yet there have evolved many genera
and species, with very striking and bizarre differences between
them. Evidently it is not necessary to ha\e local condi-
tions to which to be adapted in order to ensure that evolution

58 AGE AND AREA [pt. i

shall go on. Cordyceps, with 60 species on insects, is a similar
case (73).

Another popular theory about localised species like these
Ceylon endemics is still strongly held, though the one just con-
sidered (local adaptation) has suffered somewJiat of an eclipse
with the gradual decay of the hypothesis of natural selection.
Like the first, though completely at variance with it, this second
explanation is also founded upon natural selection, but some-
what less obviously. It is to the effect that species on very small
areas are really in process of dying out. It is evident that they
could not have arisen by aid of natural selection upon areas so
small, and therefore they are assumed to be moribund. This
hypothesis is supposed to be supported by the facts of fossil
botany, wliich unquestionably proves that many species have
existed in the past and no longer occur in the world to-day.
There is, however, nothing to show that the two cases are paral-
lel, except in a few instances where there is good evidence that
the present existing species once covered a mucli larger area. It
was simply assumed that such a species as Colcus clongatus had
once occupied more ground.

Like the previous theory, however, this explanation breaks
down when applied to the very striking facts of the distribution
of endemics in Ceylon. How can species be dying out in a chain-
mail pattern, like the R and RR species given in the diagrams
above? And why were there so many more with the smallest
areas (VR) than with areas not q\iite so small (R and RR)? Had
one arrived in Ceylon just in time to see the disappearance of
a considerable flora? Was the dying-out becoming less and less,
and if so, why? This graduation of the areas of endemic species
from many small to few large was a most difficult point indeed
to explain upon this supposition of dying-out, just as it had been
for the theory of local adaptation.

Again, why did so many of the "very rare" endemics choose
mountain-tops as a last resort? There were many widely dis-
tributed species, with very restricted areas in Ceylon, but these
did not choose such places, and why did they not? The "dying-
out" explanation supposes endemics to have ascended moun-
tains as the last refuge from the invading flora of the plains, but
in such a small and imiform country as the wet zone of Ceylon
it is hardly possible to suppose that there Avas, for example, a
separate Eugenia at every few miles; whilst some of the moun-
tains with endemic Eugemas did not even rise directly from the

59

c«- viJ AGE AND AREA

^^hJ"! fT/ ^'^^ P^"*'^^- ^"^ ^^^^>^ ^^d *h^ ^"demies
chmb right to the summits of the hills? One would have ex-
pected to find them at varying heights, pursued, so to speak,
by the widely distributed species before whose onslaught thev
ivere dymg-out, instead of finding, as is not infrequently the
case, a great gap in elevation between the two. It sucr^e^ts an
unnecessary degree of alarm about the coming competitfon, and
further suggests that they are not so incapable of adaptation to
new conditions of life that they need fear it. If they can undergo
the great adaptive changes necessary to reach a summit of 5000
feet or more, they must have a very fair capacity for modifica-
tion, and should be able to hold their own against the intruders
Queries like these might be put by the dozen (131, p 351 and
p. 88, below), and the explanation now under consideration
could give no answer. Clearly the theory of dving-out was as
untenable as that of local adaptation, so far as the Ceylon
endemics were concerned. There is no doubt that a considerable
number of species here and there, especially within the range of
the glacial periods, may be looked upon as dying-out, or some-
times as locally adapted, but these are comparatively few and
tar between, and the mass of local endemics, particularly in the
tropics, cannot be looked upon as coming within these cate-
gories.

Just before leaving Ceylon I published a Catalogue of the
flora (115), which rendered the task of enumerating the species
with their distribution a much simpler affair, and on reaching
Kio I began this work. Di^'iding the species into three groups—
those endemic to Ceylon, those found in Ceylon and Peninsular
India (cut off by a line from Calcutta to the north of Bombay),
and those with wider distribution abroad than tliis— I found
that the endemics were (VC 19), C 90, RC 139, RR 136, R 192,
VR 233, increasing fairly steadily from top to bottom of the list

Examining the distribution in Ceylon of the species (which I
termed "wides" for short) that occurred outside the island to
a greater distance than merely into the peninsula of southern
India, it was found that the areas they occupied in the island
■\vent in the reverse order, being (YC 221), C 402, RC 313 RR
209, R 159, VR 144.

If now, leaving out of account the somewhat uncertain VC
class (its greater uncertainty is largely due, as already explained,
to the fact that it is not based on actual area occupied), we
number the other classes 1 to 5 (i.e. by degree of rarity, not of

60 AGE AND AREA [pt. i

frequency), then the number 1 attached to a species will mean
that it has the maximum, the number 5 the minimum, dispersal
in Ceylon. And we can find the average distribution of a group,
whether Ceylon endemics or widely distributed, by multiplying
the number under each head from 1 to 5 by the number of that
head, adding up all the marks thus obtained, and dividing by
the total number of species. Thus we obtain :

Widelj

•dis-

Endemic species

tributed

species

No. of

No. of

^

Class
1. C

species Marks
90 X 1 90

species
462x1

Marks
462

2. RC

139 X 2 278

313 x2

626

3. RR

136 X 3 408

209 x3

627

4. R

192x4 708

159 X 4

636

5. VR

233x5 11G5

144x5

720

Total 790 2709

1287

3071

y repres

iented by 3-4

2-3

Now the actual number of species under each of these heads
in the whole flora is (VC 285), C 670, RC 555, RR 429, R 415,
\B. 455. If we take the average rarity of the last five classes,
we find it to be just over 2-7. The average rarity of an endemic
we have seen to be 3-4, and of a wide 2-3, while the remaining
species, which are endemic to Ceylon and South India, show a
rarity of 2-7, the same as the whole flora. The difference of 1-1
in average rarity between wides and endemics represents over
a quarter of that between the most and the least widely dis-
tributed species (1 and 5, difference 4). In other words, the
most widely distributed species in Cej^lon, on the average, are
those that show a distribution abroad to a greater distance than
merely to Peninsular India; then follow those that reach the
peninsula, and the least widely distributed are those that are
found in Ceylon only. Taking the estimates of actual area given
above for the different classes, the differences actually found
indicate that an endemic has an average area about 40 miles
in diameter, a "wide" one of 80 miles, or four times as large.

A cursory examination of other floras soon showed that their
species behaved in the same way, occupying areas of all sizes,
overlapping in the same manner, and with their endemics
occupying areas from many small up to few large, and the wides
the re\erse. At the same time, the figures for the Ceylon flora
indicated clearly that this graduation of areas, wides largest,

CH. VI] AGE AND AREA 61

Ceylon-Peninsular-India next, Ceylon endemics least, showed
not only for the grand total, but also for every family of 14 or
more species. It was clear that any one group of allied species
behaved like any other group, and it was therefore obvious that
nothing but a mechanical explanation Avould serve. Natural
selection could not act on all plants alike with even pressure.
The only possible mechanical explanation seemed to me to be
age, which would almost necessarily act alike upon all. If one
supposed the "wides" to be (on the average) the oldest, and to
have been the first arrivals in Ceylon, they were thus allowed
sufficient time to spread to the largest extent. On the way, they
would give rise, perhaps somewhere south of the middle of the
peninsula, to the species now found in Ceylon and Peninsular
India; these would be next oldest, and would spread in Ceylon
to the second degree of distribution. The Ceylon endemics would
arise in Ceylon, and on the whole probably later still, from one,
or more likely both, of these groups, and being the youngest,
would have spread the least. It seemed to me that I was at
last provided with a simple and feasible explanation of the dis-
tribution of species, though it involved a great break with the
older ideas, inasmuch as it indicated that the Ceylon species
were confined to Ceylon simply because they had been too young
to have had time to spread abroad.

It is clear, of course, that age in itself can effect nothing, but
it allows time for the various factors that are active in distribu-
tion to produce their effects. The mechanical regularity of the
figures giA'cn above demands a mechanical exjilanation. and the
only possible one seems to be that age is mainly responsible for
the distribution, or in other words, that the various factors that
are operative produce an average or resultant effect — so much
dispersal in so much time. Dispersal therefore becomes a measure
of age, except in so far as barriers, physical or ecological, inter-
fere. Distribution is very slow, and probably the \'ast majority
of species have not yet reached the limits that they might reach,
if sufficient time were allowed.

The greatest change from the older view of matters, however,
consists in the fact that since one can no longer acccjit either
the view of local adaptation or that of relic nature, for the great
majority of local species, and as these show definite numerical
relationships to those of wider distribution that occur beside
them, one must regard the two classes as related. But as area
goes vnth age, the endemics must be the younger, and must

62 AGE AND AREA [pt. i

therefore be looked upon as in general descended from the wides,
and as young species just commencing their careers.

I called this hypothesis, that on the average the area occupied
by species in a country depended upon their age within that
country, by the convenient jingle of " Age and Area" (123, p. 337,
footnote), and from the very first I was careful to point out that
this result Avas only true when averages of about 15 allied species
were taken. People, however, have nearly always insisted upon
applying the rule to individual cases, and then complaining that
it does not fit the facts. In regard to the facts that have just
been discussed, for example, they say "there are many VC
endemics, and a lot of VR wides, so it must be wrong." A simple
illustration will perhaps make my position more clear.

Suppose that five wides are approaching Ceylon (then attached
to the mainland), spreading at a uniform speed, and let the dis-
tance from the foremost of A to the foremost of B be represented
by 2, that from B to C by 2 also, and so on. Then A will reach
Ceylon first, and when B reaches the island A will occupy there
a space represented by 2. When C arrives A will occupy 4, and
B 2. Ultimately they will occupy spaces represented by 10, 8,
6, 4, and 2. Now let each give rise in South India to another
species a, /3, y, S, e, each always at a distance beliind its parent
represented by 2. Then a Avill arrive in Ceylon simultaneously
Avith B, ^ Avith C, and so on, and these Ceylon-Peninsular-Indian
species Avill ultimately occupy areas represented by 8, 6, 4, 2,
and 0. And if, lastly, each species, Avhen it has reached a dis-
tribution in Ceylon represented by 2. gives rise to a Ceylon
endemic, then if avc subtract 2 from the figures of distribution
of all the preceding species, Ave shall get the distribution of the
endemics. This Avill be, for the endemics derived from the Avides,
8, 6, 4, 2, -, and for those derived from the Ceylon-Peninsular-
Indian species 6, 4, 2, — , — .

NoAV the most Avidely distributed endemic, derived from A,
the first Avide to arrive, Avill have a range of 8, Avhile three out
of five of the Avides, and three out of four of the Ceylon-Penin-
sular-Indian species Avill have ranges of 6, 4, or 2, considerably
less. If one attempt to apply the rule to individual' cases, it is
at once liable to break doAvn. Bvit if Ave add up the dispersal of
all the Avides, and divide by the total of species, Ave get
(10 + 8 + 6 + 4 + 2 =) 30 ^ 5 = 6 as the average range of a
AAdde, 20 ^ 4 = 5 as the average range of a Ceylon-Peninsular-
Indian species, and 32 -^ 7 = 4-5 as the average range of an

^«- ^i] AGE AND AREA 63

endemic, figures which obviously agree with the rule. If one
take the figures in groups, one may safely say that the wides
will range the most, the endemics the least.

In the same way, one must work, not only with groups of
species, but with groups of allied species, which will have more
or less the same dispersal-mechanisms and the same reactions
to their surroundings. If A, B, C be three species with wide
separation in relationship, and great differences between them
in regard to habit, dispersal-method, or other things, their rates
of dispersal may be entirely different, and A may travel ten
times as fast as C. But with a group of ten allied species one
will be fairly safe.

Changes of condition, again, might evidently completely alter
the relative rates of dispersal of species, or might even stop
some of them altogether. And we must also take account of the
presence and action of barriers, already discussed, rememberino-
that some forms may cross a barrier when it has become quite
impassable to others.

Age shows clearly in the distribution figures because it always
pulls the same way, whereas other causes of dispersal will either
tend to cancel one another by pulling different wavs, or more
commonly to exert a practically uniform pull upon' a group of
allied species, so that when two groups of allies are compared
one will be able to see the relative effects of age upon either'
In any single species its effects are liable to be completelv hidden
by those of some of the other causes, just as the effect of gravity
which is admittedly universal, is hidden in the case of an aero-
plane, a balloon, or a mo^-ing bullet.

The most recent expression of the rule of Age and Area so
far pubUshed (133) is as follows:

The area occupiedi at any given time, in any given country
by any group of allied species at least ten in inimber, depends
chiefiy, so long as conditions remain reasonablv constant, upon
the ages of the species of that group in that country, but mav
be enormously modified by the presence of barriers such as seas^
rivers, mountains, changes of climate from one region to the
next, or other ecological boundaries, and the like, also by the
iietiuu of man, and by other causes.

Extensions, Avhich will be considered below, have since been
given to Age and Area, which appears to be a general law cover-
ing all or nearly all the plants now existing upon the globe, and
1 Determined by the most outlying stations.

64 AGE AND AREA [pt. i

to have determined their distribution thereon, in broad outhne.
When stated thus, it would appear to be almost axiomatic, but
for a very long time the simple effects of age upon dispersal have
been lost sight of, under the widely held view that distribution
was rapid, and that local species were either local adaptations
or Avere dying out.

The Ceylon figures gave strong evidence in favour of Age and
Area, but confirmation of the most satisfactory kind was soon
obtained by working out the distribution of the flora of New
Zealand (127), employing north and south diameters of areas
occupied {i.e. in the direction in which the islands run), and
obtaining these by actual measurement. This flora followed the
law with great exactness, as a quotation of actual figures will
show.

Range

in N.Z.

(miles)

Endemics

Wides

1.

881-1080

112

201

2.

641-880

120

~ 77

3.

401-640

184

53

4.

161-400

190

38

5.

1-160

296

301

Further work was then carried out upon various other similar
phenomena, the conclusions already made being confirmed by
the Orchids of Jamaica, CalUtris (a Conifer) in Australia, and
the flora of the Hawaiian Islands. A study of the ferns there
and in New Zealand also gave the same result, showing that the
law was probably quite general.

Breakwcll (13) studied the grasses of Australia, and found
that while the species of very wide distribution showed an
average rarity there of 3, those confined to Australia and New
Zealand or Asia showed 4-1, and those confined to Australia
only an average rarity of 4-6, the figures agreeing exactly Avith
those already given. He also found that the genera showed the
same thing, and that it showed in Panicum alone, while several
of the larger genera showed a very close agreement.

Taylor (105-6) has studied the endemics of Ncav York and
of the Bahama islands, obtaining results that harmonise quite
well with the general theory of Age and Area. In the latter
case, it was noticed that the difi'erence usually seen between the
distribution of the endemics and the wides was not nearly so
large as usual. This may be due to one or more causes; it may
1 Largely undoubted introductions of recent years.

CH. VI] AGE AND AREA 65

be that the pecuhar conditions of the Bahamas, with their
sterile soil and considerable droughts, suit the endemics— which
must have been developed in them, and have had, as just ex-
plained, a strenuous struggle to become estabhshed, and which,
therefore, should be unusually well suited to the local conditions.
Although the parent species were able to survive there, the
endemics were probably better suited, and woiUd therefore be
able to overtake the former to some extent.

Summary
Studying the flora of Ceylon, it was very soon noticed that
there were enormous differences between the areas occupied by
species of the same genus, some of which were endemic to the
island, some not, and this led on to a study of areas occupied in
general, when it was soon found that the endemic species occu-
pied, on the average, the smallest areas in the island, those found
also in Peninsular India (but not beyond) areas rather larger,
and those that ranged beyond the peninsula the largest areas
of all (again on the average). The two current theories about
endemic species— that they were local adaptations, suited to
special local conditions, and that they were relics — proved to
be incapable of explaining the facts when it was found, as was
ultimately done, that the areas occupied, both by endemics and
by widely distributed species, were arranged in a graduated
series, the first from many small to few large, the second in the
opposite direction. It was not possible to suppose that local
adaptation should exist in this graduated manner, nor that there
should be many relics at the final stage of dying out, and suc-
cessively fewer at all the stages leading iip to that. Some
mechanical explanation was necessary, and the only simple and
reasonable one was that the area occupied increased with age.
The actual quotation of the Age and Area hypothesis, as so far
developed, is given on p. 63.

CHAPTER VII

AGE AND AREA (contd.). CONFIRMATION
BY PREDICTION

i'J-iOT^cb Howe I?

/ ^'STtares

ohj-ity /?

tAuLcJclamL I.
CampbelL I^ ''

^ -^ TxtioodjLS I

(confirmation of the general idea advanced in the hypothesis
can be easily obtained by applying it to predict what will be
found in certain places
or under certain circum-
stances. Many success-
ful predictions of this
kind have been made for
the area comprised by
New Zealand and its
surrounding islands (the
Kermadecs, 420 miles
north; Chathams, 375
miles east; and Auck-
lands, 190 miles south).
It will be well to instance
a few of these.

To begin with simple
cases (129) ; from the fact
that to the east of these
outlying islands the
soundings are in general

of enormous depth, while the water between them and New
Zealand is comparativel}' shallow, one may infer that their
floras have in general the same sources of origin as that of New
Zealand. This is indicated also by the very few species in them,
other than their own local endemic species, which do not occur
in New Zealand. If they had received their flora by casual
transport over sea, one would expect that it would be a miscel-
laneous assortment, and that it would not show any numerical
relations to the flora of the larger island. But as such relation-
ships are shown very clearly one may, I think, take it for granted
that the connection was by land, at least so far as the bulk of
the flora of these islands is concerned. Now in this case it is
clear that on the hypothesis ol' Age and Area, this flora should
in general be very old in New Zealand, or it could not have

"^JVTouj^Cuxrte- /

New Zealand and outlying islands. The
dotted line is the 1000-fathom limit.

PT. I, CH. VII] CONFIRMATION BY PREDICTION G7

reached the islands before they were cut off. In the case of the
Chathams, more particularly, where except New Zealand there
IS no other source for the flora than casual arrivals by sea bv
currents which also run close to New Zealand, this should be
the case. The Kermadecs must have lain fairly near to any in-
commg northern current of plants, the Aucklands probably to
any southern mvasion, and both these islands therefore may
contain plants that were too late, or only just in time, to reach
New Zealand at all, but this does not apply to the Chathams

One will therefore expect, upon the hypothesis of Age and
Area that while on the average all the floras of these islands will
be old, and therefore widespread, in New Zealand, those plants
that reach the Chathams will be the oldest, and most widespread
Actual examination soon shows that those plants that reach all
three groups, and which are therefore, by hypothesis, about the
oldest of all in New Zealand in their own circles of affinity, show
the maximum possible range in New Zealand, rangino- "it from
end to end. Three of the five are Compositae, including La^eno-
phora Forsten, which is endemic to New Zealand and the islands,
and the others are Samolus repens and Deyeuxia Forsteri In
my papers upon New Zealand I have divided the plants into
ten classes by range, instead of the six of the Ceylon flora. The
average rarity of a plant in New Zealand, including all the flora
is represented by 5-6, and the rarity of these five species is repre-
sented by 1. Those plants that reach two groups of islands,
which must also, by hypothesis, be very old forms, have a raritv
represented by an average of 1-5. Of these plants there are 16
species in class 1, 4 in class 2, and a solitary species in class 8,
about whose identification there is some doubt, and whose in-
clusion brings the average from 1-2, at which it would otherwise
stand, to 1-5.

There are a great many species that reach only one group of
islands, and these show on the average less range in New Zealand,
but it is very noticeable, th.at just as was predicted above, those
of the Chathams show a much greater average range than those
of the Kermadecs or Aucklands. The average rarity for a species
reaching the Chathams is represented by 1-7, and it would be
1-5 were it not that, though there is otherwise no species below
class 4, there is one conspicuous exception in class 9, which
brings up the average figure. This exception is the Tainui of the
Maoris {Pomaderris apetala), which they assert sprang from the
rollers or skids of their invading canoe the Tainui, and which

- IS
5—2

68 AGE AND AREA [pt. i

only found in a short range on the north-west coast of New
Zealand. It is fairly clear, from the marked way in which it
forms an exception to the rule as regards distribution of these
island species, that this legend is probably the truth, and that
this species therefore may be regarded as an introduction, and
omitted from the indigenous flora. Even including it, however,
the average figure for the Chatham plants is 1-7. The species
that reach the Kermadecs show an average rarity in New Zea-
land represented by 3-6, and as each 0-1 represents 12 miles in
range, this means that they range New Zealand on the average
228 miles less than the Chatham species. Their range, however,
is still much greater than the average for the species of New
Zealand as a whole, which is represented by 5-6, or 240 miles
less than the Kcrmadec species. The number of species in the
different classes ranges down to class 7, and in class 9 there is
again a species which may be looked upon as an exception
— Ij)07noea palmata, which is possibly carried by sea currents,
and may have reached both Kermadecs and New Zealand in
this way, as they are washed, where it occurs, by the same
current.

Lastly, the species that reach the Aucklands (only) show an
average rarity in New Zealand represented by 3-5, or practically
the same as the range of the Kermadec species, with the lowest
species in class 4. The prediction as to range in New Zealand of
the various species reaching the islands is thus fully verified, and
this success lends great support to the hypothesis of Age and
Area. There is no conceivable reason Avhy ranging to one or
more of these little groups of islands, and to any one of them^,
though they differ Avidely in climate and geology, should make
a species more widespread in New Zealand than the average,
unless it be the mere fact that to have been able to reach the
islands at all it must have been above the average age in New
Zealand, and thus have had more time in which to spread.

This is confirmed by the fact that there are in New Zealand
many species, both widely distributed (reaching Australia, etc.)
and endemic, which do not reach the islands at all. These by
hypothesis should be younger, each of course, as already ex-
plained, in its own circle of affinity, than the species which reach
the islands, and should therefore be less widespread in New
Zealand. There are 213 such "wides," and they show an average

^ Kermadecs in latitude 29°-15, volcanic; Chathams in 44°-20, schists,
volcanic and tertiary ; Aucklands in 50°-35, igneous, mostly volcanic.

CH. vii] CONFIRMATION BY PREDICTION 69

rarity in New Zealand represented by the figure 4-3, i.e. 0-7
greater than the largest figure for any that reach the islands
(Kermadecs, 3-6), or a range of 84 miles less. This difference
between the two groups comes out in a very striking way if we
place the figures in columns by classes :

Ran<Te in

N.Z.

Reaching Not reaching

Class (miles)

islands

islands

1

1001-1080

45 X 1 =45

35

2

881-1000

19 X 2=38

39

3

761-880

3x3=9

26

4

641-760

3 X 4 = 12

28

5

521-640

1x5=5

19

6

401-520

1 X 6=6

17

7

281-400

3 X 7=21

12

8

1 01-280

1x8=8

14

9

41-160

2 X 9=18

7

10

1-10

. X 10 = .

10

78 with 162 marks

213 with 919

Average rarity

20 ( =range of 940 miles); 4-3 (664 miles).

Difference 2-3, representing

; 276 miles of range.

If one subtract from

class 10 in

the second column

about a dozen that

are probably introduc-

tions, one

gets 201 with 799

marks, an average of 3-9, representing 228

miles less range tlian the first column.

There are also 98 species that are endemic to New Zealand
and one or more of these island groups, but not found elsewhere
in the world. These have an a^-erage rarit}' in New Zealand
represented by 2-9, or in other words, they are a good deal more
widely ranging in New Zealand than tliose species which reach
Australia, etc. (enumerated above in the second column), but
do not reach these little islands. The difference of 1-4 in average
range represents 168 miles. Now here, still more than in the
previous case (p. 68), there is no conceivable reason why ranging
to these little groups of islands (and to auy one of them, though
they differ completely in climate and geology) should make these
endemic species more widespread in New Zealand than many
others whose distribution touches Aiistralia, etc., unless it be
simply that being older, they have had time to reach the islands,
and to range more widely in New Zealand itself.

Another interesting point shows in the table given above,
which also indicates the greater age of these species, whether
wide or endemic, that reach the outlying islands. The wides that
reach them show 45 in class 1, whose range covers Stewart
Island, a separate island near to the south coast of New Zealand,

70 AGE AND AREA [pt. i

and only 19 in the next class, which does not in all cases include
Stewart. In other words, most of these species were so old that
they were also in time to reach Stewart before it was cut off.
The endemics that reach the islands also show 41 in class 1 and
21 in class 2, but the wides that do not reach these islands
(last column in table above) show 35 in class 1 and 39 in class 2,
indicating that they were on the whole a good deal younger, so
that many of them were not in time to reach Stewart. The
endemics that do not reach the islands show 52 and 60 in these
classes respectively, in the same way.

That these outlying islands of New Zealand are not a special
case may be seen by comparing with the flora of Great Britain
those of some of its outlying islands. If we take the Orkneys
(north Scotland), Colonsay (south-west Scotland), Clare (west
Ireland) and the Scillies (south-west England), islands widely
separated, and differing very much in climate and geology,
and if we take in these, at random (37, 108), the families
Ranunculaceae,Caryophyllaceae, Leguminosae, Orchidaceae, and
Gramineae, we find that Avhile (going by the LondoJi Catalogue^
8th ed.) the average distribution of a species in Great Britain
is to 47 of the vice-counties out of 112, the 175 species of these
families that occur on the islands mentioned range on an average
to 71 (or 50 per cent, more), whilst those that reach three or four
of the islands show an average range of 99. The facts are exactly
parallel to those for the islands off New Zealand, though of course
not so striking, as the islands are very much closer to their
mainland.

Before going further we must once more consider the reserva-
tions which are laid down in the statement of the hypothesis
in the preceding chapter, and whose misunderstanding seems the
cliief stumbling-block in the way of an acceptance of Age and
Area. It is easy to pick out of the list of "wides" reaching the
islands a few that have less range in New Zealand than other
wides that occur there and do not reach the islands. The hypo-
thesis is often treated in this manner, and then rejected for non-
agreement with actuality. It must not be forgotten that if it
could be applied in such minute detail we should have at our
command a theory that would (explain more facts in distribu-
tion and phylogeny than any other that has ever been suggested.
Too much is expected of an hypothesis which claims no more
than to be a useful guide, and the reservation, that it must not
be applied to a group of less than ten allied species, is ignored.

cii. VII] CONFIRMATION BY PREDICTION 71

It may be applied to less if it be simply desired to gain an argu-
ment from greater or less probability to add to other arguments
in favour of some point, but when it is to form a main argument
it must be applied to at least ten allied species at once. By this
means the exceptional species, of which there are many, will
be lost in the crowd, and also a group of species will be obtained
M'hich react to their surroundings in much the same way, have
more or less the same rapidity of dispersal, and so on. On
averages there can be no question about the wider dispersal in
New Zealand of the Chatham plants, though individuals can be
found with little dispersal there. The herbaceous Compositae
may be enormously younger in the islands than the woody
Leguminosae, for example, and also younger in New Zealand, yet
by virtue of their better dispersal mechanism, and the fact that
they are herbs, may be much more widely distributed in the
latter, and may even have started much later from New Zealand
than the Leguminosae (which could hardly cross a strait) and
yet have reached the islands. Both groups, however, obey Age
and Area, though they cannot be compared with one another
as to relative age.

If there were, again, a" great change of conditions between
New Zealand and the Chathams, or any serious barriers like
mountains, this would completely alter the hst of plants that
might arrive. One must remember all these provisos in dealing
with the distribution of plants, but none the less one finds that
by keeping to the Age and Area rule as enwiciated, and dealing
always with groups of allied species, results may be obtained
that are fairly reliable.

To return to predictions, another upon the following lines (132)
was equally successful. A family will rarely arrive in a country
as a group of genera simultaneously; some will arrive sooner
than others. On the average, therefore, in any circle of affinity,
the families with several genera Avill be older in that coiuitry
than those M'ith one or two, as it is all but impossible that their
first genus should only arrive at the same time as the solitary
one of another family. This being so, we shall therefore expect
the larger families of New Zealand to he better represented upon
the outlying islands than the smaller, as being older. On Stewart
Island, at the south end of New Zealand, avc do in fact find this
to be the case, as the following table shows :

AGE AND AREA

[PT. I

Family

Represented in

represented

in

In New-

Stewart hv

Not represented

New Zealand

by

Zealand

there

(genera)

(families)

(families)

(per cent.)

(families)

1

36

13

36

23

2

15

6

40

9

3

15

12

80

3

4-5

10

9

90

1

6-10

9

8

90

1

over 10

6

6

100

.

91

54

59

37

We may even take the genera, and consider those represented
by most species in a coimtry to be the oldest in the country.
Testing this on the flora of Stewart Island, we get:

Genus
represented in
New Zealand by

In New
Zealand

Represented in
Stewart by

Not represented
there

(species)

(genera)

(genera)

(per cent.

) (genera)

1

155

32

20

123

2

54

22

40

32

3

29

20

68

9

4-5

29

23

79

6

6-10

36

32

88

4

1 1-20

16

15

93

1

over 20

10

10

100

329

154

46

175

Thus, just as with the families, the proportion of genera
represented in Stewart shows a steady increase with the in-
creasing number of species in the genus from 20 per cent, of
those with one up to 100 per cent, of those with more than
20 species.

If we test the same question on the farther outlying islands
of New Zealand, the Kermadecs, Chathams. and Aucklands. we
find that the average size of a family that reaches all three
groups is 47 species, of a family reaching only two is 14, reaching
one 5, and of a family reaching none is only 2. A similar result
follows a test of the genera. This fact also shows in the flora of
the islands off the British coast mentioned above.

Or again, as the wides are, according to hypothesis, the oldest
forms, one will expect to find them the best represented in the
floras of the outlying islands of New Zealand. In New Zealand
itself the wides form about 18 per cent, of the flora in number
of species, but when we pass over into Stewart Island, the plants
reaching which must, by hypothesis, be older on the average
than the plants of New Zealand proper, we find that the wides
form 30 per cent, of the flora. In the plants that reach Stewart,

CH. vii] CONFIRMATION BY PREDICTION 73

and also one of the three outlying groups so often mentioned,
the wides form 41 per cent., when two groups are reached they
form 64 per cent., and of those plants that reach Stewart and
all three, i.e. Kermadecs, Chathams, and Aucklands, they form
80 per cent. The result agrees exactly with the prediction, con-
firming the hypothesis in a very striking manner.

Or we may predict that the far outlying islands will have a
large proportion of forms in common with one another and with
Stewart, all being old in New Zealand, and that the jjroportion
will be much larger than that in common with New Zealand.
In actual fact, one finds 81 per cent, of the Stewart families,
67 per cent, of the genera, and even 40 per cent, of the species,
on the other islands, Avhile of the plants that occur in New Zea-
land, but not on Stewart, only 32, 17, and 15 percent, respec-
tively occur, an enormous difference. The prediction is com-
pletely borne out by the facts, and it will suffice to quote one
or two instances. The Kermadecs have 30 per cent, of the genera
that occur upon the Aucklands, 1200 miles away, in a totally
different climate, and only 19 per cent, of those of New Zealand.
Of 52 species occurring outside the Kermadecs, as well as in
those islands, 30 occur in the Chathams, and even 5 in the
Aucklands; and so on.

One may in the same way predict a great similarity between
the floras of the islands off the British coast, above mentioned.
On examination, one finds, in the five families before considered,
that their 70 genera have in the British Islands an average of
4-7 species, against 3-4 for the whole flora. Whilst about 37 per
cent, of the whole 175 species of these families are confined to
one island, 24 per cent, are foinid on two, 19 per cent, on three,
and 19 per cent, on all four, widely separated, and widely
different in climate, etc., though they be. The average occur-
rence of each species is upon 2-2 islajid groups of the four.

Or we may predict that the genera which are common to the
islands and New Zealand, taking at least two groups of the
three, will be very old genera, and consequently in general will
be large genera in large families. This is so obvious when one
comes to make a list, and finds it composed of Ranunculus,
Cardamine, Lcpidium, StcUaria, Colobanthus, Geranium, etc.,
that it hardly needs any further elaboration. The 32 genera
upon the islands in the first half of the New Zealand flora show
an average size of 144 species, against an average for the world
of only 12. Only five of them, Corynocarpus, Coriaria, Panax,
Samolus, and Calystegia, are below the average in size.

74 AGE AND AREA [pt. i

Or, lastly, one may take the endemics of New Zealand and
the outlying islands, and make predictions about them. We
have just seen that on the whole, each in its own circle, the
larger families and genera of a country will be the older in that
country. Now endemic species, by hypothesis, occupying small
areas, will be on the whole younger than the wides, as already
pointed out, and one will therefore expect the older families,
which have had the longest time in the coiuitry, to produce the
most endemics. That is to say, that the endemics should belong
to the largest families in the country, working in averages. The
same rule should of course apply to the genera. If now we test
this on New Zealand and its surroimding islands, we find that
in New Zealand and its outlying islands there are 22 families
above the average size, with 1100 species, of which 890 are
endemic to New Zealand or the islands, or 80 per cent.; there
are 69 families below the a^^erage, with 292 species, of which
only 110 are endemic, or 37 per cent., an enormous difference.
In Stewart Island, all the 19 local endemics belong to the 15
largest families of New Zealand, and ]0 of them to the three
largest families in Stewart, and the same thing holds for the
local endemics of the other outlying islands.

In the same way, one finds that the {local) endemics of the
Kermadecs, Lord Howe Island, and Norfolk Island, all islands
which must have lain more or less in the track of the invasions
of New Zealand by plants from the north, belong chiefly to
those families and genera of their floras which have also reached
New Zealand, i.e. to the oldest families and genera contained
in them.

On the supposition, which follows from Age and Area, that
the wides have given rise to the endemics (p. 61), one will expect
most endemics to occur in those regions where there are most
wides, and not, as on the theory of dying out of endemics would
rather be the case, in those regions where there are fewest
wides. In fact, this is at once seen to be the case, whether in
New Zealand, its outlying islands, or in Ceylon or elsewhere.

Age and Area is thus seen to be a hypothesis by whose use
one may discover great numbers of new facts, and as so far all
the predictions made by its aid have proved to be correct, on
verification, the result is to afford great support to the hypo-
thesis itself. Over 90 such predictions as those mentioned above
have now been made and verified, and one may, one is inclined
to think, regard the hypothesis, in the absence of any rival'

CH. VII] CONFIRMATION BY PREDICTION 75

explanation, as sound. The question now is to bring it into
accord with other views, theories, and facts, which often con-
flict with it, or apparently so.

Summary

This chapter is devoted to a few instances of the very suc-
cessful way in which Age and Area can be used to make pre-
dictions about distribution. For example, it was predicted— and
verified— that the outlying islands of New Zealand would have
a flora which was very old in New Zealand, and therefore very
widespread there. In fact it was found that on the average its
species ranged nearly 300 miles more in New Zealand than did
those that did not reach the islands. Further, those endemic
forms that reached the islands were found to be more widespread
in New Zealand than the species of its flora that reached Aus-
tralia, etc., but did not reach the islands— a result only explicable
by aid of Age and Area. Parallel results were obtained by a
study of the floras of various islands off the British coast, from
the Orkneys to the Sciflies.

The reservations already laid down, that Age and Area must
only be applied to groups of at least teyi species, and to groups
of allied species, are then once more insisted upon.

The successful prediction that as, on the Avhole, the larger
families and genera in a country will be the older, therefore the
flora of the outlying islands will be chiefly composed of these, is
then described. Other predictions indicate that the farther out
one goes the greater will be the proportion of wides, that the
outlying islands will have much in common, especially of large
genera in large families, that the endemics, both of New Zealand
and the outlying islands, will belong mainly to large families
and genera, and that most endemics will occur where there are
most wides. All these predictions proved successful, and as this
method has now been used over ninety times with no failures, it
is evident that Age and Area has strong foundations on which
to rest.

CHAPTER VIII

AGE AND AREA {contd.). INVASIONS

The acceptance of the hypothesis of Age and Area involves
various changes in our way of looking at many problems of
geographical distribution, and of other branches of Botany, and
we must go on to further illustrate its (published) implications
and possibilities. The facts upon which it is based, as illustrated
by the preceding two chapters, are so clear and so definite that
they cannot go without an explanation; either one must accept
Age and Area, or one must find some other explanation for them
— a thing that no one has yet attempted.

If the distribution of plants about the world has been very
largely the result of their age, it is clear that it should be com-
paratively easy to make predictions about it, as has already
been shown. The very first prediction I employed (127) was the
following, which will serve as a text for this chapter :

Let W be a species arriving at the centre of New Zealand from
abroad, and following the rule exactly in its dispersal (there is
reason to suppose that it would not do so unless the direction were
east and west, not north and south as in New Zealand; but this
does not affect the prediction). Such exactness probably never

PT. I, CH. viii] AGE AND AREA. INVASIONS 77

occurs in real life, but by taking groups of ten allied species one
may cancel out many of the effects of chance differences. This
dispersal is indicated by drawing a right-angled triangle, which
expands regularly till after a certain time it reaches both ends
of New Zealand. As it does so, and covers more country, JV is
supposed to give rise casually to new species (shown at every
increase of 200 miles of range, their locations of origin obtained
by dra^ving numbers at random). These new species, El to 10,
spread like the parent, as is shown by the similar triangles, so
that when W reaches 0 and 1000, E 1 reaches 120 and 920.

If noAv we divide New Zealand into ten zones by drawing a
vertical line at every 100 miles, and count in every zone the
number of endemics found there (derived directly or indirectly
from W), we find the number small at each end, and with a
maximum (or at times two or more) near the middle. In the
present case, for instance, the numbers in each of the zones from
left to right are: 0, 3, 5, 8, 9, 8, 7, 3, 2, 2.

If we obliterate the left-hand half of the diagram, we get the
result of entrance of W at one end of New Zealand, and find the
maximum near that end; it always tends to be near the point
of entry. If the entrance be not at a jjoint, but at a zone, e.g.
from 300 to 700 miles, at the level of E 2 and 3, then, if one
omit -E 1, 2, and 3, one finds that the remainder give the figures:
0, 1, 3, 5, 6, 5, 4, 1, 0, 0, a similar but shorter curve.

Occasionally, with a casual development of new endemic
species, it so happens that the cur^'c may show two, or even
more, maxima with a slight drop between them, but to have one
maximum only is the general rule.

One might therefore j^redict that one would find the endemic
species of any genus in New Zealand to form such a curve, and
this proved to be the case for every genus in the flora. A few
examples are here given;

Zone

in miles

0

to
100

100
to
200

200
to
300

300
to

400

400
to

500

500
to
600

600
to
700

700
to
800

800
to
900

900
to

1000

Ranunculus

2

3

5

7

11

12

18

18

10

Drimys

Pillospomm

Colobanihus

11

11

11

2
11

2

3

3

3

2
(1
3

1

r,

4

1

5

Coprosma
Meirosideros

12
8

12
8

15
8

16
8

17
5

18

G

18

6

IG
2

15
1

12

1

Ligusticum
Veronica

6

1
6

1
10

1
14

2
15

7
39

8
41

9
43

38

6
26

JJtricularia

3

3

3

1

1

1

1

1

.

.

Pimelea

4

4

5

5

7

8

8

G

5

4

78

AGE AND AREA

These curves show many things. The first point that appears
from their study is that the maxima are not casually scattered
all over New Zealand, but occur in masses at particular regions,
e.g. chiefly at the far north, at a little south of the middle of the

\*© Solomon I*

;\\eiiitt i«

x^y

C'"^

Net

t51«.

rm

%>

\*X-V

I >

/ \ ' jaorfolk

\ * /

Aft-' »»

d«* I'

'ZtAuAND

Soundings in the New Zealand area. Numbers inserted here and there

give the depth in fathoms at those points. (From the Annals of Botany.)

100 fathoms. 1000 fathoms.

South Island, and at the north end of the same island. These
last two groups are so close to one another that they are some-
what confused together. Of the examples given above, Pitto-
sporum and Metrosideros have northern. Ranunculus and Vero-
nica southern, and Drimys and Coprosma central maxima.

CH. Villi

INVASIONS

79

These are bare and unvarnished facts, and though found bv
aid of the hypothesis of Age and Area do not depend upon it in
any way, but may be examined upon their own merits It is
clear from them that the previous distributional history of these
groups of genera must have been quite different, and it would
seem to pomt to the conclusion that the present flora of New
Zealand has been the result of at least three distinct invasions
ot plants from elsewhere, which probably had their centres at
the pomts, north, south, and central, where the masses of
maxima occur.

This is confirmed by examination of the actual genera, for the
northern group is composed of families characteristic of Indo-

<)n

__

BO

^

\

70

\,

60

S

\

—

40

\

30

^

^

7.0

\

\

10

^

^^

Er

^

.^^

1 1

"

W

'

-30

0 -40

0 -so

0 -60

0 -70C

-60

0"^

0 -10

W

■10

80

Endemics

^Malaya, probably indicating an invasion thence, the southern
group belongs to Ranunculaceae, Umbelliferae, and other
families prominent in the northern hemisphere (the only ex-
ceptions being Stylidiaceae and Centrolepidaceae, both southern
families), and the central group to Stackhousiaceae, Campanu-
laceae, Violaceae, etc., which may perhaps have come from
Australia.

If now one add together all the species of the genera of the
northern invasion that occur at each zone of 100 miles from
north to south in New Zealand (including Stewart Island), one
obtains the curves shown above, from which one may peiiiaps
mfer that the invasion was at about 0-300 miles from North
Cape. The two curves fall off very steadily towards the south,

80

AGE AND AREA

[PT. I

but that for endemics much more rapidly than that for wides,
the maximum in each case being at about the same spot, and

230

/

i

/

ZIO

>

/

\

/

\

180

\

170

130

1

/

no

100

/.

^

.^^

/

/

^

-^

K

1

90

/^—

/

/

\

//

/-'

\

w

7

\

/

f-

\

50

/

40

/

^

\

10

— '

1

Wides

Endemics
(N.Z.only)

^i'm'

100 -200 -300 -400 -500 -600 -700 -800 -900 -1000 -1080

CH. VIII] INVASIONS 81

the minimum at the same. The more rapid fall of the endemic
curve is to be attributed (on the hypothesis of Age and Area)
to the fact that they are in general younger, and so have not
had time to sj^read so far.

Treating the southern invasion in the same way, one obtains
the curves on p. 80, showing both endemics and Avides falling
off towards the north. The latter are shown with a double curve;
the upper shows the grand total of wides, but many begin at
the north and do not occur in the far south, showing that they
probably really belong to the northern invasion. Subtracting
these gives the lower curve, and the diminishing distance be-
tween these two curves shows the way in which these species
diminish southwards. The endemics, being more numerous, are
split into two curves, one endemic to New Zealand only, one
endemic to New Zealand and the outlying islands (Kermadecs,
Chathams, Aucklands).

These curves provide a very formidable argument against the
supposition that endemics are djang out, for if so, why does
their number show its maximum with that of the wides, and
fall off to a minimum at the same point with the latter?

They also illustrate various other points. For example, from
the much steeper curves of the southern invasion, one may
probably infer that it was much younger than the northern,
both wides and endemics having had less time to spread widely
in New Zealand. This is confirmed by the fact that both northern
curves, and that for southern wides, show no break of any kind
between 500 and 600, where Cook's Strait lies, while that for
southern endemics shoAvs a marked drop there, indicating that
when this group (the youngest of all, by hypothesis) came along,
the strait was at any rate beginning to be formed. The same
feature shows in a much more marked way at Foveaux Strait,
between the last two figures in the curves; even the northern
"wides" show a drop here, and the southern endemics an enor-
mous one.

The greater age of the northern invasion may also be inferred
from the fact that in it the mmiber of the endemics at any zone
is always at least twice as great ns that of the wides, while in
the southern invasion the ciirve for endemics goes below that
for wides at both ends, or adding the endemics of islands, below
that for Avides at the northern end.

The greater youth of the southern invasion is also emphasised
by the fact that it is composed to the extent of 83 per cent, of

82 AGE AND AREA [pt. i

herbs, while the northern has 84 per cent, of trees and shrubs,
and, as we have pointed out above, the latter will be likely to
spread with vastly greater slowness. The average areas occupied
by the species of the two invasions are much the same.

I am informed by the well-known palaeobotanist, Mrs Clement
Reid, that geology gives evidence that invasions follow directions
which offer stability of climatic conditions to their members;
polewards when climates are warming, equatorwards when
coohng. One feels inclined to infer, therefore, that at the time
of the northern invasion New Zealand was warm in the south,
whilst the Antarctic land was habitable to the northern types
of plants that largely compose the southern invasion, and which
perhaps reached Antarctica by way of the Andes, as most of
them occur in that chain. Then as the south cooled, the southern
invasion perhaps entered New Zealand, working northwards. It
is very noticeable in the ciuves for this invasion that they fall
off much more gradually to the north than to the south.

Yet other probabilities may be deduced from the figures and
curves given. The curve in the southern invasion for endemics
that reach the outlying islands is flatter even than the curve
for wides, showing that they are probably older than the average
for wides, as we have shown ab^^^e (p. 69). But if we split the
curve for wides in the same way, into two, that for the wides
that reach these islands proves to be even flatter than that for
the endemics which do so, as we should expect by hypothesis.

From the diagram given at the commencement of this chapter,
one may deduce that the average range of endemic species that
occur in the outer zones of New Zealand will be greater than
that of those that occur in the centre, for obviously those of
short range will be mainly concentrated towards the middle.
Examination of the actual figures for the southern invasion
shows that this is very strikingly the case, the average range of
all the endemics occurring in the northern half of the South
Island being about a third of that of those occurring to the
south. Not only so, but they belong in much greater proportion
to the smaller genera of New Zealand, i.e. by hypothesis (p. 71)
the younger. The long-ranging endemics of the outer zones be-
long mainly to large genera (of the New Zealand flora).

It is clear that Age and Area can be used with considerable
directness in the study of the invasions by which a country has
received its present population of plants.

c^-yui] INVASIONS 83

Summary
The application of the hypothesis to a study of the way in
which a country has been peopled by invasions of plants is
Illustrated by the case of New Zealand. If a species enter the
country and give rise casually to new (endemic) species, then,
If the country be divided into equal zones, it will generally occur
that the endemic species occupy the zones in numbers increasing
from tlie outer margins to some point near the centre at which
the parent entered. Applying this prediction to New Zealand
It was found that all the genera in the flora showed figures of
this type. Further, it was noticed that the points at which the
maxima occurred were not scattered casually all over the coun-
try, but tended to mass together in three places— northern,
southern, and central. The most reasonable explanation of this
is that these points represent the centres of correspondin<T inva-
sions. Curves are given showing the way in which both^wides
and endemics fall off, from the centres of the invasions, the latter
much the more rapidly. As the curves of the southern invasion
are much more steep than those of the northern, one mav perhaps
infer that the latter was much the older (perhaps even a geo-
logical period older), and this is confirmed by the fact that it
consists mainly of trees, while the southern is composed chiefly
ot herbs, and also by other considerations. It is clear that 4cTe
and Area can be applied with effect in the studv of the peopling
of a country with plants.

6—2

CHAPTER IX

OBJECTIONS TO THE HYPOTHESIS

Very many objections and criticisms have already been pub-
lished, and many more are doubtless to follow. A consideration
of them, however, shows that in general they are based upon a
few general principles, and that a proper understanding of Age
and Area, and of the provisos with which it is hedged round,
will go far to remove the most of them.

The first few, (1) that the numerical results are accidental^
(2) that the figures are not reliable, and will be vitiated by
further Avork, and (3) that the figiu-es can be accounted for by
changes in climate and configuration of the countries concerned,
require no discussion at the stage which Age and Area has now
reached. Far too many facts have been accumulated from too
many places, to leave room for them to be seriously advanced.
Another, (4) that the hypothesis is an assumption, has really
little bearing upon the matter. Natural Selection, and many
other fruitful hypotheses, are also assumptions, and Age and
Area has already led to new discovery.

Some writers show a confusion of thought between (5) en-
demism and endemic species. The former, if it occur in a country,
is a sign of age, for time must be allowed for it to appear; but
the endemic species are in genei-al the youngest in the country,
in their own groups of afiinity.

Some say (6) that the wide dispersal of the wides is diie to
their wide dispersal outside the country, but give no reason for
this. It utterly fails, however, to explain the graduated dis-
tribution of the wides, those that occur farthest away showing
(on the average) the maximum local dispersal (cf. the Ceylon-
Peninsular-Indian species with the wides of greater range, or
the species that reach the outlying islands of New Zealand with
those that do not). To explain such cases the most improbable
supplementary hypotheses have to be adduced (126, p. 10).

A number of objections arise from the attempt to apply Age
and Area to individual cases; such are (7) that there are many
exceptions^ — species whose area does not at all represent their
age, and the like, (8) tliat species may die out or be killed out in
part of their area, (9) that one cannot properly compare a single

FT. I, CH. IX] OBJECTIONS TO THE HYPOTHESIS 85

species to its nearest relative, (10) that a species may owe its
wide range to being part of a wide-ranging association of plants,
(11) that a species that occurs in a greater number of associations
must have taken longer to spread than one that only occurs in
one, (12) that climate produces great effects upon the distribu-
tion of a species, (13) that altitude does the same, (14) that
latitude also does the same, (15) that of two species with equal
latitudinal range the one with the greater altitudinal range will
be the older, and so on.

It has, I hope, been made clear above that the distribution of
any one species depends upon very many factors — method of
dispersal, acclimatisation, suitability to the society of plants in
which it may find itself, local adaptation, barriers of all kinds,
whether physical, climatic or ecological, individual habit of the
species itself, and so on, as well as upon mere age. With so many
factors active, it is clear that probably in no single case does
age alone determine the area upon which a species occurs. In
exactly the same way, when a baby is born, it is very rarely
possible to say of what complaint that baby will ultimately die,
yet if one take a large number of babies, living in the same coun-
try, one can say that just so many will be accidentally killed, so
man^^ will die of tuberculosis, and so on. In India one can say
that just about so many deaths from snake-bite will occur in
a year; and there are many other similar cases of reasoning upon
large numbers, where in the large figure and the long run the
result is certain, yet cannot be predicted for the individual.
And the same is the case for Age and Area, and such objections
as just quoted have really no bearing upon its validity or other-
Av^ise.

AVhen one takes groups of ten allies, and compares them with
other related groups of ten alhes, for instance, ten Mimosas with
ten Ingas — nearly related genera in the same family, living under
much the same conditions — the effects of age will show clearly,
because all the other factors in dispersal will either be pulling
the same way upon all, or will cancel one another out by pulling
in different directions. Ten herbaceous Compositae may occupy
an area X, and ten woody Dipterocarpaceae may occupy the
same area X, but the two are not comparable. In the former
case, the herbaceous habit implies many more generations in a
given time, and therefore many more opportunities of dispersal;
the parachute mechanism of the seed-dispersal enables it to
travel better; the fact that herbs of this kind grow in the open

86 OBJECTIONS TO THE HYPOTHESIS [pt. i

also enables dispersal to be more rapid, and so on. The two cases
are quite incomparable. But if the ten Compositae be compared
with ten other nearly alUed Compositae, then the effects of the
"other" factors will be much the same, and age, which is always
pulling alike upon all species, will show its effects clearly. The
greater the number of allied forms taken, and the greater the
length of time considered, the more clearly will the effects of
age show.

Other objections come under the head of comparison of un-
allied forms. For example, it has been objected (16) that herbs
must be older than trees, because they occupy greater areas,
but that all probabihty is against this, (17) that Age and Area
shows that new species must have been formed more rapidly
among trees (because there are more of them among the endemic
forms), and that this also is against probability, (18) that local
endemics are usually unrelated to the wides that grow beside
them, and are often very unlike them, and so on. What has just
been said about comparing groups of allied forms only really
covers most of these, and a reference to such works as Hooker's
Flora ofNexv Zealand, or other systematic works, will show that
a great deal too much has been made of the supposed differences
between the endemics and the wides that accompany them. In
the great majority of cases the two are allied, and if they were
unrelated, it would be a very remarkable thing that they should
show the numerical relationships that we have seen to exist.
There are a considerable number of endemic forms, especially
within the range of the last glacial period, for example in
temperate North America, which are not related to the wides
beside them, but when groups of tens are taken, these are quite
lost in the crowd, or in some cases can not find a crowd to which
they can be attached. There are, however, at most about 400
such cases in North America^, and the endemics of most of the
world, especially the countries south of the Tropic of Cancer,
are to be counted by tens of thousands. Only very rarely, again,
will one find a group of ten allied herbs, with a group of ten
allied trees closely related to it. In such a case, which Avill very

1 Sinnott (95) instances as endemics of this class Canja, Plancra, Madura,
Garrya, Sassafras, Xanlhorhiza, Baptisia, Ncmopanthus, Ceanothus, Dirca,
Dionaea, Hudson ia, Rhexia, Ptdea, Dccodon, Ilouslonia, Sijmphnricarpus,.
etc., pointincr out that many occur as fossils in the Old World, and that they
include most of the woody endemics of north temperate America. In
dealing with such, one must, as already pointed out, include the "fossil"'
area, and in any case they are lost in the crowd when not considered singly.

CH. IX] OBJECTIONS TO THE HYPOTHESIS 87

seldom occur, comparisons on the basis of age will be impossible.
But a group of mixed trees and herbs may be compared with
another allied group of the same general composition.

These considerations also dispose to a large extent of the
objection (19) that age is only one factor of many, and (20) that
enough is not allowed for the action of other factors. Age is, as
has been pointed out above, only one factor, but it is a factor
whose action can be shown in figures which no one as yet has
been able — has e^en indeed attempted — to explain upon any
other supposition. If one were dealing with individual species,
one would have to allow for each individual factor, and could
never, or very rarely, be in a position to say how much was due
to this, and how much to that. No one has yet been able to
reduce to figures the effects of any of these factors, and their
action is still accepted upon a priori considerations. The effect
of my work is to disentangle from among them the effect of
age, and to show that it is very considerable indeed; and this
should of itself make much easier the study of the effects of the
many other factors that take part in the dispersal of a species
about the globe.

The next group of objections is to the general effect (21) that
endemic forms, Avhether species or genera, are local adaptations,
suited expressly to the spots in which they occur. In one sense
this objection is a truism, for if a species or genus were not suited
to the spot where it occurred, it would die out there, so that if
it were endemic to a very small locality, it might easily dis-
appear from the earth. But the general explanatory idea which
lies behind this objection is very hardly pressed when it comes
to explaining such a series of species, arranged in "wheels
within wheels," as those of Ranunculus in New Zealand (p. 156),
or Doona in Ceylon (p. 153), and breaks down altogether when
it is once realised that endemic species and genera, as will be
more fully shown below, represent only a special case of species
and genera in general. It is not possible to explain upon any
theory of adaptability the varying areas occupied, and occupied
in a way that can be reduced to statistics which agree for each
family and area. One cannot suggest conditions that will overlap
like the rings in a shirt of chain mail, as do the genera and
species (p. 56). Nor will this view explain the increase of
cndemism as one goes sonthAvards, or outwards from the con-
tinental areas. Nor will it enable us to do any prediction about
geographical distribution whatever, though Age and Area has

88 OBJECTIONS TO THE HYPOTHESIS [pt. i

already been successfully used in this way nearly a hundred
times. Nor, again, can it explain such cases as Castelnavia, -with
seven species in an area where there are no differences in con-
ditions (126, p. 15). Above all, it will not explain the mechanical
way in wliich every group of species behaves like every other,
as has been pointed out above.

The next group of objections takes the general position that
endemics are mostly the relics of pre-existing floras ; the first is
(22) that endemics are usually rehcts in the sense of species that
are dying out; that they are old species driven into qiiiet nooks
or odd corners ; the most recent statement to this effect is that

Very many endemics owe their limited distribution to the cir-
cunistance that they are remnants of comparatively vuisuccessful
types which have been exterminated elsewhere, and which even
in these isolated floras are waging a losing fight against more
vigorous and adaptable newcomers.

This is undoubtedly true of a great number here and there
especially in the north temperate zone (particularly North America
and China), where the influence of the last glacial period was
severely felt, and so far as the first jiart of the sentence (to
"elsewhere") is concerned. We know from geological evidence
that in the Canaries and Madeira there are many generic sur-
vivals of the Tertiary flora now extinct in Europe itself, but we
have no proof that they are dying out there without change of
conditions. Age and Area has always insisted upon the reserva-
tion "so long as conditions remain reasonably constant," though
critics and opponents frequently ignore this. Guppy has re-
cently (50) shown that the endemics of the Canaries which may
be looked upon as Tertiary relics occupy more space in the
Canaries than do the more recent Mediterranean type of en-
demies, while they also extend to the Azores or Madeira, which
the latter do not. As these Tertiary relics are mainly woody, the
conditions are naturally against them so soon as man has settled
in the covmtry (cf. p. 27).

When a species is really dying out, the fact is usually due to
some change of conditions; and, as we have shown above, dis-
persal is usually so slow and to such small distances that the
species may easily be cut off by the changing conditions, and
then gradually exterminated, through no fault of its own.
Cupressus macrocarpa is probably the most generally suitable
Conifer for average sub-tropical climates, and is planted in
milhons all over the warmer parts of the world ; yet it is dying

CH. IX] OBJECTIONS TO THE HYPOTHESIS 89

•out in its only natural habitat, the Monterey peninsula of
California, probably on account of the secular drying of the
Californian climate.

The comparative rarity of seriously broken areas of distribu-
tion among endemic forms, especially south of the influence of
the last glacial period, is much against any very large amount of
dying out. One would not expect a moribund species to retain
its area intact— though it is true that with the Cycads, often
supposed moribund, this is largely the case.

It is very hard to suppose that a genus would choose certain
spots upon the globe where its species should die out in large
numbers, yet the facts of distribution require that this should
be so under this explanation. Why should the Seiiecios retire
to die, in large numbers, to Mexico, California, Bolivia, Peru,
South Africa, Australia, etc.? The larger the genus the greater
the number of local species, and the greater the number of
places in which they occur.

As this is the principal argument brought forAvard by oppo-
nents of Age and Area, it will be well to bring up other points.
If all or most endemics are to be regarded as relics, then they
must evidently be, on the whole, older than the "wides," and
the reply to another objection (23) that greater distribution may
be due to youth, rather than age, may be given at the same time.
The great difliculty is to explain why, in most countries remote
from the influence of the last glacial period, the "dying-out" is
purely mechanical. Every family and genus behaves in the
same way, whether it has or has not wides, and whatever its
habit of growth, its origin (local or foreign), or its distribution
generally. The general type of distribution — in "wheels within
wheels" — is shown below (Chapter xv) in several maps, and not
even the most determined upholder of a general dying-out can
interpret these maps into a support for his position. There is no
doubt that a large number of species and genera in the north
temperate zone may be interpreted as dying out (cf. footnote,
p. 86), but they are insignificant in number beside the endemics
of more southern regions. North-temperate America has perhaps
400, but Ceylon alone has SOO endemics, and Brazil perhaps
12,000. The latter country has 240 endemic Eugenias alone.

A still greater difliculty for the supporters of general dying-
out is to explain why there should be many more endemics at
the point of death (VR in Ceylon, for example) than there are
a little further removed from it (R), and more of these than of

90 OBJECTIONS TO THE HYPOTHESIS [pt. i

those still further away (RR, RC, and C, in diminishing numbers
as one goes up the scale). This is a general rule for all endemics
of the tropics and the south, and is impossible to explain on any
theory of dying out.

Yet another difficulty, considered below in Chapter xv, is to
explain why the endemics should belong in larger proportion to
the large and "successful" families and genera than to the small
and broken ones which we have been accustomed to consider
moribund.

Or again, why should those genera, like Gunnera, in which
there are no wides at all, behave exactly like those in which
there are such? And why do not the moribund species congre-
gate in special regions, so to speak, reserved for derehcts, instead
of choosing each its own special location? Why should many
Eugenias in Ceylon choose each its own mountain upon which
to die?

To these one may add the following notes and queries, which
if not successfully answered, are very fatal to the view that
endemics are chiefly relicts :

{a) How, on the view that endemics are relicts, is it possible
to predict Avhat has already been successfully predicted by the
aid of Age and Area?

(6) How are the facts of the regular graduation of species, of
narrowly localised endemics up, and of wides down, to be ex-
plained at all?

(c) Why is there no difference in behaviour between endemic
genera and species?

{d) Why does a genus behave in just the same way in New
Zealand (for example), whether endemic with small area, en-
demic with large, endemic in New Zealand, endemic in New
Zealand and islands, endemic in New Zealand and Austraha, or
endemic in New Zealand and the rest of the world?

(e) Why are the endemics of the same order of rarity whether
there are or are not wides in the same genera?

(/) Why should the islands round New Zealand have more
endemics the more wides they have (129, p. 332)?

{g) Why are the endemics of New Zealand least numerous at
the ends of the islands and not in the middle, and the wides the
same (128, p. 201)?

(/i) Why do the endemics that reach the ends of New Zealand
range on the average so much farther than those in the middle
(127, p. 448)?

CH. IX] OBJECTIONS TO THE HYPOTHESIS 91

(k) Why are the endemics still less numerous in proportion
on the islands surrounding New Zealand than on New Zealand
itself, and the wides more numerous? (N.Z. Wides/Endemics
301/902, Kermadecs 45/25, Chathams 69/76, Aucklands 27/72.)
(Z) Why do the endemics of both northern and southern inva-
sions of New Zealand taper down in number with the wides, but
much more rapidly, so that in the case of the southern forms
they are actually less numerous than the wides in some zones?
(m) Why, if endemics are being driven in by the wides, do
their areas almost invariably overlap and why are there prac-
tically no broken areas among them?

{n) Why do the Ceylon-Peninsular-Indian species show a
range on the average intermediate between the Ceylon endemics
and the Avides?

(o) Why are the species endemic to New Zealand and the
islands so common in New Zealand, more so than the average of
the wides in that country (129, p. 331), and why are the wides
that also reach the islands yet more common again?

(p) Why do endemics on the average occupy so much larger
an area in New Zealand than in Ceylon, even proportionately
to the size of the country (127. p. 454)?

iq) Why do fern endemics, which must on the average be
older, show greater distribution areas than angiosperm endemics
(130, p. 340)?

(r) If the wides are the younger, there is no reason Avhy they

should be specially closely related to the endemics, and Avhy

should they sliow the same arithmetical relationships throughout?

(s) Why do endemics and Avides, in the majority of cases,

belong to the same genera?

(t) Why are the endemics so often on mountain-tops and Avhy
do separate species of endemics occur for different mountains
near together (121, p. 132)?

(w) Why do the endemics belong principally to Avidely spread
and successful genera, and this even more on \'ery isolated islands
like the Chathams? The Chatham endemics belong to Geranium,
Acijjhylla, Pseudopanox, Corokia, Coprosma, Olearia, Cotula,
Scnccio. Sonc.hufi, Cyaihodes, Myrsine, Gentiana, Veronica, Carex,
Poa, Festuca. The Auckland endemics belong to Ranunculus,
Stellaria, Colohantlms, Geuni, Azorella, Ligudicum, Coprosma,
Olearia, Cchnisia, Cotula, Abrotanella, Gentiana, Veronica, Plan-
tago, Urtica, Bulbinella, Hierochloe, Deschampsia, Poa.

(f ) ^Vhy does the maximum of the Avides, in Ceylon, Ncaa'

92 OBJECTIONS TO THE HYPOTHESIS [pt. i

Zealand, etc., coincide with that of the endemics, and both
decrease together from tliat point, the endemics much the more
rapidly?

Other formidable arguments against this view are given below,
in Chapters xv, xvi of Part II.

The hypothesis of youth (within a country) and area can only
be accepted if one be prepared to accept with it the numerous
absurdities to which it leads. In particular, it involves a most
remarkable amount of rising and falling in the scale of area of
distribution, for which we have no warrant. The distribution of
the plants of the outlying islands of New Zealand (p. 66) seems
to provide a very strong case against it, for how can youth
ensure that a species shall reach more of these little islands?

"The families Tristichaccae and Podostemaceae also afford an
excellent test case for the question of age or youth, for owing to
their peculiar morphology one can say with reasonable approach
to certainty which are the older forms. He would be a bold man
who would say that such forms as Laxvia in the one family, or
Castelnavia in the other, with their violently dorsiventral struc-
ture, shown in the lichen-like vegetative body and the extra-
ordinarily modified flowers, were older than such forms as Tri-
sticha or Podostemon, Avhich are almost radially symmetrical,
and come near to the ordinary type of submerged water plant.
Yet the latter are widespread and almost universal, covering
the whole range of distribution of families, while the violently
dorsi^'ent^al forms are all endemic to comparati\'ely small areas,
Latvia, for example, occurring from Cej'lon to Bombay, Castel-
navia in the Araguaya and one other river in Brazil. It is im-
possible to talk of local adaptation in these plants, as I have
elsewhere pointed out (124); there is nothing to be adapted to.
The non-dorsiveutral forms are just as common as the dorsi-
ventral, whether in slowly or in swiftly moving water" (quota-
tion from 128).

Mrs Arber (4, p. 306) has brought up a parallel case in the
genus CaUitriche.

Some, while admitting that in general endemics are not relicts,
say (24) that the endemics of mountains, at any rate, are usually
such, especially as the wides, not infrequently, do not ascend as
high as they do. This latter fact is a strong argument against
the explanation often given of mountain endemics, that they
have retreated upwards to escape the competition of the wides
in the plains below, for it would be very remarkable if they

CH. IX] OBJECTIONS TO THE HYPOTHESIS 93

should at oiice, so to speak, retreat as far as possible beyond
pursuit.

There is no doubt that the species of mountain chains often
show much less affinity to the species of the lowlands, than do
the species of islands to those of their mainland. In some cases
there can be no doubt that such species of mountains are relics
of a flora that once occupied the lowlands, as in the case of the
many arctic species that occur upon the mountains of the north
temperate zone. In other cases the difference may be simply
due to the fact that, as explained on p. 37, a mountain chain
may act as a road for migration to the plants of another country,
which would not otherwise be able to enter the country under
consideration, by reason of unfavourable conditions. In the
mountains of Ceylon, India, Java, and most tropical countries,
one finds two types of vegetation at least. There are the more
northern types, such (in Ceylon) as Thalictrum or Heracleum,
which may be relics of a former more northern type of vegetation
in the plains, though they are more probablj' invaders by way
of the mountains; and there are the more numerous forms like
the Eugenias, the Impatiens, or the Memecylons, wliich are re-
lated to those growing at lower elevations.

While it is clear that many mountain endemics are relicts, and
probably many more are local adaptations, the former especially
within the range of the last glacial period, the evidence for relict
nature in the tropics and the southern sub-tropics is not suffi-
ciently clear to make it safe to regard any of them as such
without some direct evidence in favour of such a conclusion.

Others, again, maintain (25) that very many endemics are
waging a losing fight against more vigorous and adaptable new-
comers. This is no doubt the case with many woody endemics
in North America, etc. — genera which once were widely spread,
and are now left as representatives of a former woody flora in
a land of herbaceous vegetation. But to say that this latter is
more adaptable seems rather stretching a point. Were its
members turned into a forest the}^ would die out there much
sooner than the woody endemics seem likely to do as things
arc. The dying-out is owing to change of conditions, which has
been carefully guarded against in the statement of the rule of
Age and Area given above.

Lastly, it is maintained that in general (26) endemics are
rehcts in the more literal sense that they are remains of floras
that have disappeared elsewhere, in whole or in part, but are

94 OBJECTIONS TO THE HYPOTHESIS [pt. i

not necessarily dying out. This is a perfectly sound position, but
is not really an objection to Age and Area, when this is properly
understood. If a genus has 5 species in one region, and an
outlying species 6 in another, and one can produce geological
evidence of former connection, whether by living or by extinct
species, then there is no doubt that 6 is a relic in the sense of
this objection. One must simply take the whole area covered
by 1-6 as the area of the genus in consideration of any matter
by Age and Area. This type of relic, however, is really rather
uncommon.

A more frequent type is that so often found in temperate
North America, where the mountain chains, running north and
south, did not offer such a barrier to the ice and cold of the glacial
period as in the Old World. Sinnott (p. 86, footnote) instances
Carya and others, pointing out at the same time that many
occur as fossils in the Old World, and that they include most
of the woody endemics of north temperate America. Such en-
demics, showing wide taxonomic separation from the rest of
their surrounding forms, are, however, comparatively rare, and
as already pointed out, in dealing with them from an Age and
Area point of view, one must include the "fossil" area.

In the tropics, or in the southern hemisphere, on the other
hand, and even in the north among the herbs, which Sinnott
has shown to be in all probability very much younger than the
trees and woody plants, and which are probably mostly forms
that have spread there since the glacial period, the endemics
are usually closely related to the forms aromid them, Avhether
other endemics or "wides." It would be absurd to apply the
"relic" explanation to such a case as Doona in Ceylon (p. 153)
or Banunculus in New Zealand, and yet on this supposition
Banunculiis in that country, or at any rate Veronica, must be
considered as a relic, though the vegetation of north temperate
type represented by Ranunculus, Veronica, and many other
genera is a very marked feature in the total vegetation of New-
Zealand.

Another very serious reply to this objection is contained in
the fact that the endemics of a country remote from the effects
of the glacial period usually belong to the large and what have
usually been considered the "successful" genera, as has been
pointed out elsewhere (Chapter xv of Part II).

The next objection (27) is based upon the supposed rapid
spread of introductions, and is urged to show that dispersal

CH. IX] OBJECTIONS TO THE HYPOTHESIS 95

Avithin a country, when a species first arrives, is rapid, not slow.
But we have already seen that the evidence of introductions
(p. 24) forms a very broken reed upon which to lean. It only
shows that the spread may be rapid when the conditions have
been changed, and cannot be twisted into meaning that spread
is always rapid even in such circumstances. Even in Ceylon or
New Zealand, only a small proportion of the introductions have
spread rapidly, although the conditions have often been changed.
Nowhere is there any indication of a whole flora, or great part
of it, spreading in this rapid way, whereas in the case of an
island like Great Britain, near to a continent, the local flora is
simply a somewhat reduced edition of that of the continent, and
the flora of such an island as Ireland, a little farther out again,
is a reduced copy of that of Great Britain. One may even go
further, and find upon little islands off the coast of Ireland a
still further reduction of the Irish flora.

A careful consideration of what has been said in Chapters ii
and V will lead to the conclusion that in general the dispersal of
plants into new areas must be exceedingly slow, so slow that
as a general rule one will notice little or no progress in a lifetime
of observation. One cannot regard this objection as sound.

An objection often brought up is (28) that in many places
characterised by the presence of endemic forms there are many
genera composed of endemic sjjecies only. This very striking fact
has been termed "swamping" by Dr Sinnott, who proposes a
hypothesis to the effect that "the longer a successfully invading
species remains in an isolated area... the less common it tends to
become until it is actually 'swamped' out of existence — quite
the reverse of the ' age and area' idea." He suggests that " some
may simply be exterminated outright, and some by continual
crossing with new forms may ultimately lose their specific
identity." Cf. also Guppy (44, Chs. xxi-xxvii), who gives full
accounts of it.

On the whole, the older and more isolated the region, the
greater is the proportion of such genera. Ceylon has 89 of 1027
(apart from actuafly endemic genera); New Zealand has 127 of
♦329, the HaAvaiian Islands have 101 out of 256 genera.

There is no doubt that the fact that genera are common in
these floras with endemics only, and no wides, is a feature which
requires explanation; but as the genera with endemics only
behave exactly hke those which also contain wides, or like the
endemic genera, the fact that it cannot at the moment be satis-

96 OBJECTIONS TO THE HYPOTHESIS [pt. i

factorily explained^ does not in the least militate against the
hypothesis of Age and Area. Age and Area may seem to disagree
with other views as to this or that, but it is based upon very
clear and definite figures, which must either be controverted or
explained in some other way — they are far too striking to go
without any explanation. It is somewhat difficult to controvert
figures which simply represent bald facts, and if Age and Area
be not accepted, it is consequently necessary to have some other
hypothesis, which must be mechanical, owing to the fact that
the figures show such mechanical regularity.

The objection is based largel}^ on the lui doubted fact that the
proportion of "swamped" genera is larger in the more outlying
of the big islands. But that mere isolation is not sufficient as an
explanation would seem to show in the fact that in the very
isolated islands round New Zealand the proportion is not so
high as in New Zealand itself. In New Zealand 127 genera out
of 329 show it, in the Kermadecs only 8 out of 62, in the Chathams
the same, and in the Aucklands 12 out of 64. In none of the
islands is the proportion anything like so high as in New Zealand,
and it is highest in the Aucklands, which Avere perhaps nearest
to an incoming stream of plants (131). On the other hand, the
number of genera which are swamped in New Zealand is 13 in
the Kermadecs, 38 in the Chathams (the most isolated), and 26
in the Aucklands, facts tending to show that the swamped
genera were in existence fairl}^ early opposite to the Chathams,
and therefore were rather old in comparison to some of the rest,
though e^'en in the Chathams the imswamped genera (29) are
almost as ninnerous.

Another test that we may apply is to find the proportion of
*'swamped"genera in the northern and southern invasions of plants
into New Zealand (p. 79). The northern shows 45 out of 75 or 60
per cent., while the southern shows 36 out of 108 or 33 per cent.
We have seen that probability is in favour of the greater age in
New Zealand of the northern invasion, so that to some extent
this speaks in favour of the objection in a general and purely
local sense. But as only one herb {Elatostema) is "swamped" in
the noi'thern invasion, and all the shrubs but one {Veronica) in
the southern, it is, it seems to me, equally possible that swamping
may go with woody habit, and further tests are necessary.

1 Small (103, pp. 189, 224) has suggested two explanations, both quite
probable; but as the phenomenon (as shown above) does not affect the
probability that Age and Area is a correct hypothesis upon which to work,
the question may be left out of consideration in this place.

CH. IX] OBJECTIONS TO THE HYPOTHESIS 97

Of the "swamped" genera, only about half are herbs, while
of the unswamped, herbs are 83 per cent. Of the unswamped
genera with no endemics, 85 per cent, are herbs, while of those
with endemics 80 per cent, are herbs. From these figures it
would seem that the evidence is just as good for the connection
of swamping and woody nature as of swamping and age.

The Coniferae are probably older than the flowering plants,
and as they have no wides at all in New Zealand, this speaks in
favour of age, but they are also all woody plants. The Ferns, on
the other hand, which are probably older again, show very little
"swamping," only 5 generaoutof 31 exhibiting this phenomenon.
Of these it may be noted that three are the only tree-ferns in
New Zealand. The remaining two, and all the unswamped genera,
are herbaceous. It is evident that the question of swamping
must be disentangled from the question of the relatively greater
age of woody vegetation, but inasmuch as woody vegetation in
general is probably older than herbaceous, it seems probable
that swamping goes to some extent with age.

Actual measurements show that the average range in New
Zealand of one species of a swamped genus is 509 miles, which
within a very close approximation is the same range as that of
the whole flora of New Zealand, and considerably more than the
average range of the total of the species endemic to New Zealand,
or New Zealand and its outlying islands, which is only 446. On
the whole, therefore, one may probably saj' that these "swamped"
genera are older than the unswamped.

Further confirmation of this ^'iew may be obtained from the
fact that 45 of the swamped genera reach the outlying islands
round New Zealand, while only 27 of the unswamped do so,
though the latter are much more mmierous.

There is a possibility that Avith mere passage of time species
may undergo change, and it may be that "swamping" is some-
thing of this nature.

An important fact must be noticed in considering this objec-
tion, that the genera without "wides" behave just like those
that include such. They have (cf. the map of Gunnero in New
Zealand, p. 158) similar local distribution ; their centres of greatest
density are the same; their proportion of species belonging to
the different classes {i.e. the classes in order of area) is the same
when several genera are taken. If the endemic species of the
genera that possess wides are dying out before the competition

98 OBJECTIONS TO THE HYPOTHESIS [pt. i

of the latter, then the same thing is going on in the genera that
possess none, i.e. they are dying out without competition, and
at the same rate, a remarkable fact. If endemics are local species
developed in response to local conditions, then it is very remark-
able that in the genera where they have not been able to kill out
the wides, the latter should occupy the largest range (cf. map
of Ranunculus, p. 156, or almost any other genus of Ceylon or
New Zealand that possesses wides).

What the explanation of "swamping" may be is not as yet
clear, though it seems probable that it goes to some extent with
the mere age of a genus, especially if of woody habit. But its
existence does not in any way prejudice the vahdity of Age
and Area as an explanation of distribution, for the presence
or absence of wides makes no difference to the behaviour of
genera.

Another objection is (29) that much detailed work is being
done in splitting up large and wide-ranging Linnean species into
micro-species, and that this will destroy the value of my work,
as I have dealt only with Linnean species. This, translated into
terms of the figures which have been given in Chapters vi-viii,
means that species are being removed from the column of
"wides" into that of endemics, and perhaps \isually to near the
bottom of this. The result will not be to undermine my work,
but rather to strengthen it. As one of our leading ecologists says
in a letter to me. and underlines, "this will be strongly in favour
of your Age and Area hypothesis."

It is also objected (30) that species with wide distribution are
usually found in an early stage of the plant succession. This is
practically the same as the old axiom of the systematists "sim-
plicity of type goes with increase of area." Later species in a
country that is undergoing change of climate will tend to be
adapted to more strictly local conditions, and their spread will
therefore be hindered by ecological boundaries. But it is to
some extent a single-species objection.

The general objection, never perhaps expressed in so many
words, but running through a number of those actually given,
(31) that Age and Area does not agree with ecological results,
is largely answered in what has been said above. Age and Area
works over much longer periods than does local ecology, and
must not be applied to single species, and it must not be for-
gotten that it is not a mere unsupported hypothesis, with no

CH. IX] OBJECTIONS TO THE HYPOTHESIS 99

facts to back it. It rests upon a large number of very clear and
definite figures, which are so consonant with one another that
they must be explained; they cannot be passed by as unimport-
ant, an)' more than can those upon which Mendel's Law is based.
Further than this, Age and Area has been used as the basis for
numerous predictions, all of which have proved to be correct.
Unless, therefore, some other hypothesis can be found to explain
the facts, and make the predictions, and that a mechanical
hypothesis, on account of the mechanical regularity of the
figures, Age and Area must be regarded as holding the field for the
present.

Ecological factors svork at right angles to the age factor, to a
considerable extent, and on groups of allied species, taken over
a long time, their influence will then rarely be visible, as regards
total areas. The objections of the ecologists should, it seems to
me, largely disappear when they fully realise the meaning of
the careful provisos with which Age and Area is hedged about.
Not only are there those already considered (groups often species,
and allied species), but it is also pointed out that conditions
must remain reasonably constant. A serious change of con-
ditions is bound to make a great change in the dispersal rate of
the plants subject to it. If it only comes after the plant has
already spread into the neighbourhood affected by it, it will
probably make little difference, unless it reach the margin of
the area to which the plant has reached. Merely to exterminate
a plant in a portion of its range does not affect the total as
marked b}' the outlying stations.

Further than this, it is expressly stated that great modifica-
tions may be introduced by barriers, including ecological changes,
changes of climate, and the like. All these provisos, taken to-
gether, seem to me to make sufficient allowance for any possible
ecological influences, and the fact remains, as just stated, that
the figures, which are incontrovertible, go to shoAv the great,
and indeed overwhelming, effect of mere age, when working
with a group of allied species over a long period.

As has already been pointed out several times, age in itself
effects nothing, but the average result of the operation of
ecological and other factors is so uniform, when one works with
long periods, that the average rate of dispersal is also very
uniform. Barriers may of course completely stop it, but usually,
perhaps, only when they are physical, or due to such a cause as

7—2

100 OBJECTIONS TO THE HYPOTHESIS [pt.i,ch.ix

a very marked alteration of climate. Ordinary ecological barriers,
which most often, perhaps, are not very broad, will usually
only be able to check it. The check may be long-lasting, but
often the succession (pp. 51, 20) which usually occurs in plant
societies may give opportunity for passage. Further, by working
with groups of ten one allows for chance differences, and by
working with groups of allies one obtains groups upon which
all the various factors will probably operate in a more or less
uniform way, so that their rate of (total) dispersal will be more
or less uniform.

Finally, one Avriter does not like big changes; (32) "if the
camel can go through the eye of the needle, the gnat can follow."
In other words, presumably, if age can produce such effects, the
various later conclusions to which we shall presently proceed
will present little difficulty. But if large changes were not
sometimes made in our way of looking at things, progress
would be remarkably slow. Even if the new point of view is
not permanently adopted, it ^^^ll do no harm to spend a little
time there.

In conclusion, it may be noted that many of these objections
will perhaps cease to be urged in view of the interesting facts
to be brought up in the next few chapters, facts which will
quite possibly educe an entirely fresh set of objections.

PART II

THE APPLICATION OF AGE AND AREA

TO THE FLORA OF THE WORLD

AND ITS IM PLICA TIONS

CHAPTER X

THE POSITION OF THE AGE AND AREA THEORY
By H. B. GuppY, M.B., F.R.S.

V\ E would sometimes infer that there is only one way of reaard-
ing the central problem of Plant-Distributioji. If tjiis wer'e so
distribution would stand alone among the great studies of plant-
hfe, and It would be particularly unattractive and uninteresting
Generally speaking, the more numerous the standpoints the
more complete will be our grasp of the problem. The surveyor
who has the most accurate conceptions of the extent and out-
hnes of a great mountain range will be the man who has viewed
It from the greatest variety of stations, and so it will be with the
student of distribution.

The fewer limitations we impose upon ourselves at the start
the better progress shall we make. Some are ine\itable, but they
should be light easy burdens that do not gall. Thus when we find
ourselves constrained to associate our point of view with the
story of E^•olution, we are at once confronted with the query as
to the kind of evolution implied. What is the genetic sequence
in the scheme of the ordinal, tribal, generic, specific, and varietal
types? It is possible to hold views in this connection that are
as far asunder as the poles. In the case where we begin with the
larger groups we ha^ e evolution on a plane, or differentiation
pure nnd simple. The basic principle here involved, the change
from the Simple to the Complex, from the General to the Par-
ticular, from the Homogeneous to the Heterogeneous, is at the
back of the development of life on the earth. It is symbolised
m all natural systems of classification and in the daily practice
of the systematist, and was a part of the faith of the old
philosophers.

102 THE POSITION OF THE [pt. ii

On the other hand, to lay down, as the Darwinian evokitionist
does, that the order of development begins with the variety,
varieties diverging into species, species into genera, and genera
into families, is to reverse the method followed in nature, since
it imiDlies that the simpler, least mutable, and least adaptive
characters that distinguish the great families are the last de-
veloped. This could never have been. Nature has ever worked
from the simple to the complex, from the general to the par-
ticular. Had she followed the lines laid down by the Darwinian
school of evolutionists, there would be no systematic botany.
All would be confusion. There would be no distribution in the
sense in which the term is generally understood, and the plant
world would be a world of oddities and monstrosities. This is
the view expressed by the writer in his volume of West Indian
Observations published in 1917, p. 820 (47).

It is this incompatibility between theory and practice that has
given Dr Willis his opportunity. Under the glamour of Darwin's
great theory Distributionists lost touch with old basic principles,
and it is as an endeavour to establish the old connections, or as
an effort to return to the pre-Darwinian position, which we have
largely abandoned or forgotten, that the Age and Area hypo-
thesis will find its place. The vain attempts to bring together
ends that could never meet, and the failures to reconcile views
that were hopelessly apart, have all prepared the way for a re-
consideration of the central problem of Plant-Distribution.

Until we are in agreement about essentials we cannot utilise
the evolutionary standpoint for a general view of the subject.
The possible standpoints need much further exploration, and
several of the oldest have been forgotten. At any time a dis-
tributionist is liable to be held up by a query that in some quaint
old-time fashion will raise an issue that has been floating in
men's minds through the centuries. Distribution bristles with
the points made by the old philosophers, and many of our new
notions can there be matched. We cannot turn up any of the
old abandoned fields of research without unearthing some of
these old notions as fresh and as sound as in the days of their
entombment. But the query may belong more to our own time.
Thus one might be asked for the real significance of the fact that
Ave could found the Institutes of Botany on much the same
principles whether we based them on the flora of China or of
Peru. One of the implications of a recent paper by Dr Wilhs (135),
in which insular and continental floras are compared, is con-

CH. x] AGE AND AREA THEORY 103

cerned with precisely the same point. The question may be im-
answerable; but there are those who might see in a primeval
jumble of family types the backgroimd of the Avhole story of
Distribution. They might regard it as the most significant indica-
tion of the great antiquity of the higher plants, and they would
see in this world-wide mixture of family types the impress of the
lost Mesozoic ages on the history of the flowering plants, ages of
unceasing revolutionary changes in the relations of land and sea.
They would see in this world-spread mixture the materials on
which the great laws of de\'elopment ha\^e operated in the later
ages. Such would be their standpoint. But the problem may
prove to be one for the biometrician; and we may perhaps be
able to learn from him in the case of other world-spread mixtures
of organisms of different types the significance of the de\elop-
ment of uniform mixtures of types in Time.

There is another way of approaching the central problem of
Distribution, and that is best typified in the case of the gold-
miner who, guided at first by a faint show of colour in his pan,
follows the clue through until he finds the reef. This is pretty
much what Dr Willis has been doing for years in the working out
of his Age and Area theory. With a history of small beginnings in
Ceylon long ago, it is still in the making, and we can watch its
development. It is assimilating as it grows numbers of ideas that
have been floating in the minds of biologists for generations, and
linking together others that ha\e alwa3's been diflicult to place.
Its tendency to luiify and co-ordinate as it develops are two of
its striking features. The writer's attitude towards it may be
thus stated. Recognising that we had here a courageous and
persistent effort to utilise the statistical method in getting behind
the distribution of living plants, the question whether it was
wrong in this or wrong in that did not seem to be of primary
importance. For years the writer had been approaching the
subject of Distribution from the opposite direction, that is, from
the a priori side. Like many a general theory that had not been
linked up A\nth the other side the one that he advocated (a theory
of differentiation of generalised types) stood still for lack of
verification; and there were echoes in his memory of the despair-
ing counsels of those in this and other lands who regarded Dis-
tribution as beyond the pale of human endeavour. So that when
he realised the possibihties of far greater extension that lay
behind the Age and Area hypothesis, the question for him was
not whether Dr Willis was right or whether he was wrong, but

104 THE POSITION OF THE [pt. ii

where he was heading for. Here was a daring attempt to get a
grip at things from the inductive side, and the question was —
Which among the general theories will prove to be its goal?

But the prospects of the new theory at the outset were not
promising. Botanists had been incHned to regard the statistical
treatment of distribution as illusory, and the believers in what
Watson termed "Species-arithmetic" and Humboldt named
" Arithmeticae botanices " were few. Yet Hooker, with the seer's
outlook, took the true meaning of things three-quarters of a
century ago when he wrote:

All seem to dread the making Botanical Geography too exact
a science; they find it far easier to speculate than to employ the
inductive process. The first step to tracing the progress of the
creation of vegetation is to know the proportion in which the
groups appear in different localities, a relation which must be
expressed in numbers to be at all tangible (57, Vol. i. p. 438).

A generation later, when Hemsley at his suggestion took up the
preliminary statistical treatment of floras in the introduction to
his great work on the botany of Central America (51), Hooker
characterised the subject as "that most instructive branch of
phj^togeography."' The lode was rich in promise, but he passed
it by. How was this?

It is clear from his lecture on Insular Floras (142), and from
different letters written in the sixties, that the Natural Selection
theory offered to him "the most hopeful future" for an advance
on the problems of plant-distribution from the inductive side.
In that lecture he also shadowed out a general notion of " Cen-
trifugal Variation operating through countless ages." It appears
almost as a suggestion, but the idea had been evidently floating
half-formed in his mind ever since he wrote his essay on the
Tasmanian flora in the late fifties. It was the nucleus of a theory
of Divergence or Differentiation that acquired more definite out-
lines as time went on, since it reappears in the intensely interest-
ing account of a talk with Darwin which is given in a letter to
Huxley in 1888 (57, ii. p. 306).

We can perhaps imderstand the long intervals of time now.
For the confirmation that such a theory would have derived
from a line of research instituted on Darwin's lines was denied
to him. The two proved to be incompatible. For no inductive
process based on Darwin's lines could have found its goal in a
theory of centrifugal variation. "1 well remember," Hooker
describes in his letter to Huxley in 1888, "the worry which that

CH. x] AGE AND AREA THEORY 105

tendency to divergence caused him (DarAvin). I believe I first
pointed the defect out to him, at least I insisted from the first
on his entertaining a crude idea which held that variation was
a centrifugal force, whether it resulted in species or not." Huxley
was in the same case. For he held views of the general differentia-
tion of types, and his road that would lead to the discovery of
the causes of evolution started from the Darwinian position.
That road was barred to him.

The secret of the success of Dr >Yillis is that he works with
limited objectives and is always free to shape his course accord-
ing to his results. A distant objective with a specified general
theory of distribution as his goal might easily have brought him
to the ground. As it is, he has struck a wonderful trail that
seems to increase in promise as he advances. But the logical
outcome of establishing his theorj^ successively for the species,
the genus, the tribe, and the family, is a general theory of
differentiation. In other words, it will bring him to the pre-
Darwinian position. Once there, he will enjoy greater freedom
in his choice of routes and methods, and new and unexpected
fields of research will be opened up all around him. This note
may be concluded with a brief reference to a few of the more
remarkable features of a theory that is still in the making.

Though the linking up of old ideas that have been without a
resting place for generations is mainly incidental, it is none the
less significant. I gather from Dr ^Yillis that his -'alliterative
series," as he terms it, which began -with "Age and Area," is
increasing in its numbers as his work proceeds. Thus we have
Antiquity and Amplitude, Rank and Range, Size and Space,
and several others, some of them OA'erlapping, but each with
its own variant, and some again capable of considerable exten-
sion and amplification. Thus Rank and Range implies Simphcity
of Type and Increase of Area, a very old principle long recog-
nised in the theory and practice of pre-Darwinian systematists.
Simplicity of Type goes with Variability, another old principle.
If, therefore, the simplest organisms of a group are the widest
distributed and the most variable (ideas old enough and true
enough) it is among them that we ought to look for examples
of genera that have arisen independently in different parts of
their areas, as in the ease of Senecio, the most primitive form of
the Compositae. Incidental as such results may be, Dr Willis
may well claim that his materials are working for him. \Vhilst
he is following a definite plan, much is happening that was

106 POSITION OF AGE AND AREA THEORY [pt. ii, ch. x

neither premeditated nor foreseen. Just as a river wearing its
way into a mountain-jnass unites in a single system widely
separated streams by capturing one water-head after another,
so the Age and Area theory in its advance is bringing about the
coalescence of principles that we have been wont to consider as
things apart.

This may be the luck of the trail. But at all events we have
to distinguish between the direct and indirect results, and one
scarcely knows which will prove to be the most important out-
come of this investigation. It is difficult to speak of work still
on the stocks, but we will expect to find in the results of the
tabulation of the genera of the flowering plants a survey of the
distribution of some 12,000 genera over the great regions of the
globe, Endemism will figure more as a world-affair than as a
peculiarity of localities, and some unexpected results are to be
looked for in a treatment of endemism in the mass. Then there
will be the story of the monotypic genera that appropriate
almost two-fifths of the total of the genera of the flowering plants ;
and their part in the forming of the curve of all the genera
grouped by the number of their species will prove to be a
triumph for the mutationists, A closing word may be said of
the great labour involved in the preparation of the tabulated
results, of the weeks of counting to establish a single point, and
of the wearisome recovering of the ground to make some doubtful
point assured. Since it was the purpose of the writer to place
rather than describe the Age and Area theory more cannot be
said here.

CHAPTER XI

THE FURTHER EXTENSION OF THE
APPLICATION OF AGE AND AREA

In most of the work so far published, and in the first part of
this book, Age and Area is used only within narrow limits, as
applying to the flora of a single given country. But this is a
purely arbitrary limitation, and was adopted in order to render
less complex its application to the problems of distribution;
and in this second part of the book Age and Area will be
applied to genera as well as to species, and to the flora of the
world as a whole.

Like Age and Area itself, its twin principle, to which I give
the name Size and Space, has also been used as yet in a limited
way, e.g. on p. 71, where it is pointed out that genera that are
represented in a country by several species are likely to be (on
the average) older in that country than genera that are only
represented there by one. The exact graduation of commonness
with number of species which is there shown indicated that this
principle was also capable of extension, and it is expanded in
Chapter xri into the more general proposition that within any
circle of afiinity, the larger genera will be the older, and when
taken in groups of ten allied genera will be older in rough pro-
portion to their numbers of species.

This supposition is very strikingly confirmed by an examina-
tion of the British flora, which shows that the distribution in
Britain of the most widely distributed species of each genus (on
the average of the whole number) varies with the number of
species that the genus possesses in Britain. The same is the case
with the second, third, foiu'th, and so on to tenth, most widely
distributed species in each genus. Extension of the principle to
the whole world is then illustrated by aid of the Helobieae, by
refcrcnci^ to Prof. Small's work on the Compositae (in the next
chapter), and also to many other cases given below. The general
result, therefore, is to show that Age, Size, and Space (or
Area) go together.

In the next chapter Prof. Small shows how Age and Area can
be applied Avith effect to the distribution of a single family, by
dealing with the Compositae. The average generic area is deter-

108 THE FURTHER EXTENSION OF THE [pt. u

mined for each group of the Compositae, and it is shown that on
the Avhole it increases with the increasing age of the group as
deduced from phyletic, morphological, and geological con-
siderations. This agreement forms a strong argument both for
the general correctness of Age and Area, and for that of the
previously deduced genetic relationships of the different groups
of Compositae.

In the second part of the chapter Prof. Small takes up the
application of Size and SjDace, showing that it holds very well
indeed as a general rule in this family, so that here, as in other
cases, "both the average generic area and the average number
of species per genus are closely related to absolute age." Age,
Size, and Space go together.

Mrs Reid then takes up the application of Age and Area to
the fossil botany of comparatively recent times, especially the
Pliocene and Pleistocene. She shows how great have been the
migrations to and fro, north and south, of the floras of the
north temperate zone, and discusses the applicabilit}^ of this
proved migration to the flora of New Zealand, leaving the
question finally open for settlement by geological evidence. Dis-
cussing then the flora that at one time occupied the complete
circle of the north temperate zone, and which is now confined
to North America or to China, or to both, and often a good deal
broken in distribution, she shows that the existing dispersal
may probably be attributed mainly to the effects of the Glacial
period.

It is then pointed out that it is this unquestionable fact that
a good many existing strictly localised or endemic species are
survivors of races that once flourished widely, that offers the
greatest stumbling block to the acceptance of Age and Area, but
that there is no insuperable difficulty in the acceptance both of
this fact and of Age and Area, for the latter is reasoning from the
mass, the former from the individual, and while perhaps 1 per
cent, of the grand total of endemic species are relics, the rest are
not, and in reasoning about the mass the former are quite lost.
There is good evidence to the effect that many or most of these
survivals are due to the effect oi' the Glacial period, and on the
whole, therefore, the verdict is in favour of Age and Area.

Endemism and Distribution of Species are then considered in
Chapter xv, and it is shown that the phenomena presented by
endemic species in their distribution are simply a miniature of
those presented by species in general, and that the distribution

CH. XI] APPLICATION OF AGE AND AREA 109

of both can be graphically represented by " hollow curves^," like
those in the fig. on p. 155 (and cf. clearer figure on p. 174), with
very many species occurring upon very small areas, the numbers
rapidly diminishing towards the areas of moderate size and then
more slowly to those of large size.

In view of these and many other facts brought up, and of
which a simimary is given upon p. 159, it is no longer possible,
except in comparatively rare cases, to regard endemic species
either as relics or as special local adaptations ; though of course
if not adapted to the local conditions as they existed at the time
of their birth, they would be promptly killed out by natural
selection. The explanation offered by Age and Area, that species
of very small area of dispersal are in general young beginners,
and that area occupied increases with age, seems the only pos-
sible one for the great majority of species. Not only so, but age
proves to be by far the most important factor in the dispersal.

In Chapter xvi Endemism and Distribution of Genera are
dealt with, and it is shown that the phenomena presented are
exactly parallel to those exhibited by species, and that the dis-
tribution of endemic genera is similarly a miniature of that pre-
sented by genera as a whole. The areas occupied by the genera
of a given family are arranged like those occupied by the species
of a given genus. There are very many upon comparatively
small areas, and many on the areas just a little larger, whilst
there are but few upon areas that are really large. As one would
expect from a consideration of the hyjDothesis of Size and Space,
one finds that the sizes of the genera themselves (in number of
their species) go mainly with the area occupied, so long as one
keeps to the allied forms of a single family. The bulk of the
genera of very small area are monotypic, or have but one species
each, Avhile the bulk of those of very large area have very many
species (average 59), those with intermediate size of area having
intermediate numbers of species. Plotting of the genera, whether
by size or by area, thus gives hollow curves. While the latter
represents their geographical distribution, the former obviously
represents their evolution.

The same hollow curve type of distribution shows ilsclf if one

^ The "hollow curve" arises when numbers are plotted as a graphic curve
which are large for the first two or three cases (e.g. in tlie fig. on p. 174 the
first three are -10, 15, and 8, or much more than half the total of 100), and
then taper away gradually in a tail (e.g. the remaining 37 are divided among
the groups of families from the 4th to the 29th). There is a large drop from
the first to the second, and from the second to the third or fourth.

110 THE FURTHER EXTENSION OF THE [pt. ir

sort into sizes the genera confined to any section of the world,
whether it be an individual island, or a larger area of territory
like Africa or South America, or whether it be the entire world
itself. Always there are many monotypes with a rapid drop
through the ditypes and tritypes, and a longer or shorter tail of
larger genera.

The supposition that endemic genera are usually rehcs, as well
as the other that they are usually local adaptations, must be
ruled out of consideration in view of the facts brought up, and
the only supposition that at present seems at all feasible is that
provided by Age and Area, that in general they are young
beginners. This is also shown by the fact that the proportions
upon islands in the different families are not unlike the pro-
portionate sizes of these families in the world.

Passing on to Monotypic Genera in Chapter xvii, it is shown
that these, which are usually much localised, display the same
phenomena. They are very numerous, over 38 per cent, of the
genera of the world containing only one species each, while there
are about 13 per cent, of ditypes, these two therefore containing
more than half the genera in the world. The proportion of mono-
types falls off with increasing size of area, and the proportions
oKgenera of other sizes bear a definite relation to that of mono-
types, showing that to explain these in general as reUcs or as
special adaptations would be absurd. They must usually be
voung beginners.

Not only do these numbers, when plotted, exhibit a beautiful
hollow cun-e for the distribution into sizes of the genera of the
world, but the same thing is shown by every individual family.
Other arithmetical relationships between the monotypes and
other genera, depending upon the size of the area considered,
are also pointed out.

Chapter xviii deals with the Hollow Curve of Distribution
and shows, by summing up what has already been said, how
universal this type of curve is, not only in the distribution of
species and genera (endemic or not) by area— Geographical Dis-
tribution or Distribution in Space— but in the distribution of
genera into groups according to their number of species— Evo-
lution or Distribution in Time. It is clearly evident throughout,
and usually in a very marked and unmistakable way, and goes
to show that Evolution and Geographical Distribution have gone
on "mechanically." The former appears to have been organised
at the start upon a definite plan, and its further unfolding, and

CH. XI] APPLICATION OF AGE AND AREA m

the distribution of species about the globe, have been chiefly
determined by age, when one is deahng with the mass of species
the various other causes that may be operative— cUmatic eco-
logical, geographical, geological, etc.-simply causing deviations
to one side or the other, but not permanently divertina the
dominant plan. Age and Area obviously, therefore, becomes a
corollary of the larger law.

But if this be so universal a rule in plants, it is obvious that it
must probably show in animals also, and Chapter xix shows
that this IS actually the case, and that it is exhibited as clearly
in the animal kingdom as in the vegetable.

The question of Origin of Species is 'then touched upon
(Chapter xx), and it is shown that probability is much in favour
of mutation as against infinitesimal variation, and that the
effect of the recent work upon distribution and evolution de-
scribed in this book is to make extremely probable the con-
tention that I have frequently put forward, and which is now
accepted by Prof, de Vries, that mutations may at times occur
of the necessary "size" to give rise at once to Linnean species
If one such mutation sur^•ived in fifty years, the whole existing
population of flowering plants could be evolved in eight million
years, which is perhaps less than 25 per cent, of the time that
has actually been available for, and occupied in, their evolution.
If c^'olution be a predetermined result, then it is clear that
advantage as guiding it is ruled out of acceptance, and it is
difficult to see, upon this ground alone (though there is strong
evidence upon other grounds), how anything but direct mutation
giving Linnean species can be effective.

In the following chapter (xxi) Prof, de Vries deals with the
relations of Age and Area to the Mutation theory, first ])ointing
out the essential diffe-ence between this and the theory of
infinitesimal variation. In the latter there is no change in the
genes, or material bearers of characters, but merely a fluctuation
or oscillation of the emphasis of the characters about a mean
value, so that in one member of a group of plants of common
descent a character may be large, in another small, and so on.
In the theory of Mutation, the changes have invoh'ed the genes,
the alterations in these resulting in permanent and usually
hereditary differences in the organism.

Prof, de Vries then points out that while Darwin recognised
that both mutation and fluctuation might result in new species,
the material of facts at hand was insufficient for any kind of

112 FURTHER EXTENSION [pt. ii, ch. xr

definite proof, and he decided in favour of the latter. The theory
of natural selection of infinitesimal variations has, however, met
with great and increasing difficulties in explaining the general
occurrence of useless characters, or the manner in which natural
selection can take hold of the first beginnings of a change. It is
now generally recognised that the bulk of the morphological
characters by which the systematic arrangement of plants into
related groups is carried out have no physiological value to the
plant at all.

At this point Age and Area comes in, showing that the dis-
persal of species is largely independent of their distinctive
morphological characters, for even in the youngest of them
(those most limited in area) no relation can be pointed out be-
tween these things, and yet the conditions under Avhich these
very confined species are living must approximate at any rate
to those under which they began. One must therefore conclude
that specific characters have evolved without any relation to
their possible significance in the struggle for existence. Area
occupied depends mainly upon age, and not upon morphological
characters (of course there are many exceptions); species spread
where they find suitable conditions, and the adaptation is not
on their side, but in the long run they choose the best environ-
ment. Prof, de Vries regards this as being the great proof which
the mutation theory still wanted for its complete acceptance.

Finally, a brief chapter (xxii), Avhich does not lend itself to
a summary in ad^•ance, is given to shoAv the general bearings of
the subject-matter of the book upon the study of distribution.
Age and Area, and Size and Space, are both so valid, and can
be so successfully used to make predictions about geographical
distribution, and these predictions are so near to accuracy, that
it is clear that in general distribution has been mainly goA'erned,
positively by age, negatively by barriers (of course including
ecological barriers). This being so, it seems probable that a very
promising line of work for the present may be the study of
invasions of plants, of course taken in connection with ecological
investigation into the formation (or disappearance) of barriers.
Age, and geographical proximity, again, will have to be taken
into more serious account in dealing with taxonomic questions,
and there are other directions in which the changes in our methods
of viewing problems of distribution that seem necessary may
produce considerable effects.

CHAPTER XII

SIZE AND SPACE

We have already pointed out, on p. 71, that on the average
the larger famihes and genera in a country will probably be the
older there, inasmuch as it is highly improbable that the sinale
species of a genus represented only by one would always arrive
as soon as the first species of a genus represented by many The
tendency will be for the latter to arrive first, and if, as A^e and
Area mdicates, there is but little killing out of species once
established, one will expect that the first arrivals will have spread
the most. It is obvious, of course, that one must work with
averages of considerable numbers to ol^tain reliable results but
It seems to me that this extension, for which I propose the name
Size and Space, may be given to the original idea of Age and
Area. Under this supposition one will sav that on the whole
keeping to the same circle of affinity, the larger families and
genera mW be the older, and will therefore occupy the most
space. This, however, involves a break with the long current
idea, that the larger families and genera are the successful ones
the smaller the (comparative) failures.

This principle obviously follows, once the central principle of
Age and Area is recognised, and it is further realised that destruc-
tion of species by natural selection takes place ^vhen tliey are
newly born and occupy minute areas of ground, and not when
they are once established on a reasonable area. Destruction
then, so far as we can see, will rarely happen, except in the case
of some great change of conditions, such as the secular drying
of climate, which (among other things) is apparently responsible
for the fact that Cupressus macrocarpa, etc., are now apparently
dying out (or rather not expanding) in California.

One may get very good evidence in fa\'our of this view by
applying it to such a flora as that of Britain, for which there are
good statistics of distribution available. If we take the distribu-
tion of the plants by the number of Watson's "vice-counties"
that they reach (37) we get the table on p. 114.

The diminution of the numbers in every line from left to right
of course means nothing, for the species are taken in order from
first to fifth most widely dispersed. But all the columns also

114 SIZE AND SPACE [pt. ii

Table showing {in the horizontal lines) the average rmmber of vice-
counties in Britain reached by the most ividely distributed species
in each genus of different sizes, and by the second, third, fourth,
and fifth, most widely distributed species in each genus.

Av

erage number of vice-counties

reached by the

1st

2nd

.3rd

4th

5th

Genus of

species

species

species

species

species

Over 10 species

108

104

96

86

79

6 to 10

103

84

64

1 49

33

5 species

98

76

39

22

16

4

89

01

35

13

3

89

48

27

2

78

r,'A

1

50

show a steady diminution from top to bottom, whether the first,
second, third^ fourth, or fifth species be taken ; and examination
of the remaining figures shows that the rule holds equally well
for the sixth, seventh, eighth, ninth, and tenth most widely
distributed species. The most widely dispersed species of a large
genus {i.e. a genus with many species in Britain) is (on the a\erage)
more widely dispersed than that of a genus with 6-10 species,
this than that of a genus with five, and so on right down the
scale, and the same thing shows with the second, third, fourth,
and fifth to tenth most widely distributed species. Nothing but
a mechanical explanation can explain such mechanical regu-
larity. If the vital, climatic, or ecological factors had many
differences, other than purely local, in their action, one would
expect some breaks in the regularity, but there are none. The
genera occupy areas in Britain in proportion to their numbers
of species there, and age has ])een the overwhelming factor in
their distribution.

As the species of those genera with one species each average
50 vice-counties, and those with two 73 and 33, one may imagine
that on the average one species in the latter genera arrived before
the solitary one of the former. In the same way (as indicated by
the vertical lines in the table) two species in the genera with
four or five, three in those with 6-10, and at least five in the
larger genera, probably did so.

Such results as this, which could be easily multiplied, go to
show that in a given country the area occupied by a genus
increases (on the average of considerable numbers) with the

CH. XII] SIZE AND SPACE 115

number of species representing that genus in that country, or,
in other words, that the principle of Size and Space is vaHd.
Very httle consideration, however, is required to show that in
general a genus of many species occupies a larger area than an
allied genus of few species. It is not perhaps always realised
how close the agreement really is, when one considers a number
of allied genera (as with Age and Area) between the size of a
genus (as marked by the number of its species) and space occu-
pied. Everyone knows that Senecio or Astragalus, with 1500 or
more species, occupies an enormous area, whilst monotypic
genera like Fatsia (Japan), or Welwitschia (south-west Africa),
or, again, like lonopsidium (Portugal) or Kitaihelia (Lower
Danube) occupy small ones, and genera with intermediate
numbers of species often occupy areas between these extremes.
But, on the other hand, people point^ to such a genus as Hijo-
puris, with one nearly cosmopolitan species, or Veronica, with
about 80 species in New Zealand, and maintain that there is no
connection between size and space. Now there is no doubt that
these exceptions to the rule are very numerous and very im-
portant, so that it would be in the highest degree dangerous to
draw a rule with limits as narrow as those for Age and Area
(ten allied species); but we are, nevertheless, of the opinion that
such a rule may be drawn, in such a form, say, as " Within any
circle of plants of near affinitv, living under similar ecological
conditions, the areas occupied, taking the genera in groups of
ten, will vary with the number of species in the genus, being
large when that is large." It is to be noted that proportionate
areas are not claimed; one would probably have to deal with
the genera by hundreds rather than tens for this.

The number of species in a genus seems to bear a distinct
general relation to the variety of conditions that exists in its
range: for example, water plants in general ha\-e much fewer
species than land plants that cover the same area. It is clear,
however, that this is not a complete explanation, for Veronica
in the compa^ati^^ely uniform conditions of New Zealand, or
Eugenia or Sirohilanthes in those of Ceylon, is represented by
iiKiu}^ species, while some species arc able to stand a variety of
conditions, such, for example, as Cissampelos Pareira or Senecio
vulgaris. On the whole, however, greater variety of conditions

^ "There is no necessary relation between the area a genus covers, and
the number of species it contains, though speaking generall}% monotypes
have a restricted area" (51, p. xxx).

116 SIZE AND SPACE [pt. ii

means greater variety of forms, and as to obtain that greater
variety of conditions means in general larger areas, size of a
genus and space occupied go largely together.

A good proof for the general correctness of Size and Space is
that, as we shall see in more detail below, the further out we go
among the islands, the larger on the average do the genera
become (in the number of species they contain in the world).
Whilst the world average for a genus is 12-13 species, the non-
endemic genera found in India contain on the average about
50 species in the world, in New Zealand about 75, and in the
Hawaiian Islands about 100.

Prof. Small (see beloAv, Chapter xiii) has worked out the hypo-
thesis of Size and Space with reference to the Compositae, and
his results form a remarkable verification of its correctness in
broad outline, and consequently a further proof that however
much the distribution of an individual form may be subject to
the many and various factors already mentioned, on the average
of large numbers the results go very largely in accordance with
the laws of probability, so that the distribution, under the steady
pull of age, is, on the large scale, much more mechanical than
we had previously been inclined to suppose.

If one take again such a group as the order Helobieae (7
families) which are chiefly water or marsh plants, and closely
related, one finds:

-1 cosmopolitan genera, witli ... ... 138 species; average 34

12 genera occupying large areas in the tropics,

with .". 83 „ „ r

2 genera, temperate and subtropical regions 7 ,, ,, 3-5

26 genera of small area ... ... ... 5.5 ,, ,, 2

showing very clearly how size goes with space. And yet it is
quite possible here as usual to pick out genera that go in the
reverse direction; e.g. Zannichellia with one species is cosmo-
politan, while Philotria with five is confined to North America.

On the whole, therefore, the principle we have laid down may
be seen to be justified by the facts when large numbers are dealt
with. But this is a recognised necessity of all statistical work^
as, for instance, in working out results under Mendel's Law.

Now, taking this principle together with Age and Area, it
is clear that Age and Size, or Antiquity and Amplitude, if an
alliterative title be preferred, go together, and on the whole the
larger a genus, the older will it be, within its ozvn circle of affinity.
No one would suggest that a herbaceous genus of 100 species was-

OH. xii] SIZE AND SPACE 117

of the same age as a tree genus with 100, but both will follow
this principle as far as possible. It goes to show that on the whole,
as the area occupied increases, a genus tends to break up into
more and more species: only at times does the original species
of the genus cover the whole of its range when it has reached a
very large area, and then most often when the conditions are
very uniform, as in the case of Zannichellia for example. In the
case of the Podostemaceae, where the conditions are perhaps
even more uniform, and yet a great many species have arisen,
it is due, as I shall hope to show in a later publication, to the
fact that the plants are always under the influence of plagio-
tropism, to the greatest extent possible.

If we take the 28 largest genera in the world (51), we find that
about 16 are cosmopolitan in their distribution, 5 are cosmo-
tropical, 4 tropical America, and Qiiercus Old World, leaving
only Erica and Mesembryanthemutn, whose large number of
species is correlated in both cases with the fact that they grow
in South Africa, where the extreme conditions seem to tend to
produce large numbers of species, though, as we shall hope to
show in later publications, there are other factors in the matter.

Nearly half the species in the world (69,000 of 162,000) belong
to 1171 genera that occur in both worlds (average 59 species per
genus), while only 66,750 belong to 9671 genera that are con-
fined to a single continent (average 7), and the 2026 genera of
the northern palaeotemperate and the palaeotropical regions,
etc. {i.e. widely distributed in the Old World) have about 26,250
species (numbers from my Dictionary), and form, as one would
expect upon the hypothesis of Size and Space, an intermediate
between the other two groups (average 13).

Of the 28 large genera named above, the British Isles contain
10, Ceylon 17, New Zealand 11, the Hawaiian Islands 14 and
the Galapagos 15. Solanum (1225 species). Euphorbia (750), and
Cyperus (400) occur on all five, and four others on four, of these
groups, the only ones that occur on none being Myrcia and Mam-
millaria. Of the 244 genera that contain o^•er 100 species, no
fewer than 166 occur in both Old and New Worlds, 28 in tropical
America, and 10 in the Old World tropics, leaving only 31 for
the remaining smaller divisions of tiie world, like tropical Asia,
which has only 7.

In the same way, the smaller families usually occupy smaller
areas than the larger, and the question arises whether they
should be considered of equal rank to the latter. Guppy has

118 SIZE AND SPACE [pt. ii, ch. xii

suggested a grouping of families into classes based upon these
principles, for which he has suggested the title Rank and Range,
and it is clear that in all future systematic work, the question
of area must occupy some attention.

Excellent examples of the application of the principle of Size
and Space may be found below, e.g. on pp. 132, 164, 165,
171-2, 174, 178, 187-8, 190, and 197.

Summary

If species spread in a country mainly in accordance with their
age, then it is clear that on the average some of those in the
genera represented by most species will have arrived before the
first of those in the genera represented by few. This principle
may be extended, and under the name Size and Space may be
thus expressed; on the whole, keeping to the same circle of
affinity, a group of large genera will occupy more space than a
group of small. The space occupied will A'ary more or less with
the number of species.

Illustrations of the operation of this principle have already
been given in Chapter vii, and further examples are drawn from
the Helobieae, and from the flora of Britain, while a good instance
is also given by Prof. Small in the next chapter. IMany other
instances can be found, too, in later chapters.

CHAPTER XIII

AGE AND AREA, AND SIZE AND SPACE,

IN THE COMPOSITAE

By James Small, D.Sc, F.L.S.

Age and Area. In a previous contribution to the study of the
geographical distribution of the Compositae (103) many of the
conchisions were based upon the Age and Area hypothesis as
far as the phenomena could be determined roughly by simple
inspection of a series of maps which included all the genera. It
Avas mentioned (103, p. 190) that although this hypothesis was
still restricted to "age within a given country, its proved exten-
sion to absolute age and total area seems to be only a question
of time and application." This extension of the original hypo-
thesis, which Avas suggested in 1916 by the writer (103, p. 208),
has now been adopted by Dr J. C. Willis, and the present con-
tribution consists of a critical analysis of the statistics for Age
and Area in the Compositae in the light both of that extension
and of previous phyletic conclusions. These previous suggestions
were simimarised as "the basis of future discussions" (103,
p. 313) in a family tree which is reproduced upon p. 125. The
statistical data are given in Table I, and were obtained by the
following methods. (Table I. pp. 120-124.)

In order to avoid the unbalanced effects of the inclusion of
new genera which have been discovered or resuscitated fre-
quenth'^ as the result of special studies of only one or a few
tribes, the data have been prepared only for the genera included
by Bentham in the Genera Plantarum. The area covered by each
genus has been determined approximately in millions of square
miles. For this purpose Mikania and Eujjatoriiitn have again
(cf. 103, pp. 133 and 204) been taken as one genus, and so have
Aster and Erigeron as two genera which are "so very closely
allied that the transitional species are comparatively numerous
and the genera in these cases are distinguished only by the so-
called indefinable characters of the taxonomist" (103, p. 307).
All genera occupying less than 1,000,000 square miles have been
included in Class 1 : while 59 other classes have been taken for
the other genera, the total area of the land surface of the world
being approximatelj^ 60 million square miles. This method is.

2
H I

.a

li

r

-5

saqii^
-qns

1 -^ ^ii >i2 o.^oo i2.<5 1 1 -£1,00 .tloo -in rr>-£l m i^o

saqux

-^■^ 1 1 1 1 1 -2"^ 1 1 1 1 1 1

PFOAV PIO 3q? IIY

uoiSdy; uBi[Ej^snv

f;Tf;ri f.Ti i f^i fsfg^i f^i

uoiSaa

f^||^.|:ri f-|.i f^f?-4«>

uoiSaa:
oiimsy p;oido.ix

1^1^' l^' ' "i^^' 3^ ^^"^ ^ ^'^"^ *^ '

uoigaH

uoi§e>j
oi?t!isY'-08Edo,ing

uoiSaji
UBauEuairpaK

TJDuauiv iiv

uoiSajj unjiiq^

CO CX) CO o >-, o- o ^ '~'

uoiSa^j auipuy

|?I^| i?ft^-|^i l^^^i 1 1

saipui ■)sa,\\

l^t^! 1^1^' 1 f-l 1-1^=^1 1 1

uoiSajj
sa'j'B'jg pa^iuq

t^fJC-f^l^l 1 1^1 f^'lsf^l 1

uoiSaa UEoixaj^

|f|^-.t^f.54?i f^f^f^i 1

1
1

^ li ^ 1 1 i 1 , , i 1

S -2 JS " S S 1 « ^ -S .= ^ 2

1 t: H -^ o f 1 K s E a ^

5 rn' CO (M" -«■ JB S r-i ei CC TJ<' L-;

^ rx^ fQ Ji ■sh^ >n-!i po^ t^ 1 -iCI o-li^ i: >o^ t^-2; txi; w-,ii t^_^ <>) -2=o ^o q^ « o <^ t-j

00

1 1 1 1 1 1 1^1 1 1 1 1 1 1 1 1 1 1 1

00

1 1^1 ^^f^l 1 r^^^t^l 1 1 1 1 1 1 1

1 1 III?, 1 1 |||?l f=, , , 1 1?, , 1

1 1 fs-s'^l 1 1 iS^SI 1 1 1 1 |-"l i 1 1

1 1 l^l^i^H 1 ^R|SI 1 1 1 1 1 |?l 1 1

1 1 1 1 1 1 l^|?|s|^i l^slallf^s-s?

1 1 1 1 1 1 1 f sf -!&£=" i tss-t'-"i?i 1 ^^

1 1 1 1 1 1 1 l^^fsf^tj f^i fs|?i s-i

,,,,,, 1 r4?|:r.i 111 ftf?f?5^^^

1 1 1 1 1 1 M44-"' "2i tt|:?i s-i

1 1 1 1 I 1 |g|gfs|^tll fSf-fsfK?! 1

1 1 1 1 1 1 |«"tr|*tsi fif 'fsfii?,*- 1

6. Relhaniinae

7. Angianthinae

1. Inulinae

2. Buphthalminae ...

3. Athiixiinae

4. (Relhaniinae)
Heliantheae

1. Verbesininae

2. Coreop-sidinae

1+2

3. Melampodiinae ...

1+2+3

4. Galinsoginae

5. Madiinae

6'. Milleriinae

6". Ambrosiinae ...
G'". Zinniinae

7. Lagasceinae

8. Petrobiinae

Is

r

>•

«i5

saqui
-qng

"^"^oo „ Sv"-. «o ;3;o

saqijx

1^1 1 I 1 1 1 |"l i 1^1 1 1 1 1

PFO.U PIO aqi IIV

<~-> OO M OO r^ 0 ■ O

ucnSay^ UBjiiBJIsny

uoiga^
ucoujy qinog

uoiSajj

1 1 I5^"i^'§^ 1 f =^ 1 1^1 "i?f?! 1 1 1

uoiSajj

1 1 I'l^t- 1 "?^ i t-t"llfl 1 1 1 i

uoiSay^
op^iSY-oacdoong

uoiSajj
ut3auBjjoiipai^

1 ^Ov-^^^r^.-^^l -^--M ^foS^N-C'M^^^-OOM^^oii,^

Eoijaiuy nv

COM!-. ^40,^ r-^

-2: S -Ti -1--^ t^Ji u-,-iC' T^-^C ^"^-^T'-^.'-^i . '^Cc ^00 u-,>0 1 -^ t^ ^.VO

uoiSajj ii'BjiiqQ

' l-l{'i,--S-t5 1 1 fjl tJt^^-S-l IB^-

uo]§a>x UTJijiztja

1 -|o^--|^-|i:;'-$>|^ 1 1 >-^l -§^1 1 1 t^ 1

uoTgajj ouipuy

1 |;lt^|^-^-f--S-H fSl fll^l 1 t-i

saipui isaAi

1 l?t-"i--^"t" ' ' 1 1 t"t"l 1 1 1

uoi§a}j

1 |?|.-| l^t^i 1 ||-|o|.|^^^, 1^^^^

uoiSa^i uBDixaj^T

1 M^l^lot^l 1 f|^?p|--l 1--^

J:

Astereae

1. Homochrominac..

2. Heterochrominae

3. Conyzinae

4. Bellidinae

5. Baccharidinae ..

6. Grangeinae
Anth'cnudeae...

1. Chrysanthemidina(

2. Antheniidinae
Cichorieae

1. Lactucinac

3. Scorzonerinae . .

4. Scolyminae

2. Allothersub-tribcs

Hyoseridinae ..

■^ r^^ f^-::j rr>-^ .^ N O -^ 01

I I I I i §;

; 1^ r»^ o tM o >o

I I I 1^1^ I

I I i ^51 I I

-£1. "P^ f^-ii V^-2 lo I -S o) "' o

^^ -^ 0) ii ^^-^ r^-^ c^ M o.

lo i^. o. T- VJ' ^-),

"os:^:^<=^:

;:?l ^^l i ^^ I I I "^^ I ^^.-^i^-S^-^:?

■^ O -—0)

■s I

I I ^c^^t.^^1 I

I I I I I

I I I T-'^ I I I "^ "^"^ '-' "^ iil--^ O^"^ ^"^ I!-^ t^ I I -^ ^ I -^ "M

I I I Ij ^ I I I J7 ^ M '~' s ^ "* '^ ^-7^ ^1 I v? ^ I ^ *-i

---. O -- ■ -~ Oq -— - "r^ I -£!■

o:?;s^^^;:3-|6f|fc^-S^|^l "l^;;^^

M O oo ^ J^ CO tn h

I -^^S I I i

^o N ^ « rv^i; r^iS. ,/-,-2; ^ I -:::r'-^^^-£i'

1

1

■-1-

f ^f*^!

^

^

?

1

1 -5:'r^i n.o
1 o^l -J;^

1

1

fsjfg 1

1 :2o^^^x^N

1 in in t^ r^ ^0 ^

n-r.'

1

1

1

1

1 |-;&-§f:

1

1

l^t^ 1

1 ?^^o^^^
1 3./^^ -i"^^

■^-0 0=0 1

1

1

1

1

1

1

1 1 1

1 1^ 1 -^^ 1

^^'f"^!

1

i

1

1 §^^^§vi

-^

^t

1 ^^ 1 1 1

^^ 1 1

"o

0)

■t

rf

1 -C-^-^^l

1

2;

"^"^-^"^1

•^ ^^^0^ 1

1 ^^1 ^^1

-^0)^=0 1

"5"

s

1

'c'

^ 1

0 ._ _

cs 5J cS g 'E

- c - ° ^

■g --3 "3 -S S

111 hi

J u K M Q

: o .5 c

I I I I 5

I?; o C5 o

rH t^i CO -^

D .'" ~ .S;

r-i ci n n

|i

>

saqu;
-qng

1 s;^::^ i^^^^-^^^o^

S° tx N_ 0 iioo

1

saqiJX

O O ;^

-2:00 1 1 ^o 1 1 1 1 1 -tL^ 1

1 1

PI'OAi PIO 8M1 IIV

■^9

uoiSay^ uBii«j'}snY

^^■^-§-§-1 ' §-1 1 1

"m"*^

1 1

1

iioiSajj
UBDUJY qinos

|::?|:'l if^ 1 1 ■=^-' 1 1 1

^f^

uoigaa

^;:r'S:?-i^^l 1 1 1 1 1 1 I 1 1 1

00 VO M

i

uoiSa}j
IIBDUJY it30Tdo.ix

f^|^§^t?l 1 1 t:ri

0,^

1 ^=^

^00

i

uoi§s>i
oil'Eis Y -08t; do jn g

1 1 1 1 I 1 1 1 ! 1 1 1 1 ts

31

U0I§8>I

1 1 1 1 1 1 1 1 1 1

1

-^■^ 1

S i

■BDU8UIY (lY

0 "^■"^"^'^M '^'?<5"^'^;i?t^^,'o^'^) 1

1

1 1

^^

^ 1

uoi8ay^ u^IIT^O

§^§^1 |If;?-§-^:rt;rf-i

1

i 1

-^s

i

uoigaji ireinzTug

|-f^k|li |s-§:t^-i I

i

1 1

1

tf

uoiSaji auipuY

KSf^|t^-?^§^^-^- i 1 1 1 1

saipui ?S9A\

|j^|--|;^|^l |-§?i -^-1 1 ! 1 1

uoigr>-a
saiBig pa:)iu;i

■|?|§§^^fg-^-.-§:ri f-,1 1 1 1 1

uoiSey; UBOixaj^

1

3
1

M 1 0 1 1 1 i 1 i

1 1 1 1 §. f 1 ;^ = 1

t

<

i 1
11

1
6

c S

OLIGOCENE

MIOCEN'E

PLEISTOCENE

© © (a) ■«iffii^;^(g)@(g)(u)@ Q ,c

-AGERAT/n»E
-AOEMOi^vLiriAE

\ASTE.kLA€

-ConyziriAt

-BACCHARlDinAC
-HETEROCHRCiMirryif

■BELUDlMflt

-PLUCHCiHAE

-EU-OriAPHflLICAEi^
-HfL)CHfiySC/1£ J

-RELHAri/IHAE
-ATHRIXIIM4E
■IMULIN/IE
-BUPHTHA.LMIM>Sf

yhmzAt

-CEnrfluREirwE

-CARLirsm^E
-ECHinoPsiDinflE

VERnon/f/^£

^pio»— r — vcRnoniiMAt

^ ^ .5EnEc/or<Eyi£

ru55;L/W)ninA£
•SEMECiOMiriAE

' — OTHOntiinAL

■CALEmULUL
AliCTOTI[)CAL

' — ARCrOTIO'riJf
■G0BT£fil/r<;9E

MLEniCAZ

- -uzLznunAL

■TAGZTirifKZ
•J/^UMElNflf
■MiRllHAt

MUTI5IEAE

— :BARr^flDE5MM«C
■MASSAtiVHNAC
■G(XWM/>T/ir(flE
■ON03CftlDiN4f

--GfReECiir<AE

CiCHOfilEAE

' SCOLVMINXE
■LACTuClfiAE

: OTHER ivBTHlSCb

yflMTWEMIDOE

7 — CHRVSOHTMCIiOinjr
-I ArtTMEMioirwc

HZimjHEAE

MiLLtO'MAE

MEUlMPOOIinylE

flMBRosnn/iE

LAGA^CEmflE

PETRCBnnac
•URBfSlMinAE

ZrtMliHAE

MAO/;/SAt
■GALIMSOG/MAE

~.0RECP5tDlHAZ

126 AGE AND AREA, AND SIZE AND SPACE, [pt. ii

of course, only one of rough approximation, but the inequalities
more or less cancel out when the genera are taken in groups of
ten or more as specified for the Age and Area hypothesis.

The average generic area has been determined by adding up
the marks (= square miles in millions) for all the genera in each
group and dividing by the number of genera. This "average
generic area" has been determined for each tribe and sub-tribe
in the Compositae, firstly, for the whole world (using my own
notes); then for the Old World, all America and, in the sub-
tribes, for each of the twelve great regions into which Bentham
divided the world in relation to the Compositae (7), using the
data given by that authority. The use of two sources for the
data has introduced some slight discrepancies in the figures, but
the value of the check also introduced by this method makes
these slight differences of no real consequence. All these data
are presented in Table I, and on the whole, taken in conjunction
with the relative ages and sources of these groups as previously
determined (fig. on p. 125), they form a striking corroboration,
both for the Age and Area hypothesis and for the pre^-ious
phyletic conclusions.

in accordance with the indications of a diphylctic origin of
the Inuleae (103, p. 301), that tribe has been given in Table I
as two, the Gnaphalieae (limited) which includes the first five
sub-tribes (with the Gnaphaliinae divided into Eu-gnaphaheae
and Helichryseac) together with half the Relhaniinae; and the
Inuleae (limited) which includes the last three sub-tribes to-
gether with the other half of the Relhaniinae (cf. fig. on p. 120).
The Inuleae as a complete tribe may, therefore, be omitted.

Taking the tribes in order of origin, as given on p. 125, we find
(Table I, col. 16) that the average generic areas range from 7-9
(Senecioneae) through 6-5, 6--4, 6-2, 5-6, oo, 3-9, 5-3, 4-9, 4-5,
4-8, 3-6 and 3-8 to 3-6 (Calenduleae). In this Hst there are three
figures not in series — 3-8 for the Arctotidcae follows 3-6 for the
Helenieae, and 4-8 for the Vernonieae follows 4-5 for the Eupa-
torieae, but these two pairs of tribes are approximately of the
same age (p. 125), and the relative positions could be reversed
without any argument. The third figure not in series is 3-9 for
the Mutisieae, a tribe in which much geographical splitting of
the genera largely increases the number of genera in proportion
to the area occupied, thus decreasing the average generic area
for the tribe.

The gradual increase of average generic area with geological

CH. XIII

IN THE COMPOSITAE

127

age IS shown even more strikingly when the mean is taken for
the tribes arising in eaeh sub-division of the geological periods;
the Mutisieae are then the only exception in the series (Table II
col. 4). \\hen the mean is taken for each period an unbroken
series, running 7-2, 5-9, 4-6, 4-3, 3-6, is obtained (Table II, col. 5).

TABLE II

Pliocene
Upper
Middle
Lower

Miocene
Upper

Middle

Lower
Oligocene
Upper
Middle
Lower
Eocene
Upper

Middle
Lo\i'er

Cretaceous
Upper

Calenduleae

Arctotideae

Helenieae

Vernonieae

Eupatorieae

Cynareae

Inuleae (ltd.)
Mutisieae

Cichorieae
Anthemideae
Astereae
Heliantheae

Gnaphalieae (ltd.)
Senecioneae

Average

Generic

Area

3-6

3-8\
3-6J

4-8)
4-5/
4-9

5-3
3-9

5-5)
5-6/
6-2
6-4

6~,\
7-9i

Average Average

for for

Divisions Periods

3-6

5-3 5

6-2

6-4

4-6

Arranging the sub-tribes within each tribe in the order of
origin as given on p. 125, we find that even there the series follow
the average generic area scries more or less. The series (Table I,
col. 17) for the following six tribes are unbroken in each case: '

Anthemideae: 6-2, 4-3.

Inuleae (hmited): 8-4, 3-5, 3-3, 1-7.

Cynareae: 5-9, 5-1, 4-5, 3-3.

Eupatorieae: 5-3, 4-1, 3-3.

Vernonieae: 5-1, 4-0.

Arctotideae: 4-7, 3 0, 2-8.

The Calenduleae has no sub-tribes, so that only half of the
fourteen tribes require special consideration.

As we are in many cases dealing with fewer than ten genera,
and since a number of the sub-tribes are more than slightly

128 AGE AND AREA, AND SIZE AND SPACE, [pt. u

artificial, while several contain genera which are exceptionally
widespread because of special dispersal mechanisms, we cannot
expect a complete correspondence. On the other hand, a family-
tree is available (p. 125) which was worked out in the first place
from the morphological characters of the styles and stamens
(103, fig. 7) and subsequently modified only slightly as a result
of the consideration of the most extensive data. It is, therefore,
interesting to examine the deviations from the numerical se-
quence in average generic area for the other seven tribes.

Senecioneae. The Tussilagininae have been shown to be a
somewhat mixed group of genera, separated from the Senecio-
ninae in a rather artificial way (103, pp. 39 and 298), and these
two sub-tribes are, in fact, fused by Hoffman in the Pflanzen-
familien. The proper statistical procedure is, therefore, to take
them as one group for comparison with the other groups within
the tribe; then we obtain another unbroken sequence — 8-9, 6-8,
3-6 (Table I, col. 17).

Gnaphalieae (limited). With the Eu-gnaphalieae and Heli-
chryseae as two distinct groups the series for the Gnaphalieae
reads 9-8, 8-8, 9-3, 4-3, 1-6, 1-7, 3-2. There are in this case two
marked exceptions to the sequence. The first is the Filagininae
(9-3) Avith only eleven genera including Micropus as a widely
spread weedy type, and Filago also of the weedy type and a
distribution suggesting either early dispersal by man or-a poly-
phyletic origin. If this genus were broken up into three, as was
done by many of the earlier syhantherologists, the average
generic area for the sub-tribe would be 7-8, and the sequence
would be imbroken except for the last sub-tribe. The second
exception is the Angianthinae (3-2), chiefly an Australian group
with only ten genera, the distribution of which in Australia may
be somewhat less on the whole than has been estimated. Such
a reduction in this sub-tribe would bring the average for the
Gnaphalieae to about 6-3, but a similarly careful revision in
detail of the other sub-tribes might result in raising the average
one or more decimals, so that such changes may be considered
negligible when the broad outlines of the history of the family
are being considered.

Heliantheae. The series for this tribe appears rather irregular,
running thus: 6-9, 9-6, 3-7. 6-5, 2-7, 3-2, 12-8, 3-8, 6-0, 1-3; but
most of the sub-tribes contain less than ten genera. It is, there-
fore, advisable to group them; the first three groups are of early
origin (see p. 125), while those numbered 6', 6", 6'" in Table I

CH. xiii] IN THE COMPOSITAE 129

are so marked because they are presumed to have arisen about
the same time. Taking these six sub-tribes as two groups we
get the series 6-7, 6-5, 2-7, 6-6, 6-0, 1-3. In this series there
is only one prominent break, 2-7 for Madiinae with only seven
genera. Considering the sub-tribes showing exceptional figures
for average generic area, there is the Coreopsidinae (9-6) with
17 genera of which Bidens is a very widely spread weedy type
with a very special dispersal mechanism, especially when the
early migrations of man are regarded as a means of dispersal.
The Melampodiinae shows a low average, but it has an average
generic area very similar to several of the other young sub-
tribes. The other exceptional figure is 12-8 for the Ambrosiinae;
in which there are only nine genera, of which both Ambrosia
and Xanthium are widely spread weeds, the latter like Bidens
with a special dispersal mechanism.

Astereae. The series for this tribe is very uniform, running 5-4,
6-0, 7-5, 7-7, 7-0, 5-5. Such a sequence, with the most primitive
sub-tribe showing the lowest average generic area, might well
seem to show that the present thesis cannot be maintained, but
only two of the six groups have more than ten genera. Further,
the division into sub-tribes is introduced by Bentham (7,
p. 402) thus:

The vast tribe of Asteroideae is neither so well marked as a
whole..., nor yet is it well divisible into distinct groups. Nearly
the whole of the 90 genera, comprising above 1400 species, pass
into each other through exceptional or intermediate forms.... The
Asteroideae not being divisible into distinct sub-tribes, we may
for geographical purposes consider a number of types with the
various divergences from them.

Bentham also gives the key to this anomalous distribution as
follows: ''Aster, taken in its most extended sense, ranges over
the whole area of the tribe; but isolation has been ancient
enough to admit of its having established special forms in
different countries, which are now admitted as genera by most
botanists" (7, p. 402); and in the Solidago type (7, p. 410):

We have here about 320 species in 24 genera, all nearly allied
to each other and only distinguished technically from Aster and
its immediate allies by the homochromous florets, the ray florets,
when present, being yellow, like the disk — a character in general
of so little value that it cannot, in Senecio for instance, be ad-
mitted as of more than specific importance.

Translating these quotations from Bentham, who makes
several other statements of a like nature (cf. op. cit., pp. 405

130 AGE AND AREA, AND SIZE AND SPACE, [pt. ii

and 412), into modern terms (46), one would say that the Astereae
was predominantly a case in which a primitively world-ranging
type has been differentiated in situ with practically no spreading
of markedly new types from definite centres of origin. Some
such explanation is almost necessary for the frequency of inter-
mediate species and the grading of the Aster, Erigeron and
Conyza types into each other,

Guppy's theory of "Rank and Range," which, although
similar to "Age and Area," is slightly different, is therefore
exemplified in this tribe of the Compositae; whereas the other
tribes are examples rather of "Age and Area." A detailed
examination of many of the sub-tribes in other tribes shows
that Avitliin the sub-tribe there are seldom groups of genera
which show markedly different average generic areas. The Eu-
gnaphalieae and Hehchryseae are exceptions. This leads to the
conclusion that, in spite of the large numbers of genera and species
in the Astereae, this group is really of the same "rank" (with
regard to differentiation and Age and Area statistics) as the
normal sub-tribes of most of the other tribes.

Cichorieae. The ten sub-tribes into which this tribe is divided
are admittedly artificial. Bcntham {op. cit., p. 475) writes: "It
is very difficult to arrange these genera into sub-tribes; and
those we have adopted are in a great degree artificial, and have
little or no connection with geographical distribution; we must,
therefore, now consider the principal genera separately."

The Lactucinae have been indicated (103, pp. 271 and 282. and
p. 125) as the primitive grou]3, while the Seorzonerinae have
been indicated as a fairly definite and advanced group (103,
p. 282). Scolymus is quite a distinct genus and the only one in
the Scolyminae. Grouping the other seven sub-tribes together
as one, we have the series— 6-2, 5-9, 3-7, 4-0. The last figure is
of little importance since it represents only one genus, while the
other figures are in the usual sequence. Of the seven sub-tribes
which are grouped only one. Hyoseridinae (42/10 = 4-2), has as
many as ten genera; while the other two sub-tribes which are
taken singly show the figures 69/11 = 6-2 (Lactucinae) and
37/10 = 3-7 (Seorzonerinae). The inclusion of the American
genera in Seorzonerinae is distinctly artificial, and if only the
Old-World genera are taken the average is 5-8. Then the series
for the Cichorieae reads 6-2, 5-9, 5-8, 4-0, and it is in complete
sequence and in perfect accord with the origins given on p. 125.
With appropriate statistical treatment, therefore, this tribe

CH. xm] IN THE COMPOSITAE 131

shows the action of the Age and Area law, and it is interesting
to note that it is a clear case of an origin in and dispersal from a
definite centre (see 103, PI. 2, fig. 31), contrasting markedly
with the Astereae. ^

Mutisieae. As stated above (p. 126) the low average generic
area for this tribe is to be explained, at least in part, by the
rather artificial splitting into geographical genera. This occurs
chiefly in the Gerberinae and Gochnatiinae, and an allowance
tor t ns "error" would bring into greater prominence a feature
which IS marked in the series as given. The origin of the Barna-
desimae is still obscure, therefore the 3-5 for that sub-tribe may
be neglected for a moment. The series for the other sub-tribes
reads 3-3, 2-6, 5-6, 4-2. A diphyletic origin for the Mutisieae, as
far as the Barnadesiinae is concerned, has already been sug-
gested (103, p. 211), but the data for Age and Area suaaest very
strongly that the origin of the rest of the l\Iutisieae has also been
chphyletic, giving a triphyletic origin for the tribe as a whole.
Some difficulty was experienced in tracing the interrelationships
of these sub-tribes (cf. 103, pp. 211 and 305), and it is quite
probable that the purely American sub-tribes, Nassauviinae and
Onoseridinae, are relatively recent, while the other two sub-
tribes are a more ancient group evolved along similar lines.
Then we have two groups and two series, 3-3, 2-6, and 5Q, 4-2,
with the Barnadesiinae (3-5) intermediate from probably a
third origin. The structural affinities combined with the Age and
Area data allow of no other explanation of the origins Sf this
miique tribe.

Helenieae. This, the last of the tribes showing exceptional
figures, gives the series 6-4, 3-8, 5-6, 6-1, 2-1; but three of the
five sub-tribes have less than ten genera so that further grouping
is required. This can be done by taking as one group the first
two sub-tribes, marked 1' and 1" (Table I, col. 1) on account of
suggested simultaneous origin, and as another group the last
two sub-tribes 3' and 3" for the same reason. The series then
becomes 4-6, 5-6, 2-7. The middle figure is not in sequence, but
it refers to the Flaveriinae with only three genera (17/3 = 5-G).
and doc-s not, therefore, vitiate the general argument which
applies quite well to the first and last groups, both Avitli more
than ten genera.

With the one exception of the Astereae, which has been
explained (p. 129), the statistical data for Age and Area in Com-

132 AGE AND AREA, AND SIZE AND SPACE, [pt. ir

positae can, therefore, be said to demonstrate in some consider-
able detail the action of the Age and Area law as far as relative
age is concerned in a group, the evolutionary history of Avhich
in time can be confirmed by many other lines of evidence.

Size and Space. Dr Willis has also forwarded another pre-
diction to the effect that, "on the whole the area occupied by a
genus (taking a great many, say ten allies at least) varies in the
same sense as the number of species it contains." This also has
been worked out for Compositae, using the number of species
recorded for each genus by Bentham in the Genera Plantaruni,
in order to avoid the inequalities of modern species-splitting
and specialisation in particular genera. The results which are
given in Table III, when analysed properly, form a remarkable
verification of this prediction. The average number of species
per genus does follow on the whole the same series as the average
generic area.

TABLE III

Species/Genera

Average

Tribes and Sub-tribes

Tribes

Sub-tribes

Generic
Areas

Seneeioneae

1266/44 = 28-7

7-9

1. Senecioninae

■°i.t":'^S} 3-^

8-7

3. Tussilaginiiiae ...

—

9-4

1+3

—

1077/33 = 32-6

8-9

2. Liabinae

—

50/5 =10-0^
139/6 =23-1/ '^ '

6-8

4. Othonninae

3-6

Gnaphalieae (ltd.)

973/100 = 9-7

6-5

Inuleae (Benth.)

—

—

6-1

1. Eu-gnaphalieae ...

—

214/20=21-4

9-S

2. Plucheinae

157/16= 9-8> _.,
46/11= 4-2/ 7 5

8-8

3. Filagininae

—

9-3

4. Helichrj'seae

393/26 = 15-1 1

4'3

5. Tarchonanthinae...

10/3 = 3-3 [10-9

1-6

6. Relhaniinae

-

91/14= 6-5->J
64/10= 6-4/

7. Angianthinae

3-2

1. Inulinae

—

131/19= 6-9

8-4

2. Buphthalminae ...

—

45/16= 2-8

3-5

3. Athrixiinae

—

39/7 = 5-5

3-3

4. Relhaniinae

91/14= 6-5

1-7

Heliantheae

1101/138 = 7-9

6-4

1. Verbesininae

—

604/57 = 10-6

6-9

2. Coreopsidinae

—

153/17= 9"0

9-6

1 + 2

7-5

3. Melampodiinae ...

—

95/20= 4-7

3-7

1+2+3

—

852/94= 9-1

6-7

4. Galinsoginae

—

81/7 =11-5

6-5_

5. Madiinae

—

52/7 = 7-4

6'. Milleriinae

. —

40/11= 3-6)
40/8 = 4-4 M-o
25/6 = 4-ii

3-2)

6". Ambrosiinae ...

—

12-8 U-6

6'". Zinniinae

3-8.1

CH. xiiij IN THE COMPOSITAE

TABLE III {Co7itd.)

Species/Genera

Tribes and Sub-tribes

Tribes

7. Lagasceinae

8. Petrobiinae
Astereae'

1. Homochrominae ...

2. Heterochrominae

3. Conyzinae

4. Bellidinae

5. Baccharidinae

6. Grangeinae
Anthemideae

1. Chrysanthemidinae

2. Anthemidinae
Cichorieae

1. Lactucinae

3. Scorzonerinae

4. Scolyminae

2. All other sub-tribes

Hyoseridinae

Lapsaninae

Crepidinae

Hieraciinae

Hypochoeridinae

Dendroseridinae

Rhagadiolinae ...

Mutisieae

? Barnadesiinae

1. Nassau viinae

2. Onoseridinae

3. Gerberinae

4. Gochnatiinae

Inuleae (ltd.)

Cyiiareao^

1. Centaureinae

2. Carduinae

3'. Echinopsidinae ...
3". Carlininae

Eupatorieae

1. Ageratinae

2. Adenostylinae

3. Piqueriinae
Vernonieae

1. Vernoniinae

2. Lychnophorinae ...
Helenieae

1'. Heleniinae

1". Tagetinae

2. Flaveriinae

3'. Jauineinae

3". Baeriiiiae
Arctotideae

1. Arctotidinae

2'. Gundeliinae

2". Gorteriinae
Calenduleae

(1386/91 = 15.2)

671/45 = 14-'
824/56 = 14-;

438/53 = 8-

(306/56 =
(992/37 = :

5-4)
6-8)

(727/35 = 20-7)

534/40=
278/60 =

236/17^
1 14/8 =

13-3
4-6

Sub-tribes

7/1 =
4/3 =

323/26 = 12-4
601/36=16-7
109/10=10-9

80/10= 8-0
256/3 =85-3

17/6 = 2-8

399/30 = :
272/15 = :

148/11 = ]

165/10=]

3/1 =

508/34 = ]

50/10 =

II/3 =
162/6
171/3

95/5
8/2

1 1/5

3-3

= 3-0
= 14-9
= 5-0
■ 3-6
27-0
57-0
19-0
4-0

2-2

1 1/2 = 5-5
166/14 = 11-8
62/10= 6-2

[Ol/lO=IO-I

98/17= 5-8

402/10 = 40-2
476/17 = 28-0
75/2 =3 7-5)
39-8 = 4-8/

595/20 = 29-7
77/8 = 9-6
55/7 = 7-8

482/29 = 16-6
52/11= 4-7

48/7
114/14

9/3 = 3-0

12/6 = 2-0

Average
Generic
Areas

5-6

6-0

1-3

5-4
6-0

7-5
7-7

6-2

3-7

6-2

3-7
4-0
5-9
4-2
7-3
6-3
9-3
ro-4

:?}^-

2/6 = 2-0\ ^

5/30=- 3-if -•^

3-5
3-3
2-6
5-6

5-9
5-1
4-5
3-3

5-3
4-1
3-3

5-1

95/3

1 1 5/8

3/2

1 1 8/7

14-3
16-8} ^3-4

3-6

3-8

3-6

5-6
6-n
21} --7

4-7
3-0

Astereae + Eupatorieae =16-7.

'^ Cynareae-f Inuleae (ltd.) =

134 AGE AND AREA, AND SIZE AND SPACE, [pt. n

In this case it is clear that the larger the groups of allies taken
the more reliable the results, therefore the Astereae has been
grouped with its derivative tribe (Eupatorieae) and the Inuleae
(limited) has been grouped with the derived Cynareae. In this
way the number of larger groups has been reduced from fourteen
to twelve, and these give the following series of averages for
number of species per genus ^r

2S-7, 9-7 II 7-9, 16-7, 14-9, 14-7 \\ 8-2, 13-9 || 13-3, 4-6 13-9 \\ 14-2.
The last two numbers are not quite in series, but they represent
groups with only seventeen and eight genera respectively. All
the four numbers out of sequence are practically in series with
each other. The low figures for the Gnaphalieae (9-7) and Heli-
antheae (7-9) may be traced to the fact that many of the genera
are plants of the plains, where the average number of species
per genus is lower, according to Harshberger (cf. 103, p. 187),
than it is along the mountain ranges with their highly diversified
topography. The low figure (8-2) for the Mutisieae furnishes a
curious piece of evidence in favour of the prediction, for it may
be noted that the average generic area is also lower than it
should be in the series; and the geographical splitting of genera
already mentioned would reduce not only the area but also the
number of species per genus. The low figure (4-6) for the Helenieae
also occurs in conjunction with a low figure for average generic
area. Thus, of the twelve groups taken only two do not fall into
the same series as that for average generic area.

The increase of average number of species per genus with age
shows even more strikingly when the mean is taken of the figures
for each geological period. The figures for the groups arising in
the five periods concerned are separated by vertical lines in the
series as given above; and the means for the Upper Cretaceous,
Eocene, Oligocene, Miocene and Pliocene read thus: 19-2, 12-8,
11-0, 10-6, 14-2. Only the last figure is out of sequence, and it
represents a single small tribe, the Calenduleae, Avith eight genera
and 114 species, which arc scarcely sufficient for reliable data.

When we take the subordinate groups the same correspondence
between the two series shows very well on the whole. For the
present purpose as much grouping of the sub-tribes as seems
reasonable has been made in order to get groups Avith more than
ten genera. Taking the tribes seriatim we get the following data:
Senecioncae. The Tussilagininae are sunk as before giving 32-6;
the Liabinae and Othonninae are grouped to get more than ten
1 Numbers in italics are in series or nearly so.

CH.xiii] IN THE COMPOSITAE 135

genera giving 189/11 = 17-1; and the series is complete. An
interesting point is the large figure (40-5) for the basal sub-tribe
of the family, which is exceeded only in the Baccharidinae (85-3)
and the Hieraciinae (57-0), both with only three genera.

Gnaphalieae (limited). The Plucheinae and Filagininae are
grouped as similar in age and in area, giving 203/27 == 7-5; the
Helichryseae is grouped with its derivatives Relhaniinae and
Angianthinae giving 548/50 = 10-9. The series for the tribe then
reads 21-4, 7-5, 10-9, 3-3. The larger figure for the Helichrysum
group may be explained as an effect of the diversified topography
in South Africa; the other figures are in series.

Imdeae (limited). Two out of the four figures are in series, but
the numbers of genera are low in all cases.

Ilelianiheae. The first three sub-tribes and those marked 6',
6", 6'" are again counted as two groups; and the series reads
thus: 9-1. 11-5, 7-4, 4-0, 7-0, 1-3. The exceptions are 11-5 for the
Galinsoginae Avith only seven genera and 7-0 for the Lagasceinae
with only one genus.

Astereae. The series (12-4, lG-7, 10-9, 8-0, 85-3. 2-8) in this
tribe again shows the series following age as previously suggested,
with two exceptions. These are 85-3 for the Baccharidinae with
only three genera, and 12-4 for the Homochroininae as compared
with 16-7 for the Heterochrominae. The latter figures, when com-
pared with 5-4 and 6-0 for average generic area, are seen to
follow the sequence for area.

Anthemideoe. In this tribe the figures are not in the proper
series.

Cichoricae. The series in this case rims practically in the oppo-
site direction to the average generic area series, but the numbers
of genera are low in three of the four groups. These two excep-
tional tribes, it should be noted, show the proper sequence as
tribes, so that it would seem that, not only the admittedly arti-
ficial subdivision in both tribes (see below), but also the small
number of genera in most of the subordinate groups has an
effect on the corresi^ondence of the series in these cases.

Mutisieae. Adopting the triphyletic origin of this tribe, which
is suggested above, we have three sets of figures Avhich corre-
spond to the three series for area and show the same sequence
li'ithin the sets.

Cynarcae. Grouping 3' with 3" as being of the same age, we
get the series 40-2, 28-0, 11-4, which corresponds completel)^
with the series for area.

136 AGE AND AREA, SIZE AND SPACE [pt. ii, ch. xiii

Eupatorieae. The series here (29-7, 9-6, 7-8) also corresponds
completely with the series for area.

Vernonieae. The series here (16-6, 4-7) also corresponds com-
pletely with the series for area.

Helenieae. Grouping 1' with 1", and 3' with 3" as being of the
same age we get the series 7-3, 3-0, 2-9, which agrees completely
with the order of origin as given previously (p. 125).

Arctotideae. Grouping 2' with 2" as being of the same age we
get the series 14-3, 13-4, which also corresponds completely with
the series for area.

The prediction that the series for the average number of
species per genus will follow those for the average generic area
may, therefore, be said to be verified for the tribes on the whole,
ten out of twelve showing a similarity; and also for the sub-
tribes on the whole, with the exception of three tribes out of
fourteen. Further, the divergences amount to two out of four
sub-tribes in the Inuleae (limited); while the subdivision of the
other two tribes (Anthemideae and Cichorieae) is admitted by
Bentham to be artificial. For the Anthemideae he records (7,
p. 451): "In the Genera Plantarum we have, for convenience'
sake, classed the genera somewhat artificially," and (op. cit.,
p. 450) "It is not easy, either, to group them into well-marked
sub-tribes." On the artificial subdivision of the Cichorieae he
has already been quoted (p. 130).

The conclusion is, therefore, quite justified that in the Com-
positae on the whole both the average generic area and the
average number of species per genus are closely related to
absolute age.

CHAPTER XIV

AGE AND AREA FROM A

PALAEOBOTANICAL STANDPOINT

By Mrs E. M. Reid, B.Sc, F.L.S.

Any student of ancient floras must feel that in its power to
meet the facts of geology and palaeobotany lies the supreme
test of Dr WiUis' theory of Age and Area. The time has not come
when such a test can be applied Avith any degree of fullness, for
the history of Tertiary floras, which are those chiefly concerned,
is still but imperfectly known; and more especially is this true
of their migrations. Nevertheless, even if we cannot make a full
comparison, it may be of use to make a beginning, by comparing
such conclusions as have been reached by the two studies; not
only for the sake of testing a new theory, but because, if it holds,
palaeobotany has much to learn from it of the past history of
plant-life and must therefore reconsider its conclusions in the
light of new knowledge.

In what follows I do not propose to go much bej^ond the range
of my own studies, but these have been largely concerned with
the questions of which Age and Area treats, the migration of
floras, the age of species, and the extermination of species. The
material of study has been the Pleistocene floras of Britain, and
some late Tertiary floras of West Europe, chiefly the following
Pliocene floras: Cromerian (East Anglia), Teglian (Holland),
Castle Eden (Durham), Reuverian (Dutch-Prussian border),
Pont-de-Gail (Cantal). These have been investigated by an
examination of seeds and fruits.

Plant Migration. If there is one fact which has emerged
more clearly than another from the study of Pleistocene and late
Tertiary floras in West Europe, it is that at different geological
times, different floras have occupied the same locahty. By
"different floras" is meant different assemblages of plants which
have lived in the past, as they do in the present, in regional, or
in ecological association, more especially in climatic association.

Thus, by the quantitative study of pollen-grains in the suc-
cessive horizons of the peat-bogs of Scandinavia, it has become
possible for Scandinavian workers to trace successive assemblages

138 AGE AND AREA FROM A [pt. ii

of plants at different periods, not only so as to gain a knoAvledge
of the species occupying the country at successive times in the
Pleistocene, but so as to gain also some knowledge of the pro-
portion in which those species flourished (60).

Or, again, we may take in our own country the succession
seen in our eastern counties. In the Cromerian (83) at the close
of the Pliocene period, Ave find a temperate flora almost identical
with that now inhabiting East Anglia. At a later period we find
a flora composed of plants now inhabiting colder regions — sub-
arctic, alpine, or cold temperate (15, 72, 82). Yet again, in the
present day, after a further interval, when the climate has once
more become temperate, we find the old temperate flora of the
Cromerian back in its former locality, shorn only of a few of its
elements.

The instances of such successions could be multiplied, but the
above are sufficient to show that we haAc definite evidence of a
continual swaying to and fro of plant-life.

Evidence of this kind can scarcely be interpreted otherwise
than as indicating the movement of plant assemblages, imder
the influence of climatic change ; in other words, migration.

But if migration has occurred, how has it been brought about?
The answer is suggested by Dr Willis' theory of Age and Area,
though the idea of plant movement embodied in it would seem
to need some modification. Dr Willis suggests, as a result of his
work, that newly arrived, or newly formed, species tend to
spread outwards in all directions from their point of arrival, or
point of origin, like rings formed by casting a stone into a pool.
In such a tendency Ave see a motive force; but migration is a
directed movement, and the combined CA'idence of geology and
palaeobotany indicates that the directing force is change of
climate. Each species flourishes best under definite climatic
conditions, Avithin limits appropriate to itself. Change of climate,
acting ecologically, Avorks as a Aveeding process, so that move-
ment, instead of being general all round, becomes a moA^ement in
one definite direction — migration.

From A'arious considerations of geology, palaeontology, fossil
and recent botany, the conclusion has been reached that if
change of climate has been from cold to heat, in a flat country
migration has been poleAvards, in a mountain country upAvards.
If the change has been from heat to cold, then in a flat country
migration has been equatorAvards, in a mountain country down-
wards.

CH. XIV] PALAEOBOTANICAL STANDPOINT 139

In the paper on the "Sources and Distribution of the New
Zealand Flora, pp„ 354-362, the conclusion is drawn that there
were two main plant-invasions by which New Zealand was popu-
lated, a northern and a southern. In a subsequent paper (134)
this conclusion is amplified. The northern invasion is split into
three a principal one from the north, and two subsidiary ones,
called the Ivermadec and western invasions respectively '

It will at once be seen that we have here postulated three
invasions (northern, western, and Kermadec), which in their
general direction are poleward, and one invasion (southern)
winch IS equatorward. Bearing in mind the conclusions we have
reached as to the relationship between direction of migration
and change of climate, it would appear that the three poleward
mvasions must have occurred whilst the climate of all the regions
involved, or possibly only that of New Zealand, was becomina
warmer: the southern in^'asion, equatorward, must have oc"
curred whilst the climate of the regions involved was becoming
colder. '^

It will be sufficient for our argument if we consider onlv the
two main divisions, the northern and southern.

Dr Willis brings forward strong evidence (132, and cf p 81)
to show that of the two, the northern was much the older We
have,- therefore, to consider one very old migration polewards
whilst the climate was warming, and one newer, equatorwards,
whilst the climate was cooling.

For the migration oi floras (plants in ecolooical association)
as opposed to the casual transport of individuals, Dr AVillis
rightly insists that land connection, complete or all but com-
plete Avith the source of dispersal, at the time of dispersal is
necessary. It is inconceivable that associated assemblages could
travel 111 one definite stream except bv land. A sea-passage
must have sifted out species with inferior powers of dispersal
across Avater in a Avay that is not found to have occurred.

We have now traced the conditions necessary for these two
main invasions as postulated. For the northern", a very ancient
land connection between New Zealand and Indo-iAIala\-a, Avith
a climate increasing in temperature, certainly in New Zealand,
and probably over the whole of these regions." For the southern,'
a very much later land connection southwards, at least as far
as the Campbells and Aucklands, Avith the climate of these
regions becoming colder. It will readily be seen that for the
western and Kermadec iuA-asions, conditions A'cry similar to

140 AGE AND AREA FROM A [pt. ii

those needed for the northern invasion must have occurred,
possibly at intermediate periods.

Such are the problems which present themselves for solution
when we attempt to apply the results of the comparative study
of Pliocene and Pleistocene floras to the postulated migrations
of the New Zealand flora. Whether the geology of New Zealand
and Indo-Malaya will bear out the possibility of these changes
of sea-level associated with the corresponding changes of climate,
it is for students of those regions to say. The answer is outside
the range of my knowledge.

Extermination. The study of West European Phocene floras
led to the recognition of an extinct Tertiary flora in West Europe.
This flora, which I have named the Chinese-North-American
Association of Plants, is now represented by two living plant
associations; the one the forest-belt flora of the East Asian
mountains, the other the allied flora of parts of North America.
There is much evidence from recent and fossil botany, and
geology, to show that all three are migrant floras, branches of a
common polar or circumpolar flora, which migrated southward
in later Tertiary time under the influence of a cooling climate in
the Northern Hemisphere. The travel southward of each branch
must have extended over manj^ himdreds, more probably thou-
sands, of miles. In the end there resulted the complete exter-
mination of the European branch, and the isolation of the other
two, in regions of the Old and New World rcspecti\xly, separated
by many thousand miles of sea and land.

In the history of this flora we see exemplified two kinds of
extermination, both of which are concerned with the questions
raised by the study of Age and Area. In the first place we have
regional extermination; no trace being left, in the region where
such extermination occurs, of the life that has been. In the
second place we have specific extermination; the species being
killed, but an alhed one taking its place.

Regional Extermination. Regional extermination, as
illustrated by the history of the Chinese-North-Ameriean flora,
may be of different degrees.

(1) It may be confined to o?2.£r region only. We have numerous
instances of this in our flora. Take, for example, the genera
Magnolia, Liriodendron, Menispermum, and Nyssa. These have
been exterminated in Europe, but have survived in East Asia
and North America ; though they are now represented by different

CH. XIV] PALAEOBOTANICAL STANDPOINT Ui

species in the two regions. Survivals of this kind in Japan and
North America, Avhich are many, led to the recognition by Asa
Gray of the fact that the floras of Japan and Atlantic North
America are allied.

(2) It may have occurred in two out of the three regions. Thus
Phellodendron, Actinidia, and Zelkowa have been exterminated
in Europe and probably in North America, but survive in the
East of Asia. Dulichium, Karzvinskia, Proserpinaca have been
destroyed in Europe and probably in East Asia, but survive in
North America. When such regional distribution has occurred,
there is nothing to indicate in the present how wide the distribu-
tion may have been in the past, or to say whether genera are
survivals or not.

(3) Extermination may have extended to all three regions. In
that case the past is completely wiped out, and in the present
there is no sign of the life that has been. We have numerous
instances of such extermination in the case of species — extinct
species of Dulichium, Euryale, Liriodendron, and so on, far too
numerous to name here; but we have also in all probability
instances of genera exterminated in the many undetermined
fossil forms which would appear to belong to living families, but
cannot be placed in living genera. These forms are mostly un-
named so cannot be referred to, but by consulting the works
enumerated they will be recognised.

It is this fact, that endemic species can frequently be proved
to be survivors from a wide-ranging past, which offered to me
the greatest stumbling-block to the acccj^tance of the theory of
Age and Area. So formidable did the difficulty appear that I
felt it must vitiate the reasoning which pointed to endemics as
the newest elements in plant-life; and yet it was hard to see
where the flaw could lie; and the theory offered so simple and
reasonable an explanation of much that one met with in palaeo-
botany.

A student of Tertiary floras must stand by the fact that in
many instances endemics are survivors from races that once flour-
ished widely, though they do so no more. Take the genus Sequoia.
It once inhabited Europe, Eastern Asia, the Arctic regions, and
large areas of North America ; now it is confined to the Pacific
coast of California. Euryale, again, was once represented by
many species scattered at different times (some at the same
time) throughout Europe; now it survives as a single species
only in parts of Cliina and Assam. Or again, with individual

142 AGE AND AKEA FROM A [pt. ii

species; Liriodendron tulipifera, Nyssa sylvatica, Pilea pwnila,
Dulichium spathaceum were once all inhabitants of Western
Europe ; now they are confined to the North American continent.
The list could be continued to great length but this is enough to
show that genera and species formerly widespread have con-
tracted their range and become endemic. Are Ave then to throw
over the conclusions of Age and Area which show that immensely
the greater proportion of endemics represent new life; and take
the position that the two lines of research are mutually contra-
dictory? It is not necessary if we make due allowance for the
differences of method and subject-matter in the two studies, and
their consequent limitations.

Throughout his work Dr Willis has insisted that his conclusions
are based upon mass-investigation, averages. Consequently he
warns us that, as with all average calculations, though the con-
clusions will be true for the mass they quite possibly may not
be true for the individual. Now the whole of palaeobotanical
research is based upon the study of the individual; consequently
we must be prepared to find that our results may not conform
to the conclusions of mass-iuA'cstigation, though we ought to be
able to explain the causes of divergence. The palaeobotanist
states that some endemics are relicts. Dr Willis replies that if it
is so, they are of no account in comparison with the vastly
greater number of endemics which are not. Not having counted
up the total of endemics in the living flora of the world, as he
has done, I am prepared to accept his estimate that relict en-
demics form only about 1 per cent, of the total. Even if the
percentage were higher, it would not vitiate Dr Willis' reasoning.
And here we come to the explanation of our difference. Whereas
Age and Area fixes its attention upon, and argues from, the
99 per cent., palaeobotan}^ has its attention fixed upon, and
seeks to argue from, the 1 per cent. In the nature of things the
99 per cent, are outside the scope of its investigations, for if
they represent the newest forms of life, then they cannot occur
fossil. Consequently, though palaeobotany is right to hold to
its 1 per cent., it must yield place in the argument to the superior
force of numbers.

Specific Extermination; Extinction and Survival of
Species ; Killing out and Dying out. Specific extermination,
the replacement of old forms by new, is continually met with in
Tertiary botany; one of the most striking instances is seen in

cii. XIV] PALAEOBOTANICAL STANDPOINT 143

the monotypic genus Stratiotes. Miss Chandler's work on the
subject is not yet published, but we may state that a succession
of species has been found at different geological horizons which
carries the history, with but few interruptions, from the top of
the Eocene to the present time. A whole series of extinct forms
lies behind the living species. The same is true of other genera
though the succession may be less completely known. To name
but a few, Dulichmm, Sparganiwn, Potamogeton, Najas, Sam-
bucus, Vitis, Magnolia, Rubus, Cotoneaster and Phellodendron are
all known to have a long fossil record of species that are now
extinct. Given time, the fate of all species is extinction, though
exceptionally they survive for long periods. The oldest living
species I have myself come across are Vitis lanata and Poly-
gonum Convolvulus in the oldest Pliocene (Pont-de-Gail), or
possibly Calla palustris in the Bovey Oligocene.

If now we turn to Age and Area and inquire what evidence it
has to offer, we find that it points to survival as the probability,
unless extermination be due to "killing out."

It is possible that we have here a real discrepancy between the
two studies, for the evidence of universal extinction of species
furnished by the pages of paleobotany is incontrovertible; but
Ave should bear that proviso "unless killed out" in mind. For
further evidence on the subject we may turn to the history of
the Chinese-North-American flora.

We have seen that all branches of the flora have suffered
extermination, either complete extermination, or partial. By a
comparison of its old constituents, as seen in the Pliocene
deposits of West Europe, with its present constituents, as seen
in the Far East and in North America at the present day, Ave
may gain some idea how the flora has changed.

In the first place we discover that not all species have been
exterminated, in spite of the great lapse of time, and the immense
distances travelled to their present homes. Even if we consider
the older deposits, the Reuverian belonging low down in the
Lower Pliocene, and the Pont-de-Gail at the base of the Pliocene,
we find some species of those remote times still living. As in-
stances we may cite Dulichiuni spathaccuiit, Brascnia pdtata,
Zelkowa keaki. Magnolia kobus, Liriodendron tulipifera, Stexvartia
pseudo-camellia, and Nyssa sylvatica from the large Rcu\-erian
flora, and Vitis lanata and Polygonum Convolvulus, as already
stated, from the Pont-de-Gail flora.

But though some species have remained unchanged, it is far

144 AGE AND AREA FROM A [pt. ii

more common for change to have occurred, and we find that
the greater the lapse of time, the greater proportionately has
been the change. That is to say, more species found in the older
deposits are extinct, than in the newer. This may very clearly
be seen by comparing the percentages of species and varieties,
which there is reason to think are extinct, in the successive
Pliocene floras. There is an element of uncertainty in such a
comparison for this reason. It has not always been possible, for
lack of living material, to discover whether a seed belongs to a
living species or not. The following figures will, I believe, have
at least some approximation to the truth. The deposits read
downwards in order of age ; they are those from which the main
evidence of the facts discussed in this paper were derived.

Percentages of extinct species belonging to the Chinese- North-
American Association of Plants in the West European Pliocene
at successive periods.

Percentage of

extinct species

Deposit

Age of deposit

(approximate)

Cromerian

Top of Pliocene

0

Teplian

Upper Pliocene

35

Castle Eden

Middle Pliocene

44

Reuverian

Lower Pliocene

70

Pont-de-Gail

Base of Pliocene

90

The figures show clearly a progressiA^e extermination of older
forms as compared with newer. In some way age has acted as
an exterminating agent. Hoav? Is it, to use the language of Age
and Area, by dying out, or by killing out?

I take it that, by the use of these terms, Dr Willis intends to
distinguish between extinction of species due to exhaustion of
vitality, and extinction of species due to external agencies. That
is, between internal and external causes. It cannot always be
possible to distinguish between these two, for frequently, as we
know, external causes are assisted in their work of destruction
by pre-disposing conditions in the individual, or it may be in the
race. Such relationships are seen in that of disease to sus-
ceptibility to disease, of change of climate to a Aveakened con-
stitution. It is the external cause which seems the cause of
death, though the internal cause may have an equal share in it.

Viewing the fact of the immense amount of extermination
that has occurred in this flora, I was at first inclined to think
that such destruction of species, one might almost call it uni-

CH. XIV] PALAEOBOTANICAL STANDPOINT 145

versal destruction, must be the effect of dying out. It seemed
that some all-embracing, inevitable, cause — "dying out" — must
be at work, which in time would kill everything; "killing out"
must be more localised and more discriminating. If due to
climate, it might act in one region, if due to disease, upon one
species, but not in regions far apart, or upon so many species.

Yet, with further consideration, the problem appeared
differently. It had to be taken into account that the mere fact
of age means so many more chances of destruction. Therefore
these older forms must have suffered far more vicissitudes, and
have been subjected to far more numerous attacks from exter-
minating agencies than the newer. The lapse of time since the
deposition of the Eocene basalt of Antrim has been the subject
of investigation by Lord Rayleigh, F.R.S. Basing his calculation
on the amount of helium as compared with radium (and hence
of uraniiun) present in haematite iron of that age, he reached
the conclusion that the interval is one of 30 million years.

We must acknowledge that the vicissitudes of 30 million years
are quite beyond the powers of our mind to grasp, and it seems
possible that they may have furnished ample cause, through
disease or other adverse conditions, to bring about all the
destruction to which paleobotany bears witness.

There is very strong evidence to show that the whole European
branch of the Chinese-North-American flora was killed out. by
being subjected to cold which it could not withstand, with no
possibility of escape: and that it perished, trapped between the
cold of the north behind, and an impassable trans-continental
barrier of mountains and seas in front.

What of the other two branches, the living ones? Has there
been extermination of species there? And if so, how has it
occurred? We have already gained some knowledge on this
subject, by comparing some of their species with those of the
European Pliocene — which it must be remembered are nearer
in time to the ancestral forms, even if they be not actually the
ancestral forms— and we find that of the few species so com-
pared somewhere about 90 per cent, have changed.

To discover something more of the changes which have taken
place, we may compare the living members of this flora >vith
what are, without any doubt, the ancestors of some of them.
The description of these is to be found in Prof. Nathorst's
account of the Post-Miocene flora of Mogi in Japan (78-9). This
is the largest of several Post-Miocene floras (mostly very frag-
^ . 10

146 AGE AND AREA FROM A [pt. ii

mentary) examined by him. With regard to the age of the flora,
Dr Kryshtofovich, who has also worked much upon Japanese
fossil plants, believes that it should be assigned to the base of
the Pliocene (63). In that case it is contemporary with the flora
of Pont-de-Gail.

Nathorst remarks of the Mogi flora that the outstanding fact
regarding it is its closeness to the living flora of the forest-belt
of the Japanese mountains; and we have only to consiflt his
lists to see how^ true the statement is. But though the living
flora be close to that of Mogi, it has changed. If we examine
Nathorst's list we shall see that about 44 species may be con-
sidered as belonging to what I have termed the Chinese-North-
American Association; and that of these. 39 are now represented
by different species or varieties. That is. about 89 per cent, have
changed. The figure approximates closely to the 90 per cent, of
changed species found, when we compared the Pont-de-Gail
species with living Chinese-North-American species. I do not
wish to press the similarity of the figures. Whether it be due to
chance, or really represents the degree of extinction of older
forms since the beginning of the Pliocene, I do not know. The
number of species in the Pont-de-Gail flora, belonging to the
Chinese-North- American plant association was small; also Prof.
Nathorst's references to living species suggest that the Japanese
Pliocene forms may be nearer to living forms than those of Pont-
de-Gail would ajipear to be. Anyhow, his work, like mine, bears
evidence that old forms have very largely given place to new.
We may say at once that the work gi\es no evidence as to how
the new forms arose, though the endemic Japanese species which
are related to these old Mogi species would seem to have arisen
in Japan ; but though Ave cannot trace the history of the new, we
can find out a little more about the extinction of the older forms.

If a list be made from all the five European Pliocene floras
(those already named), of all the Chinese-North-x\merican species
which are still living — 26 in all — it will be seen that by far the
greater number are now found living in East Asia. Twenty
species, out of the 26, about 77 per cent., are there found;
5 species, or 19 per cent., in North America; whilst one species
{Brasenia), or 4 per cent., occurs in both continents. Again,
consider the living genera represented— 55 in all — 33 of these,
or 60 per cent., are found living on both continents; 17, or 31 per
cent., in East Asia only; 5, or 9 per cent., in America only.
Therefore, whether we consider genera or species, there would

CH. XIV] PALAEOBOTANICAL STANDPOINT 147

appear to have been greater destruction in America than in
East Asia. Such a conchision is not wholly unexpected, and mav
help to suggest in part the fate of some of the older species. "

In travelling to and fro in latitude in America, the plants
would find themselves mostly in a country of great plains; not
so in East Asia, where successions of north and south mountain
chams existed in those old times, as in the present. It is known
that many of the chmatic changes which occurred during the
Pleistocene were fairly rapid. When such rapid changes happen
m a flat country, it may result that a migrating flora may be
overtaken in its travel by the change of climate, and in that
case many of its components will be exterminated, whereas in
a mountain country, by change of altitude, they may escape.

Such facts may in part account for the lesser survival of
Pliocene species in America than in East Asia. If so, it would
seem that in part extinction in America too was due to killincr
out, ^

Age and Area. Let us now turn to the main theme of the
subject and inquire whether the Chinese-North- American flora
has any evidence to offer as to present distribution in connection
with age. Taking it as a whole, we see an ancient flora associated,
either fossil or living, with the widest possible range in longitude.
Is there any evidence that it shows wide distribution in latitude
associated with age? Such distribution would appear to me,
certainly in the case of a flora Avith a dispersal originating in
polar regions, to be a far more crucial test of age than dispe*-sal
in longitude, for with a circumpolar flora as a source of dispersal,
with a cooling climate, spread would be equatorwards through
all possible regions. The distance of travel would therefore be
measured, not by span in direction of longitude, but by travel
in latitude. In the case of the Chinese-North-American flora
there is some evidence that travel in latitude is an accompani-
ment of age. It is too small to be estimated quantitatively, but
is seen in the presence, especially among the older Pliocene
floras (Reuverian and Pont-de-Gail), of such genera as Hakea,
Symplocos, Styrax, Pulanisia, and Trichomnthes, which have a
present distribution into the southern hemisphere. The evidence
as to the source and direction of migration of the Chinese-North-
American flora to which they seem to belong indicates that these
genera too are migrants from the north, and that their present
distribution in latitude is partly due to age.

10—2

CHAPTER XV

ENDEMISM AND DISTRIBUTION: SPECIES

1 HE term endemic has long been used to connote a species,
genus, or other group confined to a small area, such especially as
a single island, a group of islands, a mountain chain, or a com-
paratively small country like South Africa or ^Vest Australia,
largely bounded by the sea or by a marked alteration of climate.
In recent years species of larger areas have been spoken of as
endemic, but the term is used in an arbitrary way, for one
speaks of species as endemic to Australia, though not to Brazil,
which really has far more of them (533 endemic genera, perhaps
12,000 endemic species).

There is almost never any real and demonstrable difference
between species and genera of small and of large area, other
than in the territory occupied, but since the rise of natural
selection it has been generally assumed that such a difference
really occurs. On that theory one will expect to find many
species "going under" in the struggle for existence, and the fact
that so many are actually localised to small areas of territory,
particularly in somewhat isolated regions of the globe, provides
the necessary material for this explanation to rest upon. Botanists
have long been accustomed to look upon endemic forms as the
oldest, and very often as in some way expressly suited to the
very local conditions in which they occur. This latter must of
course be true for any species, anywhere, or it would be exter-
minated in a short time; but the study of detail which has re-
sulted in the putting forward of the hypothesis of Age and Area
gives reason to believe that in general the supposition of greater
age of endemics is incorrect.

As endemics usually occur in somewhat isolated places or
countries, the question at once arises whether endemism is corre-
lated with isolation as such, for if so, the fact will have an im-
portant bearing on the question of evolution generally. There is
also some ground, however, for supposing that the soil in isolated
regions may be less completely taken up b}^ its associations of
plants, so that a newcomer would have a better chance of sur-
vival; and this may be the explanation.

From about 48° N., to the southwards, all important islands.

PT. II, CH. XV] ENDEMISM AND DISTRIBUTION 149

and mountain chains (if over 4500 feet), besides all more or less
isolated pieces of country, like Italy, possess endemics. They are
also frequent in such localities, even in large areas of country
with large populations of plants, as are isolated in the sense that
they do not lend themselves to free interchange of plants with
their surroundings. Such are stations in large forests, or patches
of grassland in forest country, patches of country with salt soil,
and the like. The numbers and proportions increase to the south-
wards (and the isolation becomes less marked), till one finds the
maxima in such places as ^yest Australia, South Africa, Juan
Fernandez, the Mascarene Islands and New Caledonia. Beyond
about 40° to 48° S. they fall off again. "The greatest concen-
tration of species in small areas occurs in... West Australia and
South Africa" (52, p. 36). "The fertile portions of New South
Wales, Victoria, South Australia, and ^Vest Australia do not
probably... exceed in area Spain, Italy, Greece, and European
Turkey, and contain perhaps half as many more flowering
plants" (55 b, p. xxxi).

Endemism, though it is most commonly associated with
islands in people's minds, is by no means a phenomenon con-
fined to them. It is very strongly marked in comparatively
isolated mountains, such as Kilimandjaro (and cf. 117 and 122),
and in mountain chains, and in these cases the flora presents, as
a general rule, less relation to that of the plains than does the
flora of an island to that of the nearest mainland^. This may be
largely due to the fact that, as explained upon p. 37, mountains
may act as highways of migration for the plants of other coun-
tries and climates.

Endemism is also strongly marked upon continental areas,
and while the maximum proportion is in West Australia and
South Africa — regions where conditions are rather extreme — all
the southern land masses, more especially, show a great pro-
portion of their species confined to themselves.

Whilst the largest numbers and jDroportions of endemics are
chiefly in the more southern countries, there are also large
numbers and proportions in several of the northern, e.g. in
Mongolia, California, the region about the Mediterranean Sea,
etc. There are a few endemics on the west coast of Europe, the
Alps contain about 200, and Italy about the same; and the

1 "A great deal too much has been made of the assumed extreme differen-
tiation exhibited by insular floras as compared to the continental flora"
(52, p. 387).

150 ENDEMISM AND DISTRIBUTION: SPECIES [pt. ii

Iberian peninsula contains about 800, or roughly the same as
Ceylon, which, however, has only one-ninth of the area. The
really large numbers are south of the tropic of Cancer. The
Hawaiian Islands have 600, Ceylon 800, New Zealand over 1000,
Australia about 7500, Mexico and Central America about 8000,
and Brazil perhaps 12,000. They are especially common in
momitainous country, and it is worth noting that most islands
are also mountainous.

No country or island has all its species endemic, though in
several or most places where there is a very large proportion of
endemic species, like Hawaii or New Zealand, it is very common
to find genera with all their species endemic (cf. reply to objec-
tion 28, p. 95). St Helena, with a very small flora, seems to
have perhaps the highest proportion of endemic species, but of
countries with any large number, West Australia, with 85 per
cent, of its species endemic, takes the first place. The Hawaiian
Islands, with 82 per cent., are close behind. New Zealand (37;
has 72 per cent., the Galapagos 46 per cent., the Bahamas 14 per
cent., thus illustrating the fact that on the whole the further out
and more isolated an island is, the greater is its proportion of
endemic species. Fiji and Tahiti have much smaller proportions
than the Hawaiian Islands, but Fiji, with 50 per cent., is much
nearer to the mainland than Tahiti with 35 per cent., so that this
alone is not sufficient explanation. Nearly half the ferns and
lycopods in the Hawaiian Islands are peculiar to the group, in
Fiji and Tahiti only about 8-9 per cent.

A study of the areas occupied by endemic sjDeeies soon shows
that the}- may be of any size from a few square j' ards upwards,
and that there is no difference to be seen between them and
species tliat are not usually considered endemic, and which may
have areas of larger and larger size, up to one of a large portion
of the globe. It was these extraordinary differences in area
occupied, between species closely resembling one another, and
differing only in characters which could not, by any stretch of
imagination, be looked upon as fitting or unfitting them in any
way for the struggle for existence, that first caused me to begin
studj^ing areas, and searching for some more potent agent in
distribution than adaptation, a search which ultimately led me
to Age and Area.

This new point of view, that the mere area occupied by a
species has some more definite immediate interest than simph^ as
an expression of some unknown character in the protoplasm, or

CH. XV] ENDEMISM AND DISTRIBUTION: SPECIES 151

some unsuspected property in apparently meaningless external
characters, receives great support when the actual areas upon
which species occur in any country are mapped out by drawing
lines round their outermost locations. We shall begin with very
localised endemics.

Mr H. N. Ridley (90, p. 555) found two plants of Didy mo-
carpus Perdita Ridl. "on a bank in the centre of Singapore, sur-
rounded by extensive cultivation. It has never been seen again."
Dr Thwaites (37) found in the forest at Hakgala in Ceylon a few
plants of Christisonia albida Thw. (C, P. 3929). This differed
from its nearest relative, C. bicolor Gardn., in having the scales
of the scape ovate and glabrous, instead of oblong-obtuse and
pubescent; the bractlets below the flower instead of near the
base of the peduncle ; calyx glabrous instead of pubescent, with
linear instead of triangular segments; and the corolla larger.
Taken together with the fact that the whole plant was white,
instead of the brownish colour usual in the Orobanchaceae, these
differences were so large that the species Avas regarded as a
Linnean species, and accepted as such in the Flora of British
India, iv, p. 323. The plant has never been seen again, though
the area of forest at Hakgala which could be reached by the
invalid Dr Thwaites is very limited, and there is a botanic
garden beside it, in which many botanists have worked, search-
ing the forest thoroughly. Probably in both the cases just men-
tioned, the taking of a few specimens was sufficient to exter-
minate the species; and in the latter case, it is probable that the
Avhitc colour alone would have been such a disadvantage as to
ensure its extermination by nature in any event.

The next stage may be seen in such a case as that of Coleus
elongatus Trim., endemic only to the smumit of Ritigala in
Ceylon (p. 14). It occurs as about a dozen or two of plants upon
open rocky places at the very simimit, and differs so much from
other Colei that it is a very distinct Linnean species, even if not
subgenerieally separate. Its nearest relative is C barbatus,
Avhich also occurs on the summit, as well as in tropical Asia and
Africa. The distinctive characters inay be tabulated as on the
following page.

It is all but impossible to imagine that any of these characters,
and especially the two most important, the peculiar inflorescence
and calyx, have any serious effect upon the capacity of the
species to survive or progress, or that any of them can be
seriously disadvantageous. It is worth while in this connection

152 ENDEMISM AND DISTRIBUTION: SPECIES [pt. ii

C. barbatus C. elongatus

(Bot. Mag. T. 2318) (Fig. in Trimen's Ceylon Flora,

T. 74)

Stem cylindrical, tending to quad- Stem quadrangular

rangular in inflorescence

Stem pubescent with long hair Stem pubescent with short hair

Leaves oblong-oval, 1-2 inches Leaves ovate-triangular, 1-2 inches

Leaves closely pubescent Leaves finely pubescent

Leaves rather thick Leaves rather thin

Petioles rather short Rather longer and slenderer

Inflorescence of condensed cymes, Inflorescence of one-sided cymes,

each about 5-flowered, forming looking like racemes, about li

false whorls of 10 flowers at each inches long, one at each side of

node each node

Flowers large Flowers small

Bracts large Bracts small

Calyx with long hairs Calyx with short hairs

Calyx of one large ovate upper tooth Calyx of five almost exactly equal

and four small lower teeth

Corolla rich purple or white Corolla pale purple

Grows on rocky places Trails over rocks

to look at the distribution of the other Ceylon Colei, already-
described on p. 54. There is no such difference in the method of
dispersal as will account for the great differences in area occupied,
nor is there any difference in the other characters of the plants
that one can point to, as advantageous or disadvantageous.

A somewhat larger area than that of Colens elotigatus is that
occupied by Campanula Vidalii, which is found (47, p. 427) on
rocks near the sea on Flores and two other islands of the Azores.
A still more interesting case is Cenchrus insularis, which is found
only on one islet of the Alacran reef (75), about thirty miles off
the coast of Yucatan, while Cakile alacranensis and Tribulus
alacranensis are found on all the four islets of the reef, the largest
being less than half a mile long, and very narrow. There seems
some reason to imagine that the evolution of these species has
been fairly rapid, as they were not noticed by the Admiralty
expedition that visited the islands fifty-seven years previously.
And scores of similar cases of distribution might be cited.

We may go on to deal with genera containing several species
in the same neighbourhood, all or most of them endemic, gi\'ing
a few actual instances. Doona, for exainple, a Ceylon endemic
genus of Dipterocarpaceae, has 11 species, whose local distribu-
tion (fig. on p. 153) is typical of that of many local genera, or
genera with many species (mostly endemic) in one locality.
The whole range of the genus (about 4000 square miles in south-

CH. XV] ENDEMISM AND DISTRIBUTION: SPECIES 153

west Ceylon) is occupied by one of its species, D. zeylanica, while
the others occupy smaller and smaller areas within this, down
to a comparatively few square miles. This is perhaps the most
common type of distribution with genera of small area, which
upon the theory of Age and Area are to be regarded as young
beginners. Another instance is Haastia in New Zealand (fin-, on
p. 154).

Distribution of the same type, but more extended, is shown
by the (chiefly endemic) species of Ranunculm in New Zealand
(fig. on p. 156), and by very many other genera in that coun-
try. In this map the widely distributed species, i.e. those occur-
ring outside of New Zealand, are shown bv dotted lines, and it

DOONA,

will be noticed that three of them range all over New Zealand
(including the little Stewart Island to the south), and also to the
Chathams, 375 miles to the eastward, while the fourth only
ranges from the far south up to the middle of North Island. The
endemics all have ranges within that of the first three wides,
among which probably, upon the general implications of Age
and Area, one must prineipally look for their parent or parents.
The endemic with the greatest range covers slightl}' more ground
than the wide of least range, and the others occupy smaller and
smaller areas, becoming steadily more numerous in going south,
till a maximum is reached a little south of the middle of South
Island, as indicated in the following figures (cf. p. 77), which

154 ENDEMISM AND DISTRIBUTION: SPECIES [pt. ii

Diagram showing the areas occupied by tiie species of Ilaastia in

New Zealand.

(By courtesy of the Editor, Annals of Botany.)

CH. XV] ENDEMISM AND DISTRIBUTION: SPECIES 155

show the numbers that occur in each zone of 100 miles from
north to south in the two large islands :

Wides 33 3 4444444

Endemics - 2 3 5 7 11 12 18 18 10

If instead of taking the distribution in this way by zoning, one
take the actual longitudinal range of the different species, one
finds that of the 28 endemic Ranunculi, 10 have a range not
exceeding 60 miles longitudinally in New Zealand, while of
ranges 120, 180, 240, etc., there are only 1, 3, 1, 2, 4, 1, 1, 1, 2,
0, 1, 0, 1. The great bulk are obviously crowded towards the
short ranges. If one make five groups, occupying ranges from
0-200, 200-400, 400-600, 600-800, 800-1000 miles, one finds
that they contain 14, 7, 5, 1 and 1 species respectively. If, now,
one plot these figures in a curve (fig. on p. 162, curve 7),
one obtains a curve which is concave upwards, or what we may
term a hoUoiv curve. This type of curve we shall presently see to
be almost universal in distribution — and it proves of late to be
equally so in evolution itself. At first, perhaps, its presence will
not be readily noticed, but when one finds the figures for any
example of distribution or evolution showing a great accumula-
tion at one end, and the first two or three descending very rapidly,
while the remainder tend to taper away gradually, one will
generally find this type of curve shown, on actually plotting the
figures. It shows very strikingly in many of the examples
described below, e.g. the distribution of the Hawaiian endemic
species of CijHandra described on p. 160, (same fig. curve 6).

Or one may take such a genus as Epilohium in New Zealand
(37, p. 171). E. jmrpiiratum is confined to the Alps of Otago,
4000-6000 feet, E. brcvipes to the northern half of South Island,
E. crassum to the greater part of the length of South Island;
E. melanocaulon ranges the whole length of South and the southern
half of North Island, E. microphyllu7n ranges yet farther north,
E. glahellum farther again, while E. rotundifoUum ranges the
M-hole length of both islands, and reaches SteA^art and the
Chathams. E. nummulari folium reaches all this, and also Auck-
land and ]\Iacquarie Islands to the south, while E. jmllidiflorum
ranges this and reaches Australia and Tasmania. Tins, or some-
thing like it, is the common type of distribution in New Zealand.

If we take a genus— and there are many— that has no wides
in New Zealand at all (cf. p. 95), we find the same thing shown,
as, for example, in Gunnera (fig. on p. 158). Here there is one

156 ENDEMISM AND DISTRIBUTION: SPECIES [pt. ii

Diagram showing the areas occupied by species f /""""^"^^^^^^

New Zealand. Wides dotted; extension East uiehides Chathams.

(By courtesy of the Editor, Annals of Botany.)

cii. XV] ENDEMISM AND DISTRIBUTION: SPECIES 157

endemic species that covers all New Zealand, and reaches the
Chathams, and the other endemics occupy smaller and smaller
areas within this. The figures by zones show:

2235556665

a result exactly similar to that for Ranunculus. The presence of
wides does not seem in any way necessary, nor to cause the
species of a genus to behave in any way differently.

If we go to Ceylon, and take a few species of the pan-tropical
genus Eugenia, of which Ceylon has 29 species endemic to the
island and 14 found elsewhere (6 only in southern India), we
find that E. cyclophylla occurs only on Adam's Peak, E. lu^ida
on several peaks close together, E. sclerophylla on a number of
peaks and in the plains between, E. assimilis throughout the
mountains and in the moist plains, E. hemispherica in all this
and also in South India, and E. operculata in these regions, and
also in Burma, Malaya, and China. And many other genera
show the same type of dispersal, which, in fact, a little study
soon shows to be the usual type. If one go to the state of Rio
de Janeiro in South Brazil, which has an area about equal to
Ceylon, one finds 52 Eugenias endemic to the state (which is
very mountainous), and 6 going beyond it, 3 only into Minas,
the next state, the other three as far as the states of Alagoas
(1000 miles north along the coastal plain), Rio Grande do Sul
(the same south) and Goyaz (600 miles inland, across the moun-
tains). And one may find Eugenia behaving in the same manner
in many other places. In Brazil it has many endemic species in
Minas, the next state to Rio, but on the other and dripr side
of the mountains that fringe the coast.

This general type of distribution shows very clearly in the
case of very many genera, whether they be endemic genera with
all their species in a confined area, like Doona in Ceylon, or
whether they be genera of wide distribution that have developed
many endemic species within a certain small area, like Ranun-
culus in New Zealand. In such cases they do not seem as yet to
have encountered any barriers of a very serious kind. But one
may also find a great* number of genera, or sections of genera, in
which the same thing is displayed over a very much larger area
than what would entitle the contained species to be considered
endemic. In Callitris, for example, C. glauca occupies the whole
range of the genus over Australia and Tasmania (130); two
others range from New South Wales to Tasmania and to West

158 ENDEMISM AND DISTRIBUTION: SPECIES [pt. ii

Diagram showing the areas occupied by the species of

Gunnera in New Zealand.

(By courtesy of the Editor, Annals of Botairy.)

CH. XV] ENDEMISM AND DISTRIBUTION: SPECIES 159

Australia, while the remaining 15 species of the genus cover
smaller ranges. In Dilleiiia, D. indica covers practically the
range of the genus throughout Indo-Malaya, while there are
many other species covering smaller and smaller ranges within
this. In Gymnema, G. sylvestre covers almost the whole range of
the genus from West Africa to Australia, whilst in Cissampelos,
€. Pareira is found from tropical America through Africa and
Asia to the Philippines, almost covering the whole range of this
cosmotropical genus. In Najas, finally, N. marina is cosmo-
politan, Avhile five other species occupy very large areas, nine
occupy areas of moderate size, and seventeen areas of small size.

Another very frequent case in endemic genera of small area,
or in genera with a number of endemic species within a small
area, is to have one species occupying a "circle" of some size,
and another a (usually) smaller circle touching, or near to, the
first, thus giving the impression that the plants occupying it
have possibly sprung from some unusually isolated members of
the first species, in the case of an endemic genus which has no
species covering the range of both. In Ceylon, for example, in
the endemic genus Horto?ua, which has three species, H. angusti-
folia occurs in the moist plains, and to 2000 feet in the moun-
tains, while //. florihunda occurs only in the mountains above
4000 feet, and H. ovalifolia is confined to Adam's Peak. Or in
the Ceylon endemic genus Schumacheria, S. castaneaefoUa is
common to a height of 1000 feet, S. alni folia above that level,
and S. angustifolia occupies a tiny circle within the area of the
first named, but a long way from S. alnifolia.

This type of distribution, in smaller circles, usually over-
lapping one another to a greater or less extent, Avhile there is no
single one covering the whole range, is also very common. To
take an example at random from the Indian flora, Christisonia
(37, IV, p. 323) has three species in Ceylon, one frequent in the
hills, two confined each to one spot (cf. p. 151), three in the
Dekkan or the Konkan, three in both Ceylon and South India,
and one in Sikkim and the Khasias. None has individually a
very large range, yet the genus covers much ground, and there
is some overlapping of species. One may see the same type of
distribution upon a fairly large scale by taking such a genus as
Cyrtandra, whose species are distributed as follows (25, v, i):

160 ENDEMISM AND DISTRIBUTION: SPECIES [pt. ii

No.

Java, Sumatra, Singapore to Tenasserim 55

Sumatra, Penang, Borneo, Celebes, Amboina, Papua ... 56

Java, Borneo, Philippines ^

Java, Celebes, Timor 10"*

Java, Celebes, Ternate 25

.Java, Sumatra . . • 2,6,17,24,77,79,84,87,110,113,116

Sumatra 4, 8, 14, 18, 57, 58, 73, 74, 75, 78, 82, 85, 86, 88, 90, 91, 114, 163-5

Sumatra and Penang or Malacca 1,76

Java 19,20,21,23,71,72,107,108,109,118

.Java, Singapore, and Celebes, or Ternate ... 83, 106, 115

Borneo 3, 7, 9, 10, 11, 12, 13, 16, 22. 64, 66-70, 80-1, 99-102, 112, 117, 160

Celebes 15,25,65

Moluccas, Ternate ^^l

Ceram (Moluccas) l*^*

Philippines 92,93,95,119,158,159

China ^f

New Guinea 89,97,98,10.3

Carolines

New Hebrides .

96
125, 157
Fiii Islands . . 51-4, 59, 124, 128-9, 132-3, 139-40, 151-6, 161-2

Samoas .... 127, 130-1, 134-5, 141, 149-50, 150, 166-7

Societies, Low Archipelago . . . • - • • .13/

Societies 120-22, 126, 136, 138, 146-8

Sandwich Islands 27-50,60-3,123,142-5

The whole range of the genus is from Tenasserim to the Sandwich
Islands, yet no single species reaches half this distance. Most
have very small ranges, e.g. most of those upon Java or Sumatra,
which are usually confined to portions of these islands, but there
are a fair number, e.g. those in lines 3, 4, 5 and 6, which have
rather large, and two at the top Avith very large, ranges (cf. fig.
on p. 162, curve 1).

One may even follow them into more minute detail, for ex-
ample, in the Pacific Archipelagoes. In the Sandwich Islands,
1 species occupies four islands, 2 occur on three, 2 on two, whilst
there are 24 on single islands, viz. 11 on Oahu, 4 each upon
Kauai and Maui, 3 upon :\Iolokai, and 2 upon Hawaii. The same
thing may be seen upon the Samoan and other islands; this
"hollow c\m'e" type of distribution is general, as we shall see
below (cf. fig. on p. 162, curve 6).

One may follow this type of distribution into the small
varieties of Linnean species, to which specific rank is often given
by local botanists. For example, dipping into Linton's British
Hieracia, and taking the section Nigrescentia, one finds 4 species
occupying six to ten counties, and 9 in one to five, but no single
one covering all the range. These two cases, (1) that there is one
or a few widely ranging species with larger and larger numbers

CH. XV] ENDEMISM AND DISTRIBUTION: SPECIES 161

of more and more localised species scattered about within or
close to their range, and (2) that there are many species of
local range, usually more or less overlapping one another, and
themselves overlapped in many places by fewer species of
rather wider range, and in the total occupying considerable
areas, which are as continuous as intrusions of the sea and
other barriers will allow, seem to cover the case of the bulk of
existing genera. The latter case also makes, though not so
strikingly, a hollow curve, for there are more species of small
areas.

It is clear that the types of distribution shown by endemic
species, whether of endemic genera or not, are the same types
that one may see in the dispersal of genera, species, and varieties
of wider range; there is no place at which one can draw a line,
and say that here is the distinction between endemic and non-
endemic species.

But the resemblances between endemic and non-endemic
species may be carried much further. In the case of the former,
as Ave have seen above, their usual grouping in a country shows
a few in the class containing those of widest local dispersal, and
larger and larger numbers as one goes down the scale to the more
localised classes. And this grouping shows, not only for the
grand total, but for the individual families and larger genera.
The actual figures for New Zealand show that the curve so pro-
duced is a hollow one (fig. on p. 162, curve 5). The peak in
the middle of the curve is accounted for, perhaps, by the opening
of Cook's Strait having checked the dispersal of some of the
species (127, p. 455). If, dipping at random into the New
Zealand flora, one take the Boraginaceae, and divide the en-
demics into five classes, one finds 2/1 (two in class 1), 1/2, 2/3,
5/4, 13/5; or if one take Olearia, one finds 2/1, 5/2, 4/3, 6/4, 14/5.
Always the same type of curve is formed, with an accumulation
of species at one end.

But this same phenomenon shows in the case of all other
species, whether endemic or not. In Doona in Ceylon, for ex-
ample, one finds one species of large area, three of smaller, and
seven of areas smaller yet. With the largest endemic genus in
the Hawaiian Islands, Cyanea, one finds (ef. Cyrtandra above)
one species on four islands, six on two, and 21 on a single island.
Pelea, the next largest genus, shows 1/8 (one species on all
islands), 3/4, 3/3, 2/2, 11/1, again a hollow curve, running out
very much at one end. If one add up all the species of the

W.A. 11

162 ENDEMISM AND DISTRIBUTION: SPECIES Fpt. n

LHDLniC 5?LCtE5 OF IsLAMDS . ETC.
3lj areis [rom small to l^rge.
The(lj rnflicat'es Unit artu.

ArcA occuputd
Endemic species by area from small to large

CH. XV] ENDEMISM AND DISTRIBUTION: SPECIES 163

endemic genera of the Sandwich Islands, and compare them with
the endemic species in the non-endemic genera, one gets :

Table showing the numbers and proportions of species of endemic
and of non-endemic genera that occur on all the Hawaiian Islands,
or 071 four to six islands, etc. Thus on all islands there occur 3-1 per
cent, of the species of endemic, 9-5 per cent, of the endemic species
of non-endemic genera.

Species of Endemic species of

endemic genera non-endemic genera

r

..^

>ccurring

on

species

o/

/o

All islands

7

31

4-6

23

10-2

3

25

111

2

41

18-2

1 island

129

57-2

225

99-8

Average

dispersal 1-8 isl

lands

species

/o

34

9-5

51

14-3

55

15-4

72

20-2

144

40-4

356

99-8

2-3 islands

Thus the species of the endemic genera are dispersed on the
average some 25 per cent, less than those of the non-endemic
genera. This proves on examination to be a general rule, and is
a powerful argument against local adaptation. It shows with
equal clearness in New Zealand and in Ceylon (123. p. 324.).

If one go on to Ranunculus in New Zealand, which has " wides "
as well as endemics, while Olearia (above) has not, one finds 1/1,
1/2, 5/3, 7/4, and 14/5, taking only five classes instead of ten.
If one take at random in Vol. iv of Hooker's Indian Flora a few
genera, one finds in Exacum one species with large, six with inter-
mediate, and nine with small areas. In Chrisiisonia, where there
is no single widely ranging species, there are 4 with moderate
areas, and 6 with small. In Ebermaiera 1 has a very large, 6 an
intermediate, and 21 a small area. Or, finally, take Cyrtandra
(above); there are, roughly, 2 with very large areas, about 20
with fairly large, and about 145 with small, these latter again
showing gradations down to the smallest, as we have just seen
(p. 160).

It is clear that the distribution of endemics is only a sppoi.nl
case of a vdde general phenomenon — that there are, in any family
or genus of reasonable size, a few species of wide dispersal, and
others of less and less dispersal in increasing numbers, the in-
crease being more rapid as one descends the scale, so that the
curve produced is hollow. When, as in very many genera, there

11—2

164 ENDEMISM AND DISTRIBUTION: SPECIES [pt. ii

is one species covering the whole range of dispersal, the classifica-
tion can be carried into greater detail, but even in such cases as
Cyrtandra, where there is not such a species, the phenomenon
can be quite clearly seen. It is evidently perfectly general, and
we shall see many further examples of it in the next chapter,
and go on to consider its general bearings in later chapters.

When one goes on to examine into the genera and families to
which endemic species chiefly belong, one discovers that in most
countries the bulk of the endemic species do not belong to the
endemic genera. Even in a region of such marked endemism as
the Hawaiian Islands (37), for example, where there are several
very large endemic genera, only 225 out of 581 endemic species
belong to the endemic genera, or 38 per cent. In New Zealand
less than 5 per cent, do so, and in Brazil perhaps 10 per cent.

The numerous endemic species that do not belong to endemic
genera are found on examination to belong, not, as one might
perhaps expect, to small and broken genera, Avhich we have been
accustomed to consider moribund, but in greater proportion to
the larger and more important genera. The average number of
species in a genus, taking the whole world, is about 12-7, and in
the Hawaiian Islands, taking the first hundred genera in the
flora (37), we find that of the 47 that contain endemics, but are
themselves widely dispersed, 36 are above the average size in
the world, and have 102 local endemics, while 11 are below, and
have 22 endemics. Of these 11 belong to Lipochaeta, which only
occurs outside these islands as a single species in the Galapagos.
The average size of the whole 47 genera (in the world) is 97
species, or eight times the average. Of these genera 8 are cosmo-
politan in their dispersal, 11 are tropical and subtropical, 8 are
tropical, these three categories including 57 per cent, of the total
(cf. Chapter xii, Size and Space). A further 9, bringing the total
to 76 per cent., occur in both Old and New AVorlds.

If we turn to New Zealand, and take the first 100 genera (37),
of those with endemics 43 are above the average in the world,
and only 14 below, while the average world-size of one of these
57 is 73 species, or six times the world-average. The same thing
shows Avherever I have tested it. For example, if one take the
first 100 genera in Vol. iv of the Indian Flora (37), most of them
as it happens being Asclepiads, which are imusually small genera,
one finds 52 non-endemic genera, of which 38 are above, and
14 below, the average world-size. The remaining 48 are largely
endemic genera, for India, like all large areas, has a greater pro-

CH. XV] ENDEMISM AND DISTRIBUTION: SPECIES 165

portion of its genera endemic than have the outlying islands,
etc. The average size of all 52 is 52 species, or still much larger
than that for the world, though less than for New Zealand.

The further out one goes from the centres of greatest massing
of genera and species, in other words, the larger on the average
(in size in the world) do the non-endemic genera become. The
genera above mentioned in the Hawaiian Islands, New Zealand,
and India, that are below the average world size, are in all 39
with 89 endemics, while those above are 117 with 688. This fits
in with what was said above (p. 115) about Size and Space, that
on the whole the larger the genera, the larger the area they
occupy.

This fact, that the endemic species, in all regions of the world,
belong in greater proportion, not to the small and local genera,
but to the large and widespread, is one of the most striking
features that spring to attention when one begins to study
endemism. In New Zealand, for example (37), the genera that
have most endemic species are Ranunculus'^ (with 32), Epilohiuni
(24), Coprosma'^ (40), Olearia^ (35), Celmisia^ (42), Senecio^ (29),
Myosotis (21), Veronica^ (81), Carex'^ (36) and Poa^ (21), a fairly
well-known list of genera. These ten contain no less than 36 per
cent, of the endemics of New Zealand. Or in Ceylon, the largest
numbers of endemics are in Eugenia (29), Memecylon (21), He-
dyotis (16), Symplocos (17) and Strobilanthes (25), again not
altogether unknown genera, the five containing 13 per cent, of
the endemics of the island. And if one study the endemic or
local species of the world, one finds these same genera appearing
in many other places with large numbers of local species; Eu-
genia, for instance, has about 240 in Brazil (52 in the little state
of Rio). If one adopt the explanation of dying out, these great
genera must have become world-wide very early, and have left
all these endemics as stragglers, dying out before the advancing
host of those species which had proved the best adapted to the
conditions.

The view to which all this leads is simply, as has already been
mentioned (p. 61), that in the vast majority of cases endemic
species are young species comparatively recently evolved, and
still in the earlier stages of their distribution about the globe,
Mhile they show no points of distinction from species of larger

^ These genera also oeciir with endemic representatives on the outlying
islands (Kermadecs, Chathams, Aucklands), where they have 27 out of the
grand total of 73 endemics of these islands, or 37 per cent.

166 ENDEMISM AND DISTRIBUTION: SPECIES [pt. ii

area, being distributed upon exactly similar principles, and like
them showing many of small area, with numbers diminishing at
first rapidly, and then more slowly, towards the few that occupy
large areas, the effect of the figures, when plotted graphically,
being to form a hollow curve (fig. on p. 162).

One may almost regard the question of endemism as the central
point of taxonomic distribution, upon which all the rest depends.
Controversy has largely centred around it, and there are at least
three rival explanations in the field at the present time. These
are (1) that endemics are very specialised species (and genera)
suited only to the areas upon which they are found; (2) that
they are old species (and genera) which have been driven into
quiet nooks, or left in odd corners, by the competition of better
adapted species; and (3) the explanation just given, that in
general they are young beginners, descended from the "wides."

The first and second explanations were based upon incom-
plete knowledge of the distribution of endemics, and can no
longer be regarded as general. One has only to think over what
has been pointed out above (and cf. p. 55). The facts (1) that
the endemics are distributed in "wheels within wheels" (cf.
maps given above), (2) that the numbers in any genus in a
country increase from the edge up to a maximum at some point
or region, (3) that this is the same place at which many other
genera have also their maxima, (4) that there may be more than
one place in a single country (p. 78) where these maxima aggre-
gate together, (5) that the distribution of the endemics by areas
forms hollow curves, increasing most rapidly to the smallest
areas of all, (6) that these hollow^ c\irves show for country by
country, for family by family, e^en for genus by genus, (7) that
there is no difference in type of distribution between the species
of endemic genera, those of Avidely distributed genera with all
species endemic, and those of widely distributed genera with
some species endemic and some not, (8) that the species of
endemic genera show less dispersal in a country than the endemic
species of non-endemic genera, (9) that the endemic species
mainly belong, not to the endemic genera, or to small and broken
genera, but to the large, widely distributed, and "successful"
genera of the Avorld, (10) that endemic species are distributed,
and behave, just like other species, (11) that endemics increase
in numbers and proportion towards the south; to say nothing
of other facts already brought up, or of the difficulties in explain-
ing in any single case what characters are disadvantageous (as

CH. XV] ENDEMISM AND DISTRIBUTION: SPECIES 167

required for dying out), or advantageous (as required for local
adaptation), these facts, we repeat, are very much against any
explanation that is based, as are the two first named, upon
natural selection. Further, upon these suppositions it is impos-
sible to make any of the predictions that have already been so
successful!)^ made.

There remains the third hypothesis, that in general endemics
are species so young that they have not yet had time to spread
to any great extent, or in other words that they are in general the
most recent appearances of species in the genera to which they
belong. Only in some such way can one explain the appearance
of such maps as those given above for Doona or Ranunculus, or
the "hollow curves" of distribution. No valid evidence has yet
been brought up to show that this is not the correct view to take
of the existence of the majority of endemics. There can be little
doubt, however, that quite an appreciable number of existing
species must be looked upon either as relics, or as local adapta*^
tions. The relics may or may not be dying out (cf. rephes to
objections, pp. 88 to 94). The local adaptations must, of course,
be looked upon as simply a special case, i.e. as species which
appeared at first (as all species, to survive at all, must do) as
eminently suited to the local conditions that obtained at their
birthplace, but which have not been able to spread far, by reason
of ecological boundaries caused by changes of conditions at a
very short distance.

There are many points in fa^'our of tliis third hypothesis. It
explains as well as the other two all the phenomena that they
were able to account for, and also very many to which they were
quite inapplicable, as, for example, the eleven given on p. 166.
It also enables us to make predictions about distribution, Avhich
an examination of the facts shows to be justified, and it has
already been successfully employed in this way nearly a hundred
times. Under these circumstances, Age and Area may perhaps
be regarded as at any rate possessing a greater basis of probability
than either of the two hypotheses based upon natural selection.

Summary

It is shown that no real difference can be pointed out between
endemic and non-endemic species (or genera). The former are
frequent upon mountains, upon islands, and in isolated pieces
of country, or in regions in which dispersal is very slow, or

168 ENDEMISM AND DISTRIBUTION [pt. ii, ch. xv

hindered by surrounding barriers. Instances are given of the
space occupied by endemics, beginning with very minute areas,
and going on to larger; the latter show no break as one goes on
to areas larger again, up to any size possible for a species. No
difference can be seen between endemic and non-endemics.

It is shown that endemics are distributed in "wheels within
wheels" (cf. maps); and that other features obtain in their dis-
tribution, of which a brief list is given on p. 166. None of these,
or but few, can be explained on the supposition that endemics
are local adaptations, or are relics, and the only possible explana-
tion, for the vast majority, seem to be that provided by Age and
Area and Size and Space, that, in general, they are young
beginners, descended from the " wides."

The most important general featiu-e in the distribution of
endemics is probably that it is always of the "hollow curve"
type (fig. on p. 162) with most species on the small areas,
and numbers rapidly decreasing upwards to the large. This same
type of distribution proves to be the rule for all genera, however
large they may be, and however large an area they may occupy.
Endemics simply present a miniature of the general distribution
in the world.

CHAPTER XVI

ENDEMISM AND DISTRIBUTION: GENERA

We have seen that endemic species are especially common
upon islands, upon mountain chains, and in more or less isolated
localities (small or large), and that in all such regions they in-
crease, on the whole, in passing from north to south, up to a
certain limit. We have also seen that it is probable that the great
bulk of them must be regarded as young beginners. But if this
be so, there is no logical reason why the same should not be true
of endemic genera, which occur in similar places, and there is
every probability in its favour. Of course, just as in the case of
species, there are doubtless many exceptions here and there, but
we are speaking of the genera in the bulk.

When the number and proportion of endemic species is large,
there are generally to be found a fair number of endemic genera
also, but there seems no necessary relation between number of
species and number of genera; or perhaps rather, this relation
may be much interfered with by other causes. The Hawaiian
Islands have more endemic genera than Ceylon or New Zealand,
though they have many fewer endemic species ; on the other hand,
they are more isolated. This matter still requires more careful
investigation.

The number of genera confined to islands or mountain chains
seems to increase with at least three factors — with the size of
the island or mountain chain, with the isolation of the same, and
with increased southern latitude, up to 45-50° S. The effects of
all these factors may be seen in the list below, by comparing, for
example, Ceylon and Java, Ceylon and the Hawaiian Islands,
and Ceylon and New Caledonia (which is much smaller).

The greatest proportion of endemic genera to area is to be
found in some of the southern and comparatively isolated loca-
tions, e.g. in the islands of Juan Fernandez, the Mascarenes, or
NcAv Caledonia, in south-west South Africa, in parts of West
Australia, etc. But the actual numbers of endemic genera in-
crease with increasing area, as the rough figures^ on p. 170 show.

As in the case of species, no country has all its genera endemic,
and most are very far indeed from this condition. Contrary to

1 Taken, without criticism, from my Dictionary, and not revised in detail.

170 ENDEMISM AND DISTRIBUTION: GENERA [pt. ii

Endemic

Endemic

genera

genera

Islands

about

Continental

about

British Islands

0

•Mediterranean region

280

Macaronesia

20

India

320

Ceylon

23

Australia

470

Jai)an

69

Colombia

87

Fiji

56

Peru

75

Hawaiian Islands

45

Chile

140

Borneo

71

Brazil

533

Java

62

Argentina

47

New Guinea

146

South America

1731

New Caledonia

134

South Africa

523

New Zealand

32

Africa

1733

Madagascar

266

Mascarenes

64

Juan Fernandez

10

what is often supposed, the proportions of endemic genera upon
islands are usually small : they range from nothing for the British
Islands to about 12-20 per cent, upon such islands as Juan
Fernandez, the Mascarenes, and New Caledonia, being as usual
larger in the more southern islands. On larger areas of grovmd
the proportions are greater; Brazil has about 21 per cent, of its
genera endemic, and so has Chile, Australia about 30 per cent..
South Africa about 35 per cent. Africa as a whole has about
46 per cent., and the proportions increase with increasing area
till one finds 100 per cent, endemic in the world.

AVhilst in general it is true that increasing size of area, greater
isolation, and greater nearness to the southern limit of about
40-48° S. are accompanied by increasing number and proportion
of local genera, these are probably not the only factors in the
question. If the country from which the invasion of plants has
come be inhabited by great numbers of them, or if the communi-
cation between them be broad, the proportion of local genera
will be more likely to be large.

There is no definite and demonstrable difference between en-
demic genera and others, and we shall endeavour to show, just
as in the case of species, that the phenomena exhibited by them
are simply a miniature of those exhibited by genera as a whole.

One may, to a very large extent, repeat the preceding chapter,
but with genus substituted for species, and family for genus, and
find it to agree with the facts about endemic genera, which
behave like the species. Just as in their case, the areas occupied
by genera, whether so local that they are classed as endemic, or
whether of larger size, are nicely graduated from small to large.

CH.xviJ ENDEMISM AND DISTRIBUTION: GENERA 171

Inasmuch as a genus consists on the average of over twelve
species, which never all occupy the same area, it is obvious that
the average area occupied by a genus must be larger than that
occupied by a species, but that does not affect the argument.

Some endemic genera occupy very small areas, e.g. Homalo-
petalum in three parishes in Jamaica, Itatiaia on one mountain
in southern Brazil, Sphagneticola in a suburb of Rio de Janeiro,
Leichhardtia on the Daintree River, Carpolyza in the immediate
suburbs of Cape Town, Traunia and Spondiopsis upon Kili-
mandjaro, Cladopus in one or two streams in Java, Alsiiiidendron
upon Oahu Island, Neohracea upon several of the Bahama islands,
Podadenia in the neighbourhood of Ratnapura in Ceylon, and
so on. Or if one take a single country, New Zealand, for" example,
and take a few of its endemic genera, one finds Siphonidium and
Toivnsonia upon very small areas, Pachycladus upon one slightly
larger. Colensoa reaches about 80 miles along New Zealand,
Tetrachondra about 100, Anagosperma about 140, Notospartiwn
about 240, Ixerba about 300, Hoheria 700, Tupeia 1000, and
Carpodetus the whole length of 1080 miles from North Cape to
the south of Stewart Island. Of the eighteen genera endemic to
New Zealand which have one species each (37), six are confined
to areas not over 140 miles in length, or 33 per cent, of the
genera upon areas not exceeding 13 per cent, of the whole, so
that the tendency even here is to give a hollow curve (cf. pre-
ceding chapter).

In Ceylon, the Hawaiian Islands, and elsewhere one finds the
same type of distribution, and if one go on to larger and larger
areas one finds larger and larger areas for genera in the same
graduated way, until one comes to such a Avorld-ranging genus
as Senecio, or Astragalus. Though of course there are many
exceptions, on the whole the size of the genera (number of their
contained species) becomes steadily larger with the increasing
area, as we have already pointed out in Chapter xii ; of course
allied groups only being compared.

If instead of taking individual genera, or the endemic genera
of a single country, one take all the genera of a small family, one
finds the same graduation of areas. Take, for example, the
Polemoniaceae (from the Pflanzenreich). Of its twelve genera,
three, with one species each, occupy (roughly) California and
Utah, Mexico and Guatemala, and the Pacific United States.
One with five species is found in California, Utah, Nevada, and
Arizona, one with six in the Andes from Colombia to Chile.

172 ENDEMISM AND DISTRIBUTION: GENERA [pt. ii

There are two with nine species, one in the Andes from Mexico
to Chile, including Venezuela, and the other with eight species
in Pacific North America and one in Atlantic. So far the areas
occupied are closely correlated with the number of species, but
in the bigger genera there is more variation. Loeselia with twelve
species rmis from California and Texas to Colombia and Vene-
zuela. Gilia with 109 and Navarretia with 41 both occupy North
America, the Andes, and Argentina, while Phlox with 48 covers
North America and part of Siberia. Finally, Polemonium, which
has only 29 species\ covers North and South America, and north
temperate Europe and Asia. Except for this last genus, which
covers the whole family range, the area is roughly proportional
to the number of species (cf. Chapter xii) and the grouping is
just like that of the endemic species or genera.

This type of distribution is very common indeed, showing in
perhaps the greater number of the families. One genus, usually
with many species, covers the whole or most of the family range,
the smaller genera, with more restricted ranges, being the more
numerous, and on the whole increasing in number the smaller
they are, and the more restricted their range. In the Polemoni-
aceae, there are eight genera below, and four above, the average
size for the family, one of the latter occupying the whole family
range.

If one take the Cistaceae (37), one finds Halimium with 26
species covering the whole family range, while Helianthemum
with 70 covers the Old World from Macaronesia to Beluchistan
and Arctic Europe, and Lechea with 13 covers North and Central
America and the West Indies. The rest, with 20, 12, 9 and 3
species, cover smaller ranges within these.

Or if one take the large "and widespread Menispermaceae (fig.
on p. 173), one finds (37) Cocculus and Cissampelos with a
distribution practically covering that of the family, Stephania
and Tinosjjora covering most of the Old World, and Hyperbaena
most of the New World, range. Within these are many genera
of smaller and smaller range till one comes down to the 12
in West Africa, 5 in Brazil. 5 in Madagascar, etc. There are
2 genera of maximum range, 4 of rather less (including Meni-

1 This curious j)oint, that tlie most widely spread genus of all has fewer
species than some of the others, is by no means unique, but occurs in a
number of families, e.g. also in the Menispermaceae, Cistaceae, and Hydro-
phyllaceae. It requires careful investigation with the aid of palaeobotany,
for it seems to me not impossible that the deficiency in species may be
connected with the occurrence of the glacial period.

CH. XVI] ENDEMISM AND DISTRIBUTION: GENERA 173

174 ENDEMISM AND DISTRIBUTION: GENERA [pt. ii

spermum in Atlantic North America and north-east Asia), about
19 in the next class, and 24 or more in the lowest class.

This type of distribution corresponds to that of the species of
Doona, Gymnema, Cissampelos, etc., described in the preceding
chapter (p. 157). But the Cijrtandra type (p. 159) can also be
matched, e.g. by the family Monimiaceae (37), in which there
are 22 genera with small areas (the largest being New Guinea and
Celebes) and 49 species in all (average 2-2 species per genus),
5 genera with areas of moderate size (and 22 species, average 4-4),
and 5 with areas of large size (and 196 species, average 39-2).
These larger areas overlap one another to some extent in some
cases, but there is no single genus covering, or nearly covering,
the range of the family.

All these groups of genera, it will be seen, give indications, even
when considered singly, that the areas they occupy go with their
number of species, and if taken in groups, the applicability of
Size and Space is clearly obvious.

So far, in dealing both with endemic (and other) species, and
with endemic (and other) genera, we have been considering only
the areas occupied by them, and we have seen that these are
graduated from many very small areas through a good many of
a size somewhat larger up to a tail of a very few that occupy the
largest areas. Plotted graphically, as in fig. on p. 162, the
numbers always form a hollow curve.

But now, if age be the chief determinant of spread ^ as would
appear to be the case from all the figures that have been given,
and from the success of the many predictions based upon it that
have been made; and if Size and Space be equally valid, then it
would seem that the sizes of the genera {i.e. their numbers of
species) in any group of endemics should also be arranged in a
hollow curve. If Age, Size, and Space (or Area) go together, then,
as age is the only active^ factor of the three, it is clear that what-

1 As already pointed out, age of itself effects nothing, but the fact that
dispersal goes so largely with age shows that the various factors that are
operative produce an average or resultant effect, so that in twice the time,
twice the dispersal will occur, unless barriers (physical or ecological) inter-
fere. The essential difference between this view and the older one is that
under Age and Area all species (with few exceptions) are looked upon as
enlarging their area, instead of a few doing so, and many contracting theirs.

Manv people take the pojtular view, which is based, it must be remem-
bered, upon an assumed efficacy of natural selection for which as yet there
is little proof, that species with "small areas of distribution owe the fact that
those areas are small to the competition of other more successful types.
But there is little evidence for such a belief. It is simply a way of looking

CH. XVI] ENDEMISM AND DISTRIBUTION: GENERA 175

ever phenomena are shown by space should also be more or less
paralleled by those shown by size. We are thus led on to the
investigation of the sizes of genera, to see whether they may not
show some definite relationships to one another, such as might
be expressed by the aid of curves.

Very little investigation is required to show that this is indeed
the case. If we take 16/1 to mean 16 genera of one species each,
3/2 to mean 3 genera of two species each, and so on, then examine
the endemic flora of all the islands of the world, and pick out
those genera that are actually endemic to the islands, one finds
that all the islands show the same type of arrangement, as may
be seen in the following list of examples :

Table showing the numhers of Endemic Genera
of different Sizes upon a number of Islands

Azores, Canaries, Madeira 16/1, 3/2, 1/4

Borneo .... 59/1,8/2,2/4,1/5

Ceylon .... 19/1,2/2,1/3,1/5,1/11,1/15

Cuba .... 58/1,9/2,2/3,2/4,1/6

Hawaiian Islands . . 14/1,6/2,7/3,4/4,3/5,2/6,2/7,1/9,

1/11,2/12, 1/14, 1/17, 1/28
Japan .... 54/1,9/2,1/3,2/4,1/8
Java .... 57/1,2/2

Madagascar . . . 191/1,37/2,10/3,7/4,9/5,2/6,2/7,

1/8, 5/10, 1/12, 1/18, 1/20
New Caledonia . . 73/1, 27/2, 6/3, 4/4, 4/5, 4/6, 2 7,

1/9, 3/10, 2/12, 1/15
New Zealand (proper) . 22/1, 2/2, 3/3, 2 4, 1/5, 1/9
Socotra .... 17/1, 1/2, 1/3

On such large islands as Madagascar, where there are many
endemics, the same phenomenon is shown even by single families.
Thus the Madagascar Compositae show 11/1, 2/2, 1/3, 1/5, and
1/10, the Rubiaceae 14/1, 3/2, 1/4.

Every island in the world that possesses any endemic genera

at the aetual fact, which is all we have to go upon, that A occupies a large
and B a small area. My way of looking at the same fact is to suppose that
A is older than B. This is really a much more simple explanation, especially
when we remember that the areas occupied by the different species in a
genus, or the different genera in a family, usually increase fairly regularly
from very small to large. If one have areas represented by 1,2,3, 4, 5, 6,
7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, it seems an unnecessarily
oblique way of looking at the facts to say that 1, 2, 3, 4. and 5 must be
regarded as dying out, while 16 to 20 are to be looked upon as successful
and expanding species, and no two authors can agree about whetiier the
intermediate species 6 to 15 are one thing or the other. It is far more
simple to regard all as still in process of expansion, but that some, by reason
of greater age and perhaps other advantages, have grown larger than others.

176 ENDEMISM AND DISTRIBUTION: GENERA [pt. ii

shows them arranged in this way, with many monotypes (or
genera with one species only), a fair number of dit3^pes, and a
tail of a few larger genera. When plotted graphically they con-
sequently form the hollow curve that we have begun to meet so
often in dealing with distribution (cf. fig. on p. 177). One
must make allowance, in considering the figures above given, for
the "lumping" that is practised in my Dictionary, especially at
the fives and tens.

If one add up the grand total of 1582 endemic genera of all
the islands of the world, one finds that they show 1037/1 (1037
of one species), or 65 per cent, of the total, and 233/2, or 14-7
per cent., these two making up nearly four-fifths of the w^hole.
There are 104/3, or 6-5 per cent., 53/4, 49/5, and so on, the largest
endemic genus of islands being Oncostemoji with CO species. If
one take for comparison the endemic genera of Brazil, 533 in
number, one finds 334/1, or 62 per cent,, 91/2, or 15-2 per cent.,
33/3, or 6-2 per cent., and so on, the largest having 50 species.
In both these cases the same type of result, showing a well-
marked hollow curve, is obtained, and one gets the same what-
ever region of the world one may try for endemic genera, e.g. any
of the other countries of South America, or South Africa (cf. the
first two and 4th and 7th curves in fig. on p. 177).

It is worthy of notice that in these two instances, the islands
and Brazil, the percentages of genera of different sizes are much
the same, the monotypes for example being 65 per cent, in the
one, and 62 per cent, in the other. The islands, which actually
cover about two million square miles, would probably be nearly
equal to Brazil if the included seas were taken. The average
number of species per genus is also not unequal (islands
1582/3461, average 2-1; Brazil 533/1291, average 2-4).

The endemics of mountains are also as a rule small genera,
though there are a fair mmiber of exceptions to this, but only
in the large mountain chains. In the Andes, for example, there
are Chaetanthera (30 species), Cinchona (40), Cristaria (30), Nas-
sauvia (50), Psammisia (35), Fuya (25), and many more of
smaller size.

One may go on to deal with still larger floras, and find that
they are arranged in precisely the same way, so that the pheno-
mena shown by the endemic genera are exactly paralleled by
those shown by genera that occupy more area. If one take (as
usual from my Dictionary, in which uncertain fours are counted
as fives, etc.) the genera that are confined to single continents

CH. XVI] ENDEMISM AND DISTRIBUTION: GENERA 177

CO

X

X

J

—£

-d

2 -,;.

Emdlmics.

o

fA X

Genet'a- grouped by ^ire..

(0 ^-

E&ch curi/c IS 5 S'jua.res

o

u

<s

dLa.gondllvy abovfe the last.

^

</5

X

o

>_"

5

-d

05

-c

(1) %

<s <^ ol X

^

1 on

X X X -^^

X

^,(0

V ^' rj ^

o

\*

c

H

(0 J 3 J ^

^ ^ ^ ^ ^m

"t"

.1

'^

X S z: U

.1

J3

z:

-^

'

^

-

♦

\

\ ^—-~,

\ *

I

V

-

Curves showing the numbers of genera of one. two, three, five, ten (and
sometimes more) species, in various groups of endemic genera. Always
there is a great preponderance of ones and twos and a tail of few larger
genera. The monotypes are at the left-hand end of the curves (1).

W.A. 12

178 ENDEMISM AND DISTRIBUTION: GENERA [pt. ii

or continuous areas, one finds, for example, that in Africa there
are 835/1, 254/2, 136/3, 86/4, 97/5, 48/6, and so on, the largest
genus having 350 species. In tropical Asia one finds 445/1, 175/2,
90/3, 68/4, 77/5, 56/6, and so on, the largest genus having 600
species. In the north temperate region of the Old World one
finds 385/1, 135/2, 75/3, 45/4, 49/5, 29/6, and so on to 250. From
this one may go on to the world itself, and one finds (in the total
of 12,571) 4853/1, 1632/2, 921/3, and so on to 1600. All these
groups of figures exhibit markedly hollow curves when plotted
graphically.

The various figures that have just been given for islands,
countries, continents, etc., show in a very distinct way that the
larger genera are found upon the larger and more isolated areas,
whether of islands or of countries on the mainland, as would be
expected upon the principle of Size and Space (Chapter xii).
Thus, while Ja\a has no endemic genus of more than two species,
nor Socotra of more than three, Borneo reaches five, Ncav Cale-
donia 15, IMadagascar 20, and the very isolated Hawaiian Islands
28. The largest island endemic genus, Oncostemon with 60 species,
is found in Madagascar and the Mascarenes, a large total area.
Astronia, the next largest, with 30 species, occupies large parts
of the Malay Archipelago and Polynesia. The largest genus con-
fined to New Zealand proper has only 9 species, but that con-
fined to New Zealand and surrounding islands (p. 66) has 20,
In the same way, the possible size of a genus increases with the
increasing size of the area, till we reach 600 species in a genus of
Tropical Asia, and 1600 in the world.

All these groupings of genera, whether usually considered en-
demic, or not, whether confined to small areas, or found on
larger (even up to the whole world), show the same type of
arrangement, with the bulk of their number monotypic or di-
typic, and a tail running out to the larger genera, the tail being
longer the larger the size of the area dealt with. There is no
difference between the endemic genera and the rest.

It is also evident that the sizes of genera are grouped in the
same Avay as the areas occupied by their species. Both go with
age ; the older the genus, the more space will it occupy, and the
more species will it have. Of course one must only deal with
groups of say ten genera, and must only compare allied forms,
to get results that are at all reliable and comparable.

It is clear that the general types of relationship shown, whether
between endemic genera only, between genera of larger area

CH. XVI] ENDEMISM AND DISTRIBUTION: GENERA 179

only, or between these two classes, are the same, and that they
are the same whether we consider the areas occupied by the
species of the genera, or the numbers of species in the genera
themselves. The same type also appears in the population of a
country by its flora, whether some of it is endemic or not. In
all cases of distribution, whether it be distribution by areas
occupied — geographical distribution or distribution in space —
or by numbers of species in the genera — evolution or distribution
in time — the distribution seems to have been determined largely
by time. If age alo7ie were operative^, one would get much the
same distribution as at present exists, when one allows for geo-
logical happenings, and the action of barriers. Among these
latter, of course, ecological barriers are of great importance, but
the general evidence goes to show that their action is principally
negative, like that of physical barriers.

Just as with species, endemic genera have been regarded as
(1) locally adapted — a view which has largely died out, especi-
ally since it was realised how difficult it would be to find anything
to which such a list of genera as those given above for New
Zealand (p. 171) could be adapted, and a view upon which it is
impossible to explain such an arrangement of genera in order of
size as we have just seen to be the rule; (2) as survivals; and
{3) as in general new genera beginning life as such.

As islands have always been regarded as the typical location
in which to look for endemics — species, and still more genera —
we may do well to consider them.

Now if the endemic genera of islands be in reality survivals,
one would expect that they would at least show a tendency to
belong to families that are small or of broken distribution, i.e.
such families as we have been accustomed to look upon as more
or less moribund. And in any case, one would not expect the
great bulk of them to belong to the large and "successful"
families. If, on the other hand, Age and Area hold good, they
should be found to occur upon islands (provided the connection
was mainly by land) in proportions not dissimilar to the pro-
portionate sizes of existing families.

In order to test this question thoroughly (135). I have added
up from my Dictionary (1 ) all the endemic genera of all the islands
in the world, (2) all the endemic genera of West Australia, South
Africa, and Brazil, three areas very rich in endemics, and with
much variety of habitat, (3) all the genera confined to Australia,
1 I.e. if the average speed of dispersal of a species were constant.

180 ENDEMISM AND DISTRIBUTION: GENERA [pt. ii

Africa, and South America, and (4) all the genera of the world.
Arranging the families in groups of ten in the order of their size
in the world (as judged by number of genera), and taking for
each of the other three areas the number of genera in the sa?ne
ten families, one gets the following table :

Table showing in each pair of columns the numher, and the per-
centage, of genera that occur in the world, and thai are confined
to three sections of it {ending with those confined to the islafids).
The first horizotital line shows the figures for the ten largest
families in the world for each of these, and the following lines
those for the second, third, etc. tens of families in the xoorld. The
percentages are counted downwards ; the 40 per cent, at the top
of the first column means 40 per cent, of the genera of the world.

Australia,

W. Australia,

Tens of

Africa,

S. Africa,

families

World

S. America

Brazil

Islands

(world

, ^

, ^

— ^

, ^

order)

Genera

%

Genera

o/
/o

Genera

%

Genera

%

1

5019

40- 1

1579

391

459

40-5

606

38-3

2

1868

14-9

592

14-6

176

15-5

285

180

3

1094

8-7

360

8-9

86

7-6

144

91

4

874

6-9

325

80

78

6-8

115

7-2

5

695

5-5

271

6-7

75

6-6

83

5-2

6

561

4-4

216

5 3

57

50

82

5 1

7

456

3-6

83

20

19

1-6

55

3-4

8

355

2-8

111

2-7

30

2-6

48

30

9

296

2-3

99

2-4

24

21

29

IS

10

233

1-8

79

1-9

29

2-5

37

2-3

Total

11,451

91-4

3715

921

1033

911

1484

93-7

11 to 20

919

7-3

278

6-8

90

7-9

86

5-4

21 to 29-1

147

11

38

0-9

10

0-9

12

0-75

Grand total

12,517

99-8

4031

99-8

1133

99-9

1582

99-85

The percentages agree with one another in the four columns
in the most remarkably close manner, as a little inspection will
soon show. The greatest difference in the whole table occurs in
the second line, between 14-6 per cent, for Australia, etc., and
18-0 per cent, for islands, a difference of 3-4 per cent. The second
greatest is in the first line, between 38-3 for islands and 40-5 for
West Australia, etc., a difference of 2-2 only.

If these percentages be plotted as curves, they give the re-
markable figure shown.

The close coincidence of these (hollow) curves is very remark-

CH. XVI] ENDEMISM AND DISTRIBUTION: GENERA 181

able, and after looking at them it is difficult any longer to main-
tain the position that endemic genera in general are survivals of
old floras. Of course there are many single examples that are
such, but they are quite lost in the crowd when one deals with
large numbers. Survivals Avould never, so far as one can conceive,
be graduated like this.

The four columns of percentnges in the table above, plotted as curves.
Vertical readings are the percentages, horizontal the number of the group
of ten families. (By courtesy of the Editor, Annals oj Botany.)

Confirmatory evidence may be obtained in various ways.
Families that have been long enough upon islands to give rise
to endemic genera must be very old, and so must families that
have reached both Old and New Worlds. One will therefore
expect these two lists to coincide to a large extent, and in fact
one finds that 90 per cent, of the island families that contain
island endemic genera also reach both worlds. Or again, one
will expect that the oldest families Avill have reached most
islands, and should contain the most endemic genera by reason
of their age. This is easily found to be the case; the West Indies

182 ENDEMISM AND DISTRIBUTION: GENERA [pt. ii

have 195 endemic genera in 43 families that also occur in the
islands of Indo-Malaya, and only 19 in the 16 families that do not.
They have 187 in 39 families that occur on the African islands,
and 27 in 20 families that do not.

Incidentally, the close correspondence of these curves shows
that it is all but certain that the floras of the world, in the mass,
must have been distributed by land connections, and at any
rate those of the bulk of the islands, though some of the far out-
lying ones, with few endemic genera, probably were oceanic.

Just as the endemic species belonged to the large and "suc-
cessful" genera in greater proportion, so the endemic genera
belong to the large and "successful" families, and only a very
few indeed to endemic families. An analysis of the above table
of 1582 endemic genera of islands shows that 1150 of them, or
72-6 per cent., are found in the 40 largest families in the world,
which only contain 70-6 per cent, of the total genera in the
world, i.e. these families contain rather more than their proper
proportion of endemics. The remainder occur in another 110
families, leaving 141 which are not represented upon islands by
any endemic genera at all. The largest of these latter families is
the Chenopodiaceae with 86 genera, and the whole number only
contain 890 genera, or 6 per family, against 77 per family for
those which have island endemic genera. The proportion of
endemic genera diminishes from top to bottom of the table
(cf. 135, p. 509).

The further out, and more isolated, the island is, i.e. in general
the more ancient the date of its peopling with plants, the more
do the endemic genera tend to belong to the larger families. If
one divide the 150 families that possess them upon islands into
75 larger and 75 smaller, one finds that in Madagascar 62 of the
families with endemic genera belong to the larger, 18 to the
smaller. In New Zealand the proportion is 16/4, and in the
Haw^aiian Islands 13/1.

If endemic genera were really largely relics, one would expect
that there would be a fair number of endemic families, but, as
a matter of fact, these are fe^v and small, and of the five that
are found only upon islands (Chlaenaceae, Balanopsidaceae,
Corynocarpaceae, Lactoridaceae, and Cercidiphyllaceae), the
largest is upon the largest island (Madagascar) that is also a
good way out from the mainland.

Putting together all the facts about endemic genera that have
been given above, and which show that in the mass they behave

CH. XVI] ENDEMISM AND DISTRIBUTION: GENERA 183

like endemic species, and that both endemic species and genera
behave Hke non-endemic, it is clear that nothing but a mechanical
explanation will serve for the chief features of their distribution,
when one is dealing with the mass. Age supplies such an explana-
tion, but this is hardly possible to the supposition either that
they are chiefly relics, or that they are chiefly local adaptations.
It Avould thus seem to follow that endemics in the mass, whether
species or genera, are chiefly youpg beginners, descended in
general from the more widely distributed forms about them.
The smaller the area occupied, on the average, the younger the
species or genus.

Only in comparatively rare cases can we look on forms of
small area as relics. The fact that in every family the monotypes
are from two to three times as numerous as the ditypes is fatal
to any idea of relic nature for the great bulk of them. Of course,
just as in the case of species, we must make varioiis provisos for
the use of Age and Area, such as that the genera be only com-
pared in groups of ten allies on either hand of the comparison,
that they be only taken in tens in any case (to lose the relics in
the crowd), and that conditions remain reasonably constant.
Species and genera are endemic simply because they have not
yet had time to spread abroad, or because they have been pre-
vented by barriers, sometimes physical, sometimes ecological.

Summary

Endemic genera occur in similar places to endemic species,
and instances are given of the numbers that occur in various
parts of the world, from which it appears that islands in general
have the smallest proportions. Avhile the proportion increases
with increasing area, up to 100 per cent, for the world. Examples
are quoted of very small areas occupied by many endemic genera,
usually monotypic, and more detail is given of the distribution
of genera in several families, showing that on the whole the area
varies roughly with the number of species, and that both types of
distribution seen in the preceding chapter— one species covering
the whole generic range, or several species dividing it among them
—can be matched in the families, and the genera pertaining to
them.

It is then shown that endemic genera are distributed in different
countries in regular order, with many monotypes, fewer (but
still many) ditypes, and numbers tapering away to the larger

184 ENDEMISM AND DISTRIBUTION [pt. ii, ch. xvi

genera, which are usually found only in large islands or other
large areas. If plotted as graphs the figures give the usual
hollow curves, and it is clear that the sizes of the genera depend
on factors similar to those that determine the sizes of areas
occupied by species.

Still larger floras, e.g. those of single continents, or of the
whole world, show the same type of arrangement of the genera,
with many monotypes, fewer (but still many) ditypes, these two
making about half the total, while the larger genera taper away
steadily in number in a long tail.

It is clear that neither the supposition that endemics and
small genera are relics, nor that they are special adaptations
will avail to explain the phenomena presented by the great mass.

Endemic genera further prove to belong more to the large
families, just as endemic species belong to the larger genera.

The case of islands, usually regarded as the typical home of
endemic genera, is then considered in more detail, and it is
shown that the proportions of endemic genera in (1) the islands
of the world, in (2) West Australia, South Africa, and Brazil,
and in (3) Australia, Africa, and South America, are much the
same for all three, for each group of ten families in order of size,
and this proportion is the same as occurs in the world for each of
these groups. Confirmatory evidence is also given, the result of
the whole being to show that in the mass endemic genera are
simply, like endemic species, young beginners, and probably the
descendants of other genera still existing.

CHAPTER XVII

THE MONOTYPIC GENERA, AND GENERA
OF LARGER SIZE

r ASSiNG on now to deal with monotypic genera, or genera with
one species only, one soon notices that they show the same
phenomena that we have already seen in the endemic genera.
This is what we should expect upon the hypothesis of Age and
Area, as expanded by Size and Space, implj'ing as they do that
small genera, endemic or not, are on the whole younger than,
and occupy less territory than, the larger genera in the same
circles of affinity.

Few people, perhaps, have realised how numerous the mono-
types are. No less than 4853 out of the 12,571 genera of flower-
ing plants in my Dictionary (4th ed.) are monotypic, and are
usually so restricted in area that most people would call them
endemics, A number will doubtless proA'c to have more than one
species when we finally know the flora of the world, but new ones
are frequently discovered, or created by the splitting of other
genera, and there is little likelihood that the percentage will fall
much below its present figure of 38-6 per cent, of the total. The
ditypes, or genera of two species each, are also very numerous,
and include 1632 genera, or 12-9 per cent. In other words, the
monotypes and ditypes alone include more than half the genera
at present existing, or 51-5 per cent., while the tritypes include a
further 921, bringing the total to 58-9 per cent. The monotypes
are approximately three times as numerous as the ditypes, and
these almost twice as numerous as the tritypes. Beyond ten
species the figure for number of genera goes below 500, and at
twenty-five species below 200, tapering out in an enormously
long tail to the final genera Senecio (1450 species) and Astragalus
(1600).

We have already seen many instances of the hollow curve, and
when the genera of the world are plotted by numbers containing
1, 2, 3, etc., species, one gets a beautiful example of it. It is idle
to suggest that further work will alter the form of this curN-e.
The monotypes exceed the ditypes by 3221, and the ditypes
exceed the tritypes by 711, and so on right through the list.

One may even go beyond the genera, and find that the families

186 THE MONOTYPIC GENERA [pt. ii

are arranged in the same way with regard to the numbers of
genera contained in them. There are 54/1 (54 of one genus),
45/2-3 (45 of 2 or 3 genera), 40/4-6, 32/7-13, 28/14-23, 25/24-38,
22/39-63, 20/64-100, 15/101-200, 13/201-1143. The numbers
steadily decrease, while at the same time the number of species
included increases, being 1, 2, 3, 7, 10, 15, 25, 37, 100, 943, again
forming a hollow curve.

But if the whole flora of the world show such a remarkable
grouping of its genera into sizes, then one will expect the same
type of arrangement, in a hollow curve, to hold for the individual
families, and in actual fact one finds that this type of grouping
into sizes holds for the genera of any single family, with a few
trifling variations among the very small families. For example:

The families Contain

Acanthaceae (206 gen.) 119/1, 32/2, 20/3, 9/4, 15 5, and so on to 300

Aceraceae (6) 1/1, 1/3, 1/4, 1/5, and 7 and 115

Aizoaceae (20) 8/1, 3/2, 1/3, 1/4, 2/5, and so on to 15

Alismaceae (15) 5/1, 3/2, 3/3, 1/4, and so on to 33

Amarantaccae (72) 29/1, 10/2, 7/3, 2/4, 2/5, and so on to 100

Amaryllidaceae (94) 28/1 , 15/2, 10/3, G/4, 3/5, and so on to 100

Commelinaccae (38) 15/1, 4/2, 3/3, 2/4, 2/5, and so on to 110

Compositae ( 1 143) 446/1 , 140/2, 97/3, 43/4, 55/5, and so on to 1450

Coniferae (45) 14/1, 8/2, 2/3, 5/4, 1/5, and so on to 70

Saxifragaceae (96) 51/1, 12/2, 2/3, 5/4, 1/5, and so on to 225

Scrophulariaceae (241) 88/1, 32/2, 18/3, 12/4, 8 '5. and so on to 250

Simanibaceae (39) 17 1, 6/2, 2/3, 2/4, 3/5. and so on to 30

The whole nimiber of families form similar hollow curves; the
Coniferae are one of the most aberrant families of the entire list.
As a general rule, the genera with one and tMO species make up
about half the total (cf. fig. on p. 187).

This type of grouping even holds for families of lower type
than the flowering plants; for example, the Jungermanniaccae
acrogynae show 21/1. 6/2, 9/3, 4/4, Q/S and so on, the Rhodo-
melaceae 34/1 , 1 6/2, 5/3, 5/4, 6/5 and so on, the Hymenomycetineae
23/1, 10/2, 3/3, 8/4, 3/5, and so on. The numbers are more irregu-
lar, but the hollow curve is clearly shown.

It is clear that this type of distribution of the genera by the
number of their contained species is a perfectly general phe-
nomenon. There are no exceptions, when allowance is made for
the lumping in my Dictionary. If endemic genera, or monotypes,
were really mainly relics or special adaptations, such distribution
as this would be inconceivable, obtaining as it does in every
locality, and agreeing with the distribution of genera about the

CH.xvii] AND GENERA OF LARGER SIZE 187

Avorld, and with their distribution into famihes, as well as with
the distribution of species— endemic or not— by area occupied.
All show the same hollow curves.

rAniLlC.5 m ORDER or size ai )s^

SMowiMC nunbtRs or ctncKA _. ~ ~

WITH DirrtRcnT nuriBCRi or iPtciLi.

Hollow curves exhibited by the grouping Into sizes of the genera in the first
15 largest families of flowering plants. Each curve is diagonally above the
preceding one, as indicated by the heavy black dots (points of origin). Note
that the curve almost always turns the corner between the point marking the
number of genera with 3 species, and that marking the number with 5
(mdicated by the dotted lines). The number after the name of the family shows
the number of genera in it.

Not only so, but the biggest genera are in general in the largest
families, i.e. in general the oldest families. If one take (from my
Dictionary, as usual) the largest genus in each family, and average
them, one finds

188

THE MONOTYPIC GENERA

[PT. II

Average of

Size of

Size of

largest genera

largest

family

Number of

in each

genus

(genera)

families

(species)

(species)

1

54

63 j

2-3

45

600

4-6

38

7-1 S

32

79^

95

125J

14-23

28

700

24^38

25

39-63

22

161^

64-100

20

278 •

1600

over 100

28

404,

again a hollow curve. One may even find the sections of this
curve regularly arranged. The families with one genus show 12/1
(twelve with one species), 8/2, 6/3, 3/4, 3/5, 2/6-7, 3/10, 3/15,
and so on to 290 [Symplocos). This result agrees absolutely with
what has been said under Size and Space in Chapter xii.

There is no demonstrable difference between monotypes and
genera of larger size, except in the smaller number of species,
and (usually) smaller area occupied. Of the 4853 of them, 1037
occur only upon the islands of the world, usually only upon one;
887 occur in South America, usually only in small portions of it;
835 in Africa, 612 in North America. These four divisions of the
world contain in all 3371 monotypes, or 69 per cent, of the total
mmiber in the world. But if one count up the number of ditypes
they contain, one finds it to be only 59 per cent, (or a much smaller
proportion) of those in the world. The number of tritypes con-
fined to these portions of the world is only 51 per cent., or a
lesser proportion again, that of genera with 5 species 46 per cent.,
of genera with 10 33 per cent., of genera with 50 20 per cent., of
genera with 100 species 10 percent., and of genera with more than
100 species they contain only 3 per cent, of those in the world.
It is clear that what Avas said above under Size and Space is in
general correct, and that the larger genera tend to occupy larger
areas in proportion to the number of species that they contain
(for the proportions decrease with perfect regularity). Distribu-
tion about the Avorld, and number of species, go mainly with
Age. It is inconceivable that natural selection should group
genera like this.

This regular curve for the occurrence of genera not only shows
with a large number, such as those just considered, but also with
much fewer. If we divide the world into continents and larger
areas, and enumerate for each region the genera confined to it,
we get:

CH. xvii] AND GENERA OF LARGER SIZE 189

Table showing, in each line, the percentages of genera confined to
the Islands, Australia, etc., and containing 1, 2, 3, or other nujnher
of species. The percentage is of the total number of genera con-
taining 1, or 2, etc., species, not of the total number of genera
confined to the islands, etc. 21ms 21 per cent, of all the mono-
types are found upon the islands, 49 per cent, of the genera -with
75-125 species occur in both Old and New Worlds.

Percentages of Genera of different sizes (numbers of species)

No. of species in genus

1

2

3

4

5

10

50

75- ,
125

Above
125

Islands

21

14

11

8-9

7-8

3-6

—

—

Australia

4-9

4-5

5

4-7

4-3

1-7

1-6

1-7

16

Africa

17

15

14

14

15

10

10

3-4

10

South America

18

16

13

12

13

11

7-3

6

0-5

Tropical America

2-9

5-7

9

11

11

17

14

13

10

North America

12

13

12

10

9

8

2-4

0-8

05

N. Temp. Old World

8

8

8

7

7

7

7-3

6-3

1-6

Tropical Asia

9

10

9

11

12

11

G-5

4

2-7

Palaeotropical

0-7

2-9

5

6

6

9

14

13

5-9

New and Old Worlds

1-3

4-2

4-9

6

7

14

31

49

73

Miscellaneous, mostly

of large areas in Old

World

51

66

9

9-3

7-8

7-6

5-8

2-7

2-5

100 100 100 100 100 100 100 100 100

This is a veiy remarkable table. In the case of Islands, Africa,
South America, and North America (with a slight exception at
the monotypes), the proportions of genera of different sizes
decrease regularly (allowing for the limiping of uncertain fours
and sixes as fives). This fact seems to me practically to exclude
the idea of local adaptation, as well as that of relic nature, for
the great bulk of genera, though there must of course be many
exceptions to this rule. But if this be so, then the idea that
plants have been guided in their evolution by natural selection
must also suffer something of an eclipse. One cannot imagine
natural selection producing genera in careful graduation of sizes
(and areas) like this. One would get distribution almost exactly
of this type by the simple operation of the "mechanical" prin-
ciple of Age and Area^ as expanded by its corollar}^ Size and Space.
If these two worked alone, and absolutely, one would get this

1 As already several times explained, the general meaning of Age and
Area is simply that on averages and in the long run species and genera
spread at a more or less uniform rate, interfered with by barriers, physical or
ecological. On the older view it was imagined that distribution was do
rapid that all forms had already reached their limits, and that many were
in process of contracting their area of dispersal.

190 THE MONOTYPIC GENERA [pt. ii

type of distribution shown in minute detail; and working upon
large numbers, one gets it shown quite clearly.

One may, however, go further than merely splitting up the
world into continents. If one take the genera endemic to South
America, and divide these up among the countries to which the
bulk of them are confined, and then take, for example, Brazil,
one finds that it contains G-8 per cent, of the world's monotypes
(genera of one species), but only 5-5 per cent, of the ditypes,
3-5 per cent, of the tritypes, 0-7 per cent, of the genera with ten
species, and so on. The other countries of South America show
similar, but not quite so regular, results (on account of the smaller
numbers). Individual islands, when they ha^'e sufficient endemic
genera, also show the same. Thus Madagascar contains 3-9 per
cent, of the world's monotypes, 2-2 per cent, of the ditypes, and
1 per cent, of the tritypes, the numbers afterwards becoming
irregular on account of their insignificant totals, but none of
them approaching the figure for the ditypes (the highest is
1-4 per cent.).

If now, returning to the table, one look at the figures for the
largest area (New and Old AVorlds), which includes in general
genera that occur throughout the north temperate zone, the
tropics, or the world, but also includes a number that are only
found in eastern Asia and in North America {i.e. really quite a
small area), one finds the figiu'cs to go in the reverse direction,
from 1-3 per cent, of monotypes to 73 per cent, of the large
genera. This agrees absolutely with what has been said above
under Size and Space; the surprising feature is that the figures
increase regularly.

If now one take the Palaeotropical region (tropical Asia and
Africa, North Australia, Polynesia), one finds the proportions to
increase up to about genera of 50 species, and regularly, and
then to diminish regularly. Tropical America behaves in the
same wa}^ but the decrease begins sooner. In other words,
genera of larger size tend to occur in both Old and New World
tropics. In tropical Asia, a much smaller area, the falling-off
begins much sooner, and so it does in Australia. In the north
temperate regions of the Old World it does not begin till about
the size of 50 species (the flora, however, is more herbaceous).

There are many very interesting points to be made out from
the study of such statistics as these, and still more interesting
features can be discovered by breaking them up, and studying
individual regions, and families, or types of vegetation, in detail.

CH. xvii] AND GENERA OF LARGER SIZE 191

but it must suffice to have drawn attention to them, and to the
very clear way in which they show that on the large scale and in
the long run distribution is a very mechanical process, i.e. that
the various factors causing it act at a very imiform rate, and
that it is usually only stopped by actual barriers.

The individual area occupied by a monotype genus may vary
enormously, but is usually rather limited. 1037 of them occur
upon islands, and when the island is of any large size are usually
restricted to a portion of it. The great bulk of those mentioned
as found only in South or North America, or in x^frica, and the
241 of Australia, are similarly restricted, and so are most of
those in the other great regions of the globe. When one comes
to genera found in both worlds, one finds that only 66 of them
are monotypic, or a mere 5-6 per cent, of the genera that occur
in both. All but about 20 of these are found only in the north
temperate zone, which by reason of its connections by way of
the arctic regions, formerly passable for plants, is not really so
large in proportion as it seems. Bolhoschoenus, Brasenia, Hakon-
cchloa, Hijypuris, Montia, and Zannichellia are more or less
cosmopolitan, and of the remaining genera three are coastal
plants carried by sea currents, and four are tropical American
and West African— countries united by a current that crosses in
about three months. Pistia is a water plant, and some of the
others are doubtful identifications, so that there remain a bare
half-dozen that have a very large range, evidently acquired by
land, or nuich less than 0-25 per cent, of the total of monotypes.
These are Christiana, Eidophidium, Manisuris, Remirea, Rliabdia,
and Sphenoclea. In fact, it is fairly evident that if one were to
determine accurately the areas of the 4853 monotypes, one would
obtain the usual hollow curve, beginning with a great many of
very small area, and tapering aAvay to the other end as areas were
reached of larger and larger size.

In any country in which there are many monotypes, their
areas tend to overlap like those of the endemic species. Thus in
New Zealand, in any zone of 100 miles from north to south on
the main islands, there are never less than se\^en monotypic
endemic genera, though of the eighteen such genera six are
northern, ceasing towards the south, and twelve are southern,
two only of which reach the far north. Just as with the species,
the genera show a maximum number about the centre of New-
Zealand. What reason (in adaptation or relic nature) can one
find for the fact that one genus reaches from the far north to

192 THE MONOTYPIC GENERA [pt. ii

about halfway down New Zealand, while another begins there
and reaches the remainder of the distance? Further, these mono-
typic endemics have an average range of about 446 miles, and
in a varied country it is a little difficult to imagine conditions to
which they can be just suited in such a range.

If one take the families in groups of ten, in order of their size
in the world (as measured by the number of genera given in my
Dictionary), one finds that the column of monotype numbers
follows that of numbers of genera with wonderful closeness; the
first exception comes only at the seventeenth group of ten
families, a group including only 59 genera, or six per famil5^
Even beyond this the numbers continue closely parallel, and
there is only once an exception. The percentages also show clearly
that (just as with endemics) the greatest proportion of mono-
types is in the largest [i.e. on our hypotheses, in general the
oldest) families, falling steadily from 40 per cent, in the first
forty families to 30 per cent, in the final group of 131 very small
ones.

Analysing from my Dictionary, as corrected to date, the pro-
portion of monotypes in the various families, one finds that in
the families with over 100 genera the percentages vary between
28 and 56, With three-quarters of the M'hole total between 33
and 44. Those below 100 genera vary betAveen 11 and 68 per
cent., or tAvice as much, with three-quarters betAveen 23 and 50.
The percentage in the larger families is evidently a little higher,
as has already been pointed out.

There is a fair amount of difference, therefore, between indi-
vidual families. In the first ten, the largest percentage is in the
Asclepiads (54 per cent.), the loAvest in the Orchids (35 per cent.),
but there is not the least reason to suppose the former to be a
specially moribund family. Other families Avith more than 50 per
cent, of monotypes are Burseraceae, Lythraceae, Menisperm-
aceae, Portulacaceae, Saxifragaceae, Juncaceae, etc.

Explanations of the facts of monotypism have followed much
the same lines as those of endemism, the genera being regarded
as local adaptations or as relics, according to taste. But Avhat
has been pointed out above shoAvs that there is a very definite
arithmetical relationship betAveen monotypes and genera of
larger size, not only on the total, but also in very fair detail.
This alone is almost a conclusive argument against either of the
suppositions just mentioned as a general explanation, though
of course there must be many individual exceptions, better

CH. xvii] AND GENERA OF LARGER SIZE 193

explainable by their aid. How could local adaptations be gradu-
ated in this regular order, or how could there be a vast number
at the last stage of relicdom, and fewer and fewer at the stages
leading up to that, and that in every family or country?

Another great difficulty for the older explanations is provided
by the increase of monotypes, as of endemic genera and species,
as one goes southwards and outwards. Why should New Cale-
donia, the Mascarenes, and Juan Fernandez require so many
more per thousand square miles than the Sandwich Islands,
Formosa and Cuba, in similar northern latitudes, especially as
their non-endemic genera are in general very large and "suc-
cessful" world-ranging genera? AVhy should Chile have about
100 local monotypes, while there are only about 77 in Europe,
with more than ten times the area? Why should W^estern Asia
require so many more than Europe? and so on.

The only reasonable explanation of the very striking facts that
have been set forth in the last three chapters, so far as I can at
present see, is that provided by Age and Area with its corollary
Size and Space, that the smaller genera are as a rule the younger,
that they are probably the descendants of the larger genera,
that they gradually increase their area with their age, and that
as the area increases, so does the number of species, these also
increasing their area with their age. As a general explanation
of the phenomena seen in the distribution of plants about the
globe, this commends itself by its extreme simplicity, and by the
fact that it explains what has hitherto been regarded as an in-
soluble problem. Distribution is an extremely slow process,
allowing time for acclimatisation, and the effect of all the various
factors that act upon it is to cause it to take place at a regular
rate, so that it becomes a measure of age, or vice versa. Barriers
alone interfere with it, but they may be of many kinds.

Summary
The monotypic genera are very numerous, being 4853 out of
12,571 in the world, or 38-6 per cent. The ditypes are also nume-
rous, but are only 1632, or a drop of over 3000 from the mono-
types, while there is another drop to the 921 tritypes, and then
the numbers of genera of different sizes taper away in a long tail
to Astragalus at 1600 species. The mono- and di-types include
more than half of the total, and a very regular hollow curve is
formed. The individual families are arranged in the same way,

13

194 THE MONOTYPIC GENERA [pt. ii, ch. xvii

each commencing with a great number of monotypes, and giving
a hollow curve.

In the smaller areas of the world, like the single continents,
one finds the proportion of monotypes very high, while that of
dit3^pes is lower, and it falls off steadily to an insignificant figure
for the larger genera. In the genera found in both worlds, on
the other hand, the exact reverse is the case, and intermediate
phenomena show in intermediate areas.

The area occupied by a monotj^pe may vary enormously, but
in general is small; only 66 of them occur in both Old and New
Worlds.

The greatest percentage of monotypes is in the larger families,
and it diminishes steadil)^ with the lessening sizes of the families,
when these are taken in groups of 40.

This marked arithmetical relationship of the monotypes to
other genera shows that the usual explanations — that they are
relics, or that they are special adaptations — are in general in-
applicable, and that the explanation offered by Age and Area,
with its implications, that as a rule they are young beginners,
and probably descended from the larger genera, must in all
probability be correct for the great majorit}'.

CHAPTER XVIII

THE HOLLOW CURVE OF DISTRIBUTION

-Dy far the most remarkable feature that stands out through all
the work described in the preceding pages is what may be termed
the " Hollow Curve of Distribution." It was first noticed in 1912,
when working up the flora of Ceylon for the first paper upon Age
and Area (123). This flora of 1028 genera was composed of 573/1
(573 genera of one species in Ceylon), 17G/2, 85/3, 49/4, 36/5,
20/6, and so on, the niunbers becoming somewhat irregular after
six, but decreasing fairly regularly if taken in twos (genera of
7 and 8 species, 9 and 10, etc.). Having already the knowledge,
familiar to systematists, that genera of one and two sj^ecies were
the most numerous, it was thought that the regular decrease of
the numbers might be accidental, and time did not then permit
of comparisons with other floras^. The hollow curve, however,
appeared again in 1916, in coimting up the areas of distribution
of the endemics of New Zealand. Unlike Ceylon, New Zealand
was treated by actual measurement, and when the endemics
were divided into ten classes by area, it was found that the lowest
class, though it occupied much the smallest area (40 miles by
length of New Zealand against 120 for most classes), contained
much the largest number of species, having 168 out of 902, while
the ninth class came next with 128. The two classes contained
32-8 per cent, of the whole number of endemics (of New Zealand
proper), although their area was only barely 15 per cent, of the
total.

1 In actual fact, as may be quickly verified, all (or most) local floras
show the same thing, with their jrenera arranged in hollow curves when
grouped by the number of their (local) species. This is what one would
expect if genera are produced from other genera at a more or less uniform
rate, and in a more or less "casual" way. The subject will be treated in
greater detail in another place, and it will suffice for the present to call
attention to the fact that the hollow curve is regularly shown, as by the
Ceylon local flora (above, and cf. curve 4 on p. 237). The British flora shows
223/1, 90/2, 35/3. 32/4, Ifi/o, la/fi, 5/7, 7/8, 2/0, fi/lO, and so on to 71, the
numbers becoming rather irregular after (i. The flora of (.;aml)ridgeshire
(Babington, omitting Rubus, Uicraciuyn, and Salix) shows 210/1, (il,2, 3G/3,
21/4, 14/5, 6/6, and so on {I.e. curve 6). That of Wicken Fen, which is only
a very small area in the same county, is graduated in the same manner.
The flora of Italy (I.e. curve 9) shows the same thing, and so do the floras
of Greece, British India, New Zealand, the Bahama Islands, and others
that have been tested.

13—2

196 THE HOLLOW CURVE OF DISTRIBUTION [pt. ii

These results are shown on pp. 162, 237. The middle class for
New Zealand is rather high, but this is probably due, as I have
elsewhere shown, to the occurrence of Cook's Strait in the middle
of New Zealand, and it is worthy of notice, that New Zealand,
from which such strong evidence has been derived in support of
my contention that age is the main factor in distribution, shows
the most irregular curve that has as yet been met with in
examining many hundreds. The fact of the division of the islands
by straits, and the probable occurrence of several different in-
vasions of plants (127, 131) are likely enough to account for this.

As both these curves agreed in type, and as the figures for the
endemics of Ceylon, though only estimates and not actual
measurements, seemed to hint at something of the same kind,
my attention was thus roused, and especially so when the next
figures that I obtained, those for the distribution of the endemics
in the outlying islands of New Zealand (129, pp. 329, 331 ), showed
the same curve, but in a reversed direction, the maxima being
upon the largest areas. It was next shown by the endemics of
the Hawaiian Islands, where 47 per cent, were confined to one
island, and 20 per cent, more to two (out of seven), and the num-
bers rapidly diminished upwards (p. 162, curve 3); then by the
species of Callitris in Australia (130) and their local distribution,
and afterwards by other things.

Numerous instances of the hollow curve have been given above,
for example (in species first of all), in the distribution of the
species of Banunculus in New Zealand (pp. 153-6), in the
general distribution of the species of Cijvtandra and their local
distribution within the Hawaiian Islands (p. 160), the distribu-
tion of the Boraginaeeae in New Zealand (p. 161), of Olearia in
that country (p. 101), of Doona in Ceylon, of Cyanea and Pelea
in the Hawaiian Islands (p. 161 ), of the species of endemic and
non-endemic genera in the Hawaiian Islapds (p. 163), ot Exacwn,
Christisonia and Ebermaiera in India (p. 163), and so on. In-
numerable instances of its applicability could if necessary be
produced. Most of these curves are shown on p. 162.

In the same way, the curve applies to genera, and instances
have already been given, for example, in the geographical dis-
tribution of the Poiemoniaceae (p. 171), the Cistaceae (p. 172),
the Menispermaceae (p. 172), and the Monimiaeeae (p. 174).

The curve is thus a general feature of the distribution of
species by areas occupied, and goes to show that age is of enor-
mous importance in geographical distribution. In view of these

CH.xviii] THE HOLLOW CURVE OF DISTRIBUTION 197

facts, and of the striking way in which it has been found, in
regard, for instance, to the flora of New Zealand, that predictions
as to distribution may be made upon a basis of age only, and yet
be reasonably accurate, it would seem probable that age is by
far the principal factor in determining geographical distribution.

This of course simply means, as has already been explained,
that the resultant effect of the many factors that are operativ^e
upon any individual species (and still more upon any group of
ten allied forms) is so uniform, when long periods of time are
dealt with, that dispersal goes very largely with age. The great
difference between this and the older view is that we can no
longer look upon the dispersal of species as having reached its
limits. Before the rise of the theory of natural selection, as has
been pointed out on p. 3, the effects of age were recognised, but
in the last sixty years they have been more and more lost to
view. The figures that have been given above, however, show
that in reality they are perhaps the principal features that are
apparent in distribution.

But Size and Space also enters into the question, and if we
consider this principle also to be vahd, as indeed seems sho\vn
by the many cases of its application that have been given above,
then we shall expect that as Age, Size, and Space (or Area) go
together, the phenomena exhibited by Size will be more or less
like those exhibited by Space, inasmuch as Age is the only active
factor of the three. Actual examination soon shows tliat this is
the case, and that genera of one coimtry, endemic or not, ar-
ranged by sizes, form a hollow curve like those formed by species
in order of area; they begin with many monotypes, and a good
many ditypes, and taper off into a more or less long tail of larger
genera. This of course means that the hollow curve enters not
only into geographical distribution, but also into evolution, for
nothing but evolution could produce the size of a genus.

The hollow curve shows in the distribution into sizes of the
endemic genera of all islands (p. 17G), of the endemic genera of
individual islands (p. 175), of those of Brazil (p. 176), and other
countries. It shows again in the composition, by sizes of genera,
of the floras of Great Britain, Ceylon, New Zealand, India, etc.,
and shows in the division of these into portions of the country,
single families of reasonable size, and so on; it shows again in
the composition of the lists of genera with one, two, three, or
more species. Once more it shows in the composition of the
lists of genera confined to larger areas of the world, such as single

198 THE HOLLOW CURVE OF DISTRIBUTION [pt. ii

continents, and to the whole world (p. 185). It shows, and that
very conspicuously, in the composition of the various families
by sizes of genera (p. 186), as well as in the average size of the
largest genus in each family (taking the families in order of size,
p. 188). It shows in the four curves of percentages given on
p. 181. And it shows, finally, with great regularity of expression
in the curve for all the genera of flowering plants grouped by
sizes (p. 185), and in other features. There is no limit to the
number of instances that could if needful be produced.

Now this is really a very remarkable state of affairs, and that
it has not been discovered at a much earlier period can only be
attributed to the fact that the rise of the theory of natural
selection diverted effort from the lines which it is clear (cf. p, 3)
that it was beginning to follow in 1853. Until, however, the
theory of evolution Avas firmly established, it seems doubtful if
much could have come of any demonstration of the effects of
age. The clear arithmetical relationships that exist between the
various groups of plants, "Avides"' and endemics for example, are
only explicable if one consider that they are mutually related.
The Darwinian theory established for us the law of evolution,
and it now remains to carry the Avork a stage further.

It is somewhat difficult to perceive why the noAv clearly
demonstrated fact, that age is the most poAverful clement in the
dispersal of species, should rouse so much opposition. That an
older species should occupy more area than a younger one that is
closely related to it, seems almost axiomatic, and Avas cA^dently
clearly recognised by Lyell and Hooker (cf. p. 3). If tAvo species
A and B have much the same dispersal methods, and are suited
to much the same soils and climates, then it is clear that if Ave
call these three factors a, b, and c. the dispersal of these tAvo
species Avill be represented by the formula:

dispersal = {a + b + c) x age.

If the dispersal is the same, therefore, the age Avill be about the
same, while if the dispersal of A is greater than that of B, its age
Avill be greater. If avc transfer age to the left-hand side of the

equation, avc get - - = a + b + c. shoAving that dispersal

age
goes AA'ith age only. But age simply represents the total effect
of the operative factors a, b, and c, Avhich Avill be the greater the
longer the time during Avhich they have been acting.

For the last half century, however, we have been under the

CH. XVIII] THE HOLLOW CURVE OF DISTRIBUTION 199

sway of the theory of natural selection, which demands origin
of species upon large areas, as well as the occurrence of many
species that are "going under" in the struggle for existence.
The result has been, consequently, that the species of small areas
have been regarded as the failures, and this has derived support
from the fact that fossil botany shows that there are vast num-
bers of extinct forms. Most botanical work has been done in
the regions that were affected by the last glacial period, which
has left very many survivals in them (cf. p. 86, footnote). It is
not fully realised that though there may be perhaps a thousand
of such survivals, they are completely lost in the crowd when one
deals with the forms of limited area, or with the monotypic
genera as a w^hole. It would be absurd to apply the explanation
of relicdom in face of such facts as those given in Chapters xv
to XVII. Few people would now be found to express themselves
in support of natural selection as a cause for origin of species, but
though the premises of the argument are damaged or abandoned,
they hold strongly to the deductions that were made from them,
chief among which, in the present connection, is that species
have reached their limits of possible dispersal, and that those of
small area are the defeated in the struggle for existence.

So long as such a view was taken of distribution, so long would
it have seemed absurd to expect to get any result from statistical
investigations. But the figures that have actually been obtained,
and of which many instances are given above, show that what we
have called the hollow curve obtains throughout. It obtains, for
example, in the grand total of genera in the world, and for the
totals of genera in every single family : for the distribution of
endemics, and of local floras, whether for areas occupied by the
species, or for the sizes of the genera; for animals as well as for
plants. The hollow curve is apparently a universal principle of
distribution, Avhether it be distribution in space— geographical
distribution— or distribution in time— evolution. A species as
it increases in age, expands its area, while a genus increases its
number of species, the younger occupying smaller and smaller
areas, usually within the area of the first species, until that
becomes very large (and sometimes even then).

The very "important bearings of this work upon the general
theory of evolution must be left for later publications. It will
suffice to have called attention to the facts.

CHAPTER XIX

APPLICABILITY OF AGE AND AREA TO ANIMALS

At an early period of my studies of Age and Area, when once
I had found how universally operative it was in the Vegetable
Kingdom, it seemed to me that in all probability it must also
apply to animals, though perhaps with less force on account of
their capacity for movement. Accordingly, I asked Professor
J. Stanley Gardiner, F.R.S., for help, which was given in the
most unstinted manner, and for which I take this opportunity
of expressing my most grateful thanks. By his advice I investi-
gatecl some groups of Land Mollusca — animals whose locomotive
capacity is somewhat limited — and I found that their distribu-
tion agreed fairly closely with what would be expected under the
hypothesis of Age and Area. One or two other groups that he
also recommended showed the same thing. The great difficulty
in applying Age and Area to animals rests upon the fact that
Professor Stanley Gardiner pointed out, that in very many
groups either the systematic grouping or the geographical dis-
tribution is but imperfectly known, and that there are com-
paratively few groups in which our knowledge of both is fairly
complete. And of coiu'sc in applying a new principle like Age
and Area to the Animal Kingdom one must be very sure of one's
facts, and not leave it possible for any one to say that a more
complete knowledge of the subject would yield quite different
results.

At this stage I left the subject for a while, being much occupied
with the extension of its application to plants. At a later period
Professor Stanley Gardiner recommended me to apply for help
to Mr Edward Meyrick, F.R.S., the well-known investigator of
the Micro-Lepidoptera, who had at his command all the known
facts about the systematic grouping and geographical distribu-
tion of this group. Mr Meyrick was so kind as to furnish me with
the figures of the numbers of species that occurred upon New
Zealand and upon other islands, and the genera to which they
belonged, and from these I was able to determine that this group
also had closely followed Age and Area in its distribution; not
so closely, perhaps, as the plants, but with sufficient approxima-
tion for the fact to be unmistakable.

PT. II, CH. XIX] AGE AND AREA IN ANIMALS

201

humbtr 0} sp«{.i,«.s pf g4ny.s.

Hollow curves of distribution of sizes of genera in various families of
animals, plotted in the same way as those for plants on p. 187. The
almost exact parallelism of the curves for both animals and plants
may be seen in the fig. on p. 237.

202 APPLICABILITY OF AGE AND AREA [pt. ii

It was thus becoming gradually clear to me that -with perhaps
rather greater deviations than in plants, Age and Area was also
a rule for animals, and in the latter half of 1920 I began to write
a paper, which I hoped might be published by one of the zoo-
logical journals, upon the application of Age and Area to such
questions. But as about this time the work upon the "hollow
curve," described above, began to show promise of very striking
results, I decided that it would be better to leave the matter
alone for the meanwhile.

Since finding that the hollow curve is practically universal in
the distribution, and also in the actual evolution, of plants, and
that it can be traced by merely adding up, and sorting into sizes,
the genera that make up any group, the application of the theory
to animals has been rendered a much more simple matter.
Professor Stanley Gardiner once more came to my assistance,
and gave me a start with the names of reliable catalogues of
genera and species, such as those of Boulenger (Lizards, Snakes,
Amphibia, Perciform Fish); and ]\Iiss Taylor, Librarian of the
Balfour Library, Cambridge, showed me a number of others.
Finally, Professor Stanley Gardiner recommended me to apply
to Dr Hugh Scott, the Curator in Entomology, and an authority
upon the Beetles. With his assistance, which was freely and
liberally given, I have been able to enumerate a number of
families of this group.

The result of all these enumerations is to show that the
"hollow curve" is as well mnrked in zoology as in botany, for
I have found it to sJiow clearly even in such small groups as
the lizards and the snakes (fig. on p. 201), and it is as evident
in the Ungulate Mammals (Lj'dekker, 1916), the Chiroptera
(Anderson, 1912), the Amphipodous Crustacea (Bate, 1862),
the Marsupials (Oldfield Thomas, 1888), the Mycetozoa (Lister,
1894), and even in such small groups as the Cyclostomatous
Polyzoa (Buck, 1875).

Some of these curves are shown in the fig. on p. 201. As it
might be thought that parasitic animals would show a different
curve, I counted the Ichneumonidae (de Dalla Torre, Cat. Hijm.,
1901) upon Dr Scott's suggestion, and the illustration shows
that this group also exhibits the hollow curve, though there is
one irregularity shown at an earlier stage than usual. There are
60 with four species and only 50 with three, whereas the numbers
usually do not show much irregularity till one comes down to
about 20.

CH. XIX] TO ANIMALS 203

In the Beetles the curve shows clearly in all groups counted.
For example, the Tenebrionidae (Gebien, Col. Cat. 15, 22, 28, 37,
1910-11) show 489/1, 154/2, 103/3, 73/4, 40/5, 48/6, 32/7, 32/8,
24/9, 10/10, and so on; the same kind of figures are shown by the
Coccinellidae (Gemminger and Harold, Col. Cat. 1876), and the
Chrysomelidae {lb. 1876), as well as other smaller groups that
were counted. Unfortunately, I counted the Coccinellidae and
Chrysomelidae from an old catalogue, and the new catalogue is
not yet sufficiently complete to enable a comparison to be insti-
tuted. The result of comparing floras of different dates and by
different types of systematists, however, leads one to suppose
that the result Avould be very similar.

Not only does the curve shoAV in general lists of the animals of
the world, like these, but also, just as in the case of plants, it
can be seen in local faunas. Thus taking Barrett's British Lepi-
doptera (London, 1905), one finds that the genera whose names
begin with A, B, C or D show 62/1 (62 of one species in Britain),
28/2, 13/3, 4/5, and so on; those with E, G, H or L show 54/1,
18/2, 14/3, 10/5, and so on; those with M, N, O or P 63/1, 15/2,
9/3, 8/5, and so on; and those with R to Z 36/1, 7/2, 6/3, 4/5,
and so on. The total shows 215/1, 68/2, 42/3, 26/5, and so on.

The British Echinoderms (Bell, 1892) show 39/1, 16/2, 5/3,
and so on. Even so small a group as the British Spiders does its
little best to follow the curve. It is clear that the rule holds as
well for animals as for plants, as Avill be seen by examining
the fig. on p. 237, Avhere the curves for animals and for plants,
for local floras and for local faunas, etc., are mixed up together.
The same rules have evidently guided the evolution and the
geographical distribution of both groups, and the extraordinary
parallelism of the curves goes to show that both evolution and
geographical distribution were largely guided by factors that
acted in a mechanical Avay. The very interesting suggestion has
been made that the parallelism may be due to the fact that
animals are (in the long run) a function of plants. But it does
not seem to me that this is quite sufficient to explain, for instance,
the fact that the Ichneumons show^ a curve parallel to the others.

CHAPTER XX

THE ORIGIN OF SPECIES

No subject in biology has been the cause of such excited debate
and controversy as has this, since the pubHcation of Darwin's
Origin of Species in 1859. Were it not that Age and Area seems
to have some not unimportant bearings upon the subject, we
should not bring into this book so thorny a matter of dispute.

If in this chapter or elsewhere I seem severely to criticise the
Darwinian theory, it is not because I do not appreciate its many
strong points, nor is it that I am trying to throw contempt upon
it. The theory is as legitimate a subject for criticism as is any
other. It does not seem to me that it has been properly realised
that the "Darwinian theory" has two separate sides. Darwin's
immortal service to science lies in the fact that he established
the theory of E^'olution— until then regarded with contempt—
in an unshakeable position, which all subsequent research has
only strengthened. But to establish it he had to invent some
machinery by means of which it might be supposed to work, and
for this purpose he devised the very simple and beautiful mechan-
ism of natural selection. So strong was the a jniori evidence in
favour of this, and so well did natural selection seem to explain
almost everything in animated nature, that within a short time
it was accepted all round, and with it the theory of evolution,
which is now established as the rider of thought, not only in the
scientific world, but outside of it. The mechanism of natural
selection has, however, been for a long time subject to an in-
creasing severity of criticism, and as a working theory is now
becoming largely moribund.

No theory as yet brought forward in biology has been for so
long a time a stimulant to research, nor has any proved so
fruitful in educing valuable work. It may suffice to call attention
to the very different position of biology in 1859, and at the
present time. It is hardly too much to say that all, or nearly
all, the work done during that time owes its incejition, at least,
to the influence of the Darwinian theory. Not only so, but it
has produced the most far-reaching effects in all branches of
human thought.

The literature in praise of the theory is already very bulky,

PT. II, CH. XX] THE ORIGIN OF SPECIES 205

and my present object is not to add to it, but to criticise certain
aspects of the theory, and to show in what directions it has
failed to give us satisfactory explanations of phenomena, or
fruitful subsidiary hypotheses upon which to work. To suggest
a doubt of its enormous value in the advance of knowledge would
be to rank all the workers of the last sixty years as upon the
intellectual level of the Bushman or the Esquimaux.

There can be little doubt, however, that during recent years
the theory of natural selection has become what we may call a
limiting factor in the progress of biology, and the time seems to
me to have arrived when we ought to consider the advice given
by Sir Joseph Hooker soon after his first acceptance of the
theory :

"The advocate of creation by variation may have to stretch
his imagination to account for such gaps in a homogeneous
system as will resolve its members into genera, classes, and
orders, but in doing so he is only expanding the principle which
both theorists {i.e. special creationists and natural selectionists)
allow to have operated in the resolution of some groups of indi-
viduals into varieties:... Natural Selection explains things better
...it is to this latter that the naturalist should look... holding
himself ready to lay it down when it shall prove as useless for
the further advance of science, as the long serviceable theory of
special creations, founded on genetic resemblances, now appears
to me to be."

Went (112, p. 270) has said that we ought to drop all teleo-
logical explanations, and not consider nature as having any aim.
This may seem somewhat drastic, but as yet we are without
any e^•idence as to Avhat is the aim of nature, though the work
that has been described above seems to show that she perhaps
has one, for it seems evident that the evolutionary clock was
wound up to run on a very definite plan. But for what nature
is aiming in this definite way, we are completely ignorant, and
it will, it seems to me, prove more wise, in the present state of
science, to follow Went's advice, leaving out of serious account
as yet any suppositions as to the ultimate aim of nature.

We have shown in the preceding chapters that the phenomena
of distribution, whether it be distribution in space of species
and genera, or distribution in time, as exhibited by the grouping
of species into genera of various sizes, can be graphically repre-
sented by hollow curves, which could if required be produced
in tens of thousands. It is clear that such a general phenomenon
must have a general explanation, and that this must be largely

206 THE ORIGIN OF SPECIES [pt. ii

mechanical. These phenomena, as has already been shown, can-
not be satisfactorily explained by any of the many suppositions
that have long been current, based upon natural selection. A
differentiating cause like natural selection could not produce
such uniformity of expression, and at the present time, the only
feasible explanation in the field seems to be that provided by
Age and Area, which explains the species and genera as developed
in successive order and gradually expanding their area (and
their number of species, in the case of genera) as time goes on.

But if this explanation be correct, it is clear that the smaller
the area occupied by a species, the younger on the average will
it be, in its own circle of affinity. The only logical conclusion to
this is that in general the minimum area is that occupied by
species just commencing their life as such. But, as already shown
(pp. 54, 55), this may be very small indeed; a species may be
easily hmited to a dozen or two of individuals, if it does not
actually begin wirh one or the progeny of one^ It is clear that
we cannot regard as the formative cause of the genesis of the
species a struggle for existence resulting in the conservation of
favourable variations, especially if these be of the kind that we
understand as infinitesimal or fluctuating.

The new species just commencing will have to undergo a
struggle for existence, usually of a very strenuous kind, imme-
diately, and if in any way unsuitable to the conditions that pre-
vail at the exact place and time of its birth, will at once die out,
as a rule leaving no trace. If it survive, it may contiiaie to
spread so long as it finds conditions in which it can grow, and
the ultimate area that it covers will depend upon that and upon
its age (cf. de Vries, below, p. 227).

One of Darwin's innumerable services to the cause of science
was to call attention to the struggle for existence. Even he,
however, perhaps hardly emphasised sufficiently the intensity
of that which j^robably takes place at the birth of a species,
except upon more or less virgin soil. If in any xvay unsuited to
the conditions obtaining at the time and place, it will be all but
certain to succumb. IMere heredity, however, will tend to make
it more or less suitable. But even if well suited, the new species

1 A few days before I left Rio, Dr Lofgren found, on a little island about
three miles off the coast, a new and very distinct Rhipsalis, of enormous
size. He told me that there were only four examples on the island. I could
only find, on the summit of Ritigala, about a dozen examples of Coleus
elongatus (p. 54). And cf. Didymocarpus and Christisonia, p. 151.

CH. XX] THE ORIGIN OF SPECIES 207

must probably have some degree of luck. It may not chance
upon a spot where it can grow, the ground being'alreadv fully
occupied by a closed association of plants; or it may easily be
destroyed by a fire or a flood or other accident (cf. Didtjmocarpus
and Christisonia on p. 151).

The older view, that species arose by gradual accumula-
tion of infinitesimal or fluctuating variations (up and down
variations, such as always show in any character, as when a leaf
varies in length from one to one and a half inches on the same
species), is now dying out in favour of mutations, or sudden
alterations of form which have their origin in changes that have
occurred in the material bearers of heredity. It is conceivable
that any changes, however great, might be brought about by
the accumulation of fluctuating variation, provided (1) that the
variations were fully hereditary, (2) that they were not linear,
showing the same character in greater or less degree, but dif-
ferentiating, a simple leaf, for example, showing a tendency to
compoundness, (3) that the necessary variations appeared, and
(4) that natural selection should be able to act, i.e. that the
appearance of the variation should give to the plant or plants
possessing it such advantages as should ensure their survival in
at least the majority of cases.

In regard to the first supposition, so far as we know, infini-
tesimal variation is not fully hereditary, but always regresses or
falls back, so that while one may make great impro\-ement.s by
selection (as, for example, in the speed of trotting horses, oi the
content of sugar in the root of the beet) there always comes a
point beyond which one cannot go. It is sometimes stated that
the wonderful varieties of our cultivated crops owe their origin
to the selection of infinitesimal variations, and that when left
to themselves they go back to the wild form, but this is not the
case, however; as Hooker long ago pointed out, the cultivated
apple goes back, not to the crab, as is popularly supposed, but
to crab types of cultivated forms.

These facts agree with ordinary observation, which gi\es no
reason to suppose that continuous change is going on. Hooker
(55 a, p. x) has so Avell put the argument in favour of the general
permanence of species that it would be presumptuous to try to
better it (I have shortened it).

"(1) The fact that the amount of change produced by ex-
ternal causes does not warrant our assuming the contrary as a
general law.

208 THE ORIGIN OF SPECIES [pt. ii

(2) The permanence with which widely dispersed species re-
tain their characters, whether naturally or artificially dispersed.

(3) With comparatively few exceptions, plants are confined
within well-marked limits; sporadics (discontinuously dis-
tributed) are rare. If they varied indefinitely, sporadic distribu-
tion would be the rule.

(4) A multitude of allied species of plants grow close together
wthout any interchange of specific character.

(5) The individuals that inhabit the circumference of the area
occupied by a species are not found passing into other species,
but ceasing abruptly... may meet or overlap similar species.

[(6) A negative argument in favour of distribution from one
centre,]

(7) The species of the lowest orders (now families) are not
only the most Avidely distributed, but their specific characters
are not modified by the greatest changes of climate.

(8) The fact that no plant has been acclimated in England
\vithin the experience of man."

A httle consideration will show that these arguments, with
the possible exception of the eighth and last, are as sound to-day
as Avhen they were written, and all the work and experience of
Jordan (62), Johannsen (61), and the many ecological writers
of recent years has but added strength to them. But the stronger
they become, the greater is the argument in favour of sudden
change by miitation.

The second and third provisos (about fluctuating variation)
above given really go together, for we have no evidence that
differentiating variations can appear at all, unless so large and
sudden that they are really mutations, not connected with the
preceding form by infinitesimal stages. Fluctuating or in-
finitesimal variation is simply up and down in the same charac-
ters; one never finds a leaf varying by imperceptible stages in
the direction of a tendril, or of compoundness, or towards a
pitcher.

A great difficulty for the theory of natural selection, though
indeed it is no less for any other theory, is to explain the occur-
rence of correlated variations. Why, Avhen a plant produces
tendrils, or climbing leaves, should its stem at the same time be
weak and flexible? Yet the one would be useless, if not dis-
advantageous, without the other. It often happens, in these
correlated characters, that while one confers advantage, the
other is disadvantageous. It is not altogether wise or reasonable
to talk about advantage as having determined progress in nature.
To take the single instance of Coleus elongatus (p. 151), its two

CH. XX] THE ORIGIN OF SPECIES 209

most marked characters of difference from its allies are its
peculiar inflorescence and different calyx. Neither could be
"useful" under any conceivable circumstances, nor could any
of the other characters of difference between it and C. harhatus,
but neither can one say that any are disadvantageous. But to
get the one calyx from the other means that one sepal must
narrow, while the others broaden, and all experience of fluctu-
ating variation shows that all homologous members vary in the
same direction, so that nothing but a mutation can produce this
difference.

There are so many characters in plants to which no amount of
ingenuity can attach any quality of advantage or the reverse,
that though at first the natural selectionists said that avc did
not know enough about them, it soon became evident that this
would not serve as a general explanation, and it was then said
that they were correlated with useful characters. Gradually,
however, it has been realised that the bulk of morphological
characters come under this head, and that the useless structural
characters in plants outnumber the useful by an enormous
percentage.

Lastly comes the question under the fourth proviso above,
whether natural selection can act upon the first beginnings
of characters. While there are some cases in which it might be
imagined to do so, there is no doubt that in the vast majority
of cases, where no use can be even suggested for the mature
character, it could not take hold of the first rudimentary be-
ginnings. Take, for example, at random, the pollen patterns in
Acanthaceae (genus and subtribe characters), the adnation in
Solanaceae (genus), the adventitious shoots in Podostemaceae
(family), the translators of the pollen in Asclcpiadaceae (family),
the various aestivations of the corolla (genus and family), the
dehiscent or indehiscent fruit (ditto), the monoclinous or diclinous
flower (ditto), the ruminate endosperm of Anonaceae (ditto),
the phylloclades of Asparagus (genus), the valvular opening of
anthers in Berberidaceae (family), the septifragal opening of
capsule (tribal), the "boragoid" inflorescence (family or genus),
tJie tubers of potato, the bulbils of Agavi\ and hundreds more.
One cannot conceive of natural selection getting any grip upon
the early stages of these, and indeed, in great numbers of these
and other characters, early stages are not conceivable.

Not only so, but many things that were once explained as
adaptations to something or another are now proving to be in

210 THE ORIGIN OF SPECIES [pt. ii

reality of little or no value to the organism concerned. It will
suffice to recall to memory the controversies about Drosera and
its insectivorous habits, the work of Kamerling upon xerophily,
or the characters of the Podostemaceae and Tristichaceae (124),
which are extremely striking and varied, though there are no
differences in conditions to which to be adapted. Stomata with
an outer pit entrance undoubtedly diminish transpiration, and
were once considered an adaptation to that end, but one finds
them on the inner wall of the fruit in the opium poppy, where
transpiration cannot matter (112). Ant-plants were supposed
to gain from their association with ants, but one may see the
Cecropia flourishing without ants all over the forests of Southern
Brazil, and the ants bring aphides, which must do much harm
to the plants. Epiphytes were supposed to be a particular
adaptation, till Schimper showed that plants became epiphytic
when they had three properties in common — easily dispersed
seeds, clasping roots, and capacity to resist drought for long
periods. And so on; the old adaptation explanation has been
shown to be of service in many fewer cases than had been
supposed the case.

There can be no doubt that the idea of adaptation was pushed
to extremes, and that adaptations were found in many features
that have since proved to be almost or quite indifferent. Went
(112, p. 260) has treated this subject so fully that there is no
need to repeat his criticisms, and he has also ])ointed out that
when real adaptation exists, it is chiefly in plants that live under
extreme conditions, and that it is rare in mesophytic types, to
which probably the bulk of plants belong. It is quite possible
that it is in this way that one may explain the fact that in the
Bahamas the local endemics are almost as widely distributed as
the "wides" (p. 64),

Another great difficulty for natural selection is that in many
cases the distinguishing characters do not appear (119) until the
struggle for existence is long over, for there is no doubt that the
vast proportion of the mortality is among the young seedlings.
What possible difference can it make to a plant that does not
flower till it is thirty years old, to take a single instance, whether
its calyx is smooth or ribbed?

The' fact that allied species usually live near together is a
strong general argument against the idea that advantage has
anything to do (in any important measure) with the origin of
most species. Another is that for selection to produce any great

CH. XX] THE ORIGLN OF SPECIES 211

effect, it should be between large numbers, whereas a plant can-
not on the average have more than six like itself around it.
^ A consideration of the instances just given, or of still more
"important" differences, such as that between the embryo in
Dicotyledons and Monocotyledons, soon shows that infinitesimal
or fluctuating variation, though it occurs in every character of
every plant, is inconceivable as a means of effecting the great
differences that actually exist in the vegetable kingdom. In-
finitesimal variations would at once be lost by crossing with
their surrounding unmodified neighbours, and only if all were
modified in the same direction by the action of some definite
cause, e.g. the environment, would there be any likelihood of
the survival of the new form. And even then, it is hardly con-
ceivable that such changes as those instanced above should take
place in gradual stages.

The y'lQVf that evolution is not directly guided by the need
of response to the actual necessities of plants, but' is a more
mechanical process, going on in compa^ati^'e indifference to
them, but with the disadvantageous ^'ariations at once thrown
out by natural selection, has been gaining in definition for many
years, especially since the rise of the study of genetics on Men-
delian lines; and the "hollow curve" observations, described
above, seem to show clearly that it has followed a definite more
or less determined course.

Nothing but mutation, understanding by that a change of
measurable amount, hereditary, not connected by infinitesimal
stages with the more "tj'pical" form of the parent, and usually
differentiating, seems capable of explaining the bulk of the
specific, generic, and family characters that at present exist.

Large mutations, often covering se^^eraI characters of a plant,
are by no means unknown, and go by the name of sports. Actual
observation has shown that a great number of these are here-
ditary, as in the well-known instance of the cockscomb. But
that such sports can give rise to new species has been strenuously
denied, because no instance of their undoubted survival in nature
has been recorded. But, so far as I am aware, no instance of the
formation of even a small variety by natural selection of in-
finitesimal or fluctuating variations has been recorded, and the
theory was accepted on account of its a priori probability. When
this difficulty is cast up to the supporters of natural selection,
they insist that there has not been time enough for the formation
of anything since man began to observe such things. But, as

212 THE ORIGIN OF SPECIES [pt. ii

we shall endeavour to show, the same reply is valid in the case
of mutation. To expect to see the formation of a new species,
i.e. the survival of a mutation, in the short time since man began
to observe such matters, is rather sanguine than reasonable, and
especially in the north, where the adjustment of plant associ-
ations to the environment is probably very perfect, and where
consequently the establishment of a new form is correspondingly
difficult. Lord Rayleigh has estimated the period since the
Eocene alone, which covers but a portion of that occupied in
the evolution of the higher plants, at 30,000,000 years. But if
we suppose one mutation in 50 years to survive, we should get
the whole of the existing 160,000 species of flowering plants in
8,000,000 years, which is only 26 per cent, of that time. And
this mutation, be it remembered, may appear upon any small
spot anywhere in the world, most of which is not under sufficiently
close observation for us to be able to say whether or not any of
the many species that are confined to very minute areas has
arisen within the history of human record. If Tribidus ala-
cranensis (p. 152), or one of the other two Alacran species, has,
as is possible, arisen in the last 50 years, then there is no need
for any more species to arise for 50 (perhaps 150) years to come,
to keep up nature's average rate of evolution.

When one considers how difficult it is for seed to get a chance
of germinating, growing, and surviving upon any given spot,
well covered, as most spots are, with a dense association of plants
that have already pro\'ed their suitability to the locality and its
conditions, it is clear that a new form must have the most com-
plete suitability at its birth to the local conditions, to get any
foothold. Not only so, but it must suit those conditions as they
will be modified by its OAvn appearance and addition to the
association of plants already there. Clearly, therefore, to talk
about advantage as having guided its evolution is to go some-
what beyond the warranty afforded by any of the facts as yet
at our disposal.

Man can, and does, easily propagate a noA'elty^ by clearing
the ground of rivals, but in nature this will rarely happen. It
viay be that the very common presence of young species upon
islands and upon mountains is due to the fact that these places.

1 "We have no reason to suppose that we have violated nature's laws in
producing a new variety of wheat — we may have only anticipated them;
nor is its constitution impaired because it cannot, unaided, perpetuate its
race; it is in as sound and unbroken health and vigour during its life as.
any wild variety is " (55 6, p. ix).

CH. XX] THE ORIGIN OF SPECIES 213

being somewhat isolated, have comparatively small floras, which
have not, therefore, been able as yet to form very elaborate
plant societies suited to their various conditions, and into which,
therefore, a newcomer ma}^ more easily enter. In the same way,
the frequent, and apparently quite casual, appearance of young
and localised species in the great forests may be due simply to
the fact that the fall of a great tree has for the time so changed
the conditions as to give the newcomer a better chance of estab-
lisliing itself before the old conditions are completely restored.
The further out one goes, the smaller on the average does the
number of species j^er genus become, aiid perhaps therefore the
plant societies may tend to be more open.

Whether a new form upon its aj^pearance will or will not
survive, will depend chiefly upon natural selection, for it will
at once have a struggle for existence of the most remorseless
kind. It will also depend appreciably upon mere chance (cf.
Didymocarpus and Christisonia on p. 151). A fire or a flood may
easily kill it out, howe\'er perfectly suited to its environment it
may be.

As our object in the present work is simply to criticise some
of the directions in which existing theories do not seem properly
to meet the facts, and to suggest some directions in which it is
conceivable that they may be improved, there is no need to go
into an}^ discussion of possible causes of mutation. If, as is not
impossible, they depend immediately upon some chemical change
that has somewhere taken place, one can understand why
changes should be mutational, for chemical change docs not
usually take place by continuous variation.

On account of the insuperable difficulties in the way of evo-
lution by means of the natural selection of infinitesimal variations,
opinion has for a long time been steadily coming round to favour
the idea of change by mutation. Even the most enthusiastic
supporters of infinitesimal variation now generally begin with a
measurable change, impro\4ng it afterwards by the old method.
A recent writer of this school, for exam])le, cites a change from
inches to feet as an infinitesimal ^-ariation.

The A\'ork upon Age and Area outlined in Part I provides, as
has already been indicated, strong arguments against infinitesimal
variation, and the further work gwaw in the last few chapters,
which seems to show that when one deals with large numbers
and the long run cA-olution of new genera and species, and their
distribution about the world, is very much a process which has

214 THE ORIGIN OF SPECIES [pt. ii

gone onward in a mechanical way, and whose progress can to
some extent be predicted from the laws of probability, supple-
mented by the principles of Age and Area, Size and Space, etc.,
provides a still stronger argument. If the "hollow curve" type
of distribution of numbers of families in the world, of numbers
of genera in families, of nmnbers of species in genera, of dis-
tribution of families, genera and species by area, of distribution
of genera in a given flora, of the bulk of the phenomena of evo-
lution and geographical distribution, etc., held only for grand
totals, it might still be possible to say that natural selection had
had much to do with the guiding of evolution, and that simply
because one was dealing with very large numbers the final result
came out more or less in accordance with the laws of probability
and of compound interest. But when, as has been shown, this
result is exhibited family by family, genus by genus, country
by country, and in animals as well as plants, it seems clear that
in general evolution and distribution, in some detail, have
followed "mechanical" laws, some of which, perhaps, in the
shape in which we have described them — as Age and Area, Size
and Space, etc. — the work described in this book may have done
something to bring into more clear definition.

One cannot imagine species or genera arising by gradual change,
and producing such an arrangement of "wheels within wheels"
as that shoAvn in the figure upon p. 156, or such curves as those
upon pp. 177 and 187, with the monotypes in a fairly definite
relation to the ditypes, these to the tritypes, and so on, the
curve practically always turning the corner between 3 and 5. To
produce such an arrangement by gradual variation, natural
selection is evidently incompetent, and some definite law to guide
it, at i^resent inscrutable, is required. In this connection one
must not forget that very strong evidence against such a sup-
position is provided by the fact that one finds very few con-
tinuous really intermediate stages, whether living or fossil, be-
tween species or between genera; in the enormous majority of
cases they are discontinuous. One may easily find species that
have say four characters of one genus and five of another, or
varieties behaving in the same way between species, but really
intermediate characters are very rare; and indeed, as avc ha^e
pointed out above, the\^ are frequently impossible.

We shall see in Chapter xxii that the hollow curve really
represents an approximation to the compound interest rule, and
one cannot imagine it to arise by continuous variation, though

CH. XX] THE ORIGIN OF SPECIES 215

one can imagine a genus arising from another by successive
mutation of a large number of the characters of the latter.

But if evolution be thus to an appreciable, if not (as Jeems
more probable) a very large, extent predetermined and governed
m Its unfoldmg largely by definite laws, or by mechanical con-
siderations like age, then it is clear that it is no longer safe to
consider that advantage to the species has had anything to do
with the actual evolution of that species, though" it will have
determined to a very large extent whether or not that species
shall survive. It maij have been directly concerned in the evolu-
tion, but it will be safer to leave it out of consideration, and to
study evolution in much more detail before committing our-
selves. This study must be especially from an experimental
standpoint, perhaps largely Mcndelian, and we must, it seems
to me, work without any ulterior idea of any aim to which
evolution may be directed (even the very local one of immediate
advantage to the species), until we really possess some facts
upon which we may reconstruct a theory of its operations. The
work described in this book is largely iconoclastic, and I do not
propose, in the present volume, to try to substitute any new
theory of evolution for that which has for so long held the field,
but merely to suggest a point of detail in which the latter theory
may in my opinion be altered with advantage, by the acceptance
of the theor,y of mutation, whilst in a later work I shall attempt
to bring forward some of the conclusions about evolution to
which the latest extension of the work upon Age and Area
has led.

If we remove advantage from the list of factors that may be
operative in evolution— and it is clear that at most it can only
be a small one— then it is evident that the mutations that dis-
tinguish species from one another cannot proceed in easy stages,
unless there be, as is of course by no means impossible, some at
present inscrutable law guiding them. The whole change, it
would seem, must take place at once. And this brings us to the
question of how large a mutation may be.

Size of Mutations. Many people think that a mutation must
be very small, like the differences in the ".lordauian" species of
Erophila verna which are so numerous in Europe, or in the
British Ruhi or Hieracia. My own opinion, which I ha\-e held
for the last eighteen years, and have published on various
occasions (especially in 123, p. 329), is that this is simply placing
an unnecessary handicap, for which there is no positi\'e evidence,

216 THE ORIGIN OF SPECIES [pt. ii

upon the theory of mutation. We have no evidence to show that
a Jordanian species will proceed further towards a Linnean
species. One cannot imagine the 11 Doo?ias, or the 15 species of
Stemonoporus in Ceylon (p. 152) arising in this way. The
Jordanian varieties show the same phenomena of dispersal as
do the Linnean species of Ceylon and elsewhere, and often
occupy as large areas, while they still remain true-breeding, and
show no sign of variation. As a general rule, it is not hard to
place a Jordanian species in its proper "Linnean" aggregate.

In Aaew of the large mutations that have been recorded, e.g.
CapseUa Heegeri (104) and others (cf. list in 39, p. 308), and
upon general grounds of comparison of the characters used in
systematic work upon the classification of the flowering plants,
it seems to me that mutations may at times be of the necessary
size to give rise at once to Linnean species. One cannot con-
ceive of the many species of Ranunculus in New Zealand, for
example, arising by the gradual separation of Jordanian varieties,
especially when these breed true. ^Ye have no evidence to
show that the intermediate forms, as would be necessary on this
hypothesis, die out. The struggle for existence comes at the
moment of birth of a species, and if it survives it may spread.

The view that mutations are necessarily small rests upon the
opinion, often put forward as if it were a general rule {e.g. 67),
that a Linnean species consists of a great assemblage of micro-
species, which breed true, as has been shown to be the case in
Erophila verna, for example. But this opinion requires a com-
plete re\'ision in \\e^y of the facts that have been set forth above
in regard to Age and Area. A species can only consist of such
an assemblage, obviously, if it consist of many individuals, and
occupy a large area of ground. Now in the north temperate zone,
where most of our botanical research is carried on, this is in fact
true of nearly all species; and only a few are localised, for in-
stance in the Alps or the Rockies, or to a less extent in the
plains, particularly of North America and West Asia. These
localised species have been looked upon as relics or special local
adaptations, and often disregarded from an evolutionary point
of view. But the work that has been done upon Age and Area
shows that such species, except to some extent within the range
of the effects of the glacial periods, must be regarded as young
beginners. Now in their case, where often the Avhole species is
only represented by a few individuals, it is clear that unless every
plant or two is different in hereditary characters, the species

CH. XX] THE ORIGIN OF SPECIES 217

cannot be composed of many true-breeding micro-species, but
that the formation of these must be later in the hfe of a species
than the formation of the species itself, and that it is after it is
formed that a species breaks up into micro-species, not that a
species is formed by the accumulation of micro-differences. This
agrees with what Bateson has said in his Presidential Address
(6); and simply expresses what has long been an axiom with
workers in ordinary systematic botany, that it is in large and
widely distributed species that much \ariation is found, '\vork-
ing in regions where most species actually occupy fairly large
areas, people have acquired an exaggerated opinion of the
variability of Linnean species, and unless it can be shown, by
genetic or other investigations, that local Linnean species, whicJi
exist in enormous numbers, especially in the south, are equally
variable, we must prefer to go upon the positi^•e facts shown
by Age and Area, confirmed as they are by the ordinary experi-
ence of every systematist.

It has long been the fashion to sneer at the "mere systema-
tist," and to regard him simply as a useful hod-carrier for the
real work of Botany, and this especially since the incoming of
modern theories of evolution, of which, by a kind of instinct, he
has rarely been a supporter in any enthusiastic way — in itself
an offence to those who think that by this or that theory botany
will at last come to an end of its difficult and slow beginnings.
No great systematist has taken up, for example, the modern cult
that the only species that are species, and that are worth con-
sideration, are the minute varieties of Jordan and other writers.
It will be worth while, in this connection, to quote some of the
axioms of the great systematists, as they are in danger of being
forgotten in the enthusiasm for the study of micro-species. For
example, Darwin uses as headlines in the Origin of Species the
following, which have never been disputed. "Wide-ranging,
much diffused, and common species vary most." "Species of
the larger genera in each country vary more frequently than the
species of the smaller genera." "Many of the species included
Avithin the larger genera resemble varieties in being \ery closely,
but unequally, related to each other, and in ha\-ing restricted
ranges."

From Hooker (55 a and b) I take (order in his sense is now
called family) "The varying species are relatively most numerous
in those classes, orders, and genera, which are the simplest in
structure." "As with species, so Avith genera and orders... upon

218 THE ORIGIN OF SPECIES [pt. ii

the whole those are the best limited which consist of plants of
complex floral structure." "Those classes and orders which are
the least complex in organisation are the most widely distributed,
that is to say they contain a larger proportion of widely diffused
species. ...This tendency of the least complex species to be most
widely diffused is most marked in Acotyledons (Cryptogams),
and least so in Dicotyledons." "The most widely distributed
and commonest species are the least modified."

It is clear, after reading these axioms, that another explana-
tion of the greater commonness of new (endemic) species upon
islands, southern land masses, and mountains is thus opened,
and one which may prove to be of great importance. Age and
Area shows that these widely distributed forms, which are the
most variable, are the oldest, and probably the parents of the
forms of lesser distribution. But at the edge of the dispersal of
any genus or other group, one will get, most markedly, the
oldest types; these being the most variable, will be the most
likely to give rise to new forms, and this, with the probable
comparative openness of the associations, may be the simplest
explanation of the frequency of endemics in the regions we have
indicated. A cursory examination of a number of genera shows
that this is very probably a general rule, but it would lead too
far to go into it in more detail at present; this must be left for
later work.

There is as yet practically no evidence that several mutations
are required to form a Linnean species. We have no reason to
say that a new and strictly local species is appreciably better
adapted, in the great majority of cases, than the older one,
unless for the conditions in which it first finds itself upon its
evolution. If species A give rise to species B at a certain point

AAAAAAAAAAAAAAAAAAAAAAA X

B

then, unless B is suited to the conditions that obtain at that
point in the year in which it was evolved, it is going to die out
again. For the immediate conditions at B, then, it may be
better adapted than A (as for example, perhaps, the endemics
of the Bahamas, p. 64), but when both species arrive at .Y, there
is no reason why B should be better adapted than A to the con-
ditions there. It will be mainly a matter of chance.

This being so, there seems no reason why intermediate muta-
tions, if they were formed, should die out, especially as the

CH. XX] THE ORIGIN OF SPECIES 219

original species, which we must look upon as the probable parent
{e.g. Dillenia indica, p. 159), often survives in the same locality.
A few cases like Acrotrema dissectum (Trimen's Flora ofCeijlon,\
p. 9), where intermediate forms {possibly hybrids) occur, have
been noticed, but more usually the local species is fairly well
distmguished from the wide-ranging form. And in some instances
transitions are impossible, as, for example, with Coleus elongatus
(detailed characters given on p. 152). I may quote here what
has already been said about it in a paper of 1907 (118):

''The species is too entirely different from the other species
ot Coleus, whether we take C. barbatus or one of the others, for
evolution by means of continuous variations to have been
possible. To take some of the characters, especially those that
are most prominent, how is the one type of innorescence goin^
to develop into the other by any possible continuous variation?
Ihe mind cannot conceive of such a process, unless it be by dis-
continuous variation. Still more, how is a calvx with one big
tooth on top and four small ones below going to develop into
one with five equal teeth? The study of infinitesimal variation
shows that the maximum change to be expected in one generation
would be a mere fraction of the width of a tooth, and how is this
to prove of sufficient advantage or disadvantage to be of any
material import in the struggle for existence? The question is
equally hard if we suppose a common ancestor, for what kind of
calyx or inflorescence will be intermediate?"
And cf. above, p. 209, as to changes in caUx teeth.

In this case the species that one must regard as ancestral,
C. barbatus, is also found in the same locality; it is as frequent
on the summit of Ritigala as C. elongatus, and grows in similar
spots on the exposed rocks. Both suit the same conditions, and
if they have descended from a common ancestor, not one from
the other, it is very remarkable that one should be confined to
Ritigala, one common to tropical Asia and Africa.

Nearly seventy years ago, Lyell (69, p. 39-^) said "Might not
the births of new species, like the deaths of old ones, be sudden? "
and it appears to me, that when one puts together the facts of
distribution as understood in the light of Age and Area, and the
still more surprising fact of the agreement of the type of dis-
]3ersal of species, both by area and into genera, and of genera by
area and into families, etc., as more fully described above, one
can hardly arrive at any other conclusion. Ad\-antage as a cause
in evolution seems to be ruled out with practical completeness,
though it will determine whether the newly evolved form will
survive or not; and if advantage cannot be adduced, then one
can hardly conceive of the changes that distinguish one species

220 THE ORIGIN OF SPECIES [pt. ii

from another having taken place gradually, whether by in-
finitesimal stages, or by small mutations, unless there be some
at present inscrutable law that determines that such shall be
the case. It will be much safer for the present, at any rate, to
leave out of account such a su])position, and to work upon the
idea that the whole distii\ction of a species may appear at once.

Now the ncAv and distinct forms that ha\-e come into existence
range from the minute varieties of the Drabas and Hieraciums
of northern Europe to differences of well-marked Linnean-
specific rank, and one must therefore suppose that mutations
giving rise to such forms may be of similar variation in size.

It must not be supposed that this is being laid down as an
absolute rule, but it would seem probable that it is a very
general one. Individual forms may owe their origin to many
causes, but in most cases it would seem to have been due (im-
mediately) to a mutation small or large, which differentiated the
new form from its predecessor, but there seems no reason to
suppose that the new form is necessarily better adapted than its
predecessor, and Avill kill it out in competition. The widety dis-
tributed, and presumably parental, Ranunculi of New Zealand
are just as common in the south of South Island, where there is
such a mass of endemics (fig. on p. 156).

Natural selection comes in, not as a causative and positive
agent, but as a destructive and negative one. The new form will
instantly have a most strenuous struggle for existence, so that,
if not perfectly suited to the conditions that obtain upon the
spot where it is born, and at the moment of its birth, it will be
remorselessly killed out. If it passes successfulh^ through this
competition, it may be regarded as eminently suited to that spot
and those conditions, and may then spread as long as it can find
suitable conditions into which to travel. Not infrequently it will
meet with conditions that suit it even more perfectly than those
to which it was born, and we shall be liable to imagine it specially
adapted to them, when really it is only they that are suited to
it. Actual experience of the great changes in climatic conditions
that go on from year to year shows that most species are really
suited to a somcAvhat wide range of conditions. This being so,
there is little reason Avhy the child should suppress the parent in
competition. The latter will have proA^ed its suitability to the
conditions, and will probably have a much wider range, and the
chance of a direct and severe struggle between the two is but
small. Even if the child should suj3press the parent in portions
of its range, it will not be likely to overtake it o\'er the whole,

CH. XX] THE ORIGIN OF SPECIES 221

and the parent will probably survive in the outer portions of its
range at any rate.

It is clear, if the Age and Area explanation of the facts
of distribution be accepted— and as yet no other satisfactory
hypothesis is forthcoming— that the endemics must in general
be younger than the "wides," and it seems natural to suppose
that they have been derived from the latter. But if this be so,
then both parent and child occur together, or near together, in
most cases, and if one push this consideration to its logical con-
clusion, one will see that there is no reason why the whole tree
of the evolution of a genus (or even family) should not survive
upon the earth at the present moment, as I have contended for
the last fifteen years (120). Destruction such as that wrought
by the glacial periods, or other geological convulsions, might of
course kill out genera or families, but so long as conditions
remain reasonably constant, there seems no reason why they,
or intermediates, should be killed out.

If, as seems probable, destruction in the struggle for existence
is to fall largely out of consideration as potent in the evolution
that has gone on (except that it must have destroyed tens of
thousands of incipient species, many of which might have been
of great value had man been there to preserve and investigate
them), we cannot regard Jordanian species as stages in the evo-
lution of Linnean, for to get the localised Linnean from Jordanian
species, Avholesale destruction must have gone on, killing out
altogether many of the latter.

\Vhilst the exclusion of advantage to the species as a serious
factor in its evolution (though of great importance in deter-
mining whether or not it shall survive) practically compels us
to accept the theory of mutation, and that such as may give rise
at once to Linnean species, it also seems to me, when taken in
conjunction with other phenomena which are now clearly visible,
to involve other changes in our views. Chiefly important among
these is the new view of evolution, first proposed by Guppy in
1906, and by the writer in the following year, that evolution did
not proceed from indi^^idual to variety, from variety to species,
from species to genus, and from genus to family, but inversely,
the great families and genera appearing at a very early period,
and subsequently breaking up into other genera and species. The
final results of the study of Age and Area, with its demonstration
of the universality of the hollow curve, seem to me at present
almost to involve the acceptance of this view, and the subject
will be fully developed in a subsequent book.

CHAPTER XXI

AGE AND AREA AND THE MUTATION THEORY
By Hugo de Vries, F.M.R.S.

The main principle of the mutation theory is that species and
varieties have originated by mutation, but are, at present, not
known to have originated in any other way. Originally this con-
ception has been derived from the hypothesis of unit-characters
as deduced from Darwin's Pangenesis, which led to the expecta-
tion of two different kinds of variability, one slow and one
sudden.

Freed from the assumption of a transportation of gemmules
through the organism, the conception of Pangenesis is the clear
basis of the present manifold theories of heredity. An organic
being is a microcosm, says Darwin, a little universe, formed of
a host of self-propagating organisms, inconcei^'ably minute, and
numerous as the stars of heaven. In honour of Darwin, I have
proposed to call these minute organisms pangenes, and this name
has now been generally accepted under the shortened form of
genes. They are assumed to be the material bearers of the unit-
characters of species and varieties.

This principle leads almost directly to the distinction of two
different kinds of variation. For the first, no material change of
the genes is required; they remain what they are. No two leaves
on a tree are exactly alike; no tAvo individuals of a species are
the same in exery detail. These two well-knoAvn propositions are
the essence of what we now call fluctuating variability. In their
visible features characters usually oscillate around a mean value,
but this does not affect their material bearers. The researches of
Quetelet and Galton have shown that such oscillations follow
the law of chance. Starting from this idea, fluctuating variability
of animals and plants has now become a main branch of bio-
logical study.

Besides these, changes may be expected, which involve the
material bearers of heredity, or the genes, themselves. Some
may be lost, either really or apparently, and new ones may be
added to the stock, this latter process consisting probably in the
transformation of old genes into new types. In consequence of
such changes the external features of an organism may become

PT. II, CH. XXI] THE MUTATION THEORY 223

altered, and these alterations are now generally called mutations.
The theory assumes that these only are connected with the origin
of species and varieties.

Darwin recognised both mutation and fluctuation as steps in
the general process of evolution. For this assertion he mainly
relied on his studies of the variation of animals and plants under
domestication, since organisms in the wild condition did not, at
his time, afford a sufficient basis for controlling his conception.
He assumed mutations to be of subordinate significance, explain-
ing the main lines of the evolutionary process on the assumption
of individual or gradual variation. This variation he had shown
to occur everywhere, but as to its capability of achieving lasting
changes, he had no facts at hand to give a definite proof.

In my book on the Mutation Theory I have given an elaborate
"Review of the Facts," especially on the botanical side, in order
to show that fluctuating variability does not lead to durable
changes in the hereditary composition of a type. Wherever such
changes occur they may be shown to be historically, or at least
probably, due to saltations. These critical considerations led to
the proof that the conception of mutations was in full harmony
with our knowledge of the variability of plants, as it occurs
everywhere in nature as well as in horticultural and agricultural
breeding.

The mutation theory is intended to be a support and a corol-
lary of the selection theory of Darwin. There can be no doubt
that Darwin correctly set forth the essential steps in the evo-
lutionary process and that changes in his views mostly relate
to those minor points, for which, at his time, the material of
facts was not adequate to a correct decision. The mutation theory
claims to remove many of the difficulties, inherent to the Dar-
winian doctrine, as e.g. the general occurrence of useless charac-
ters and the impossibility of explaining the first beginning of a
selection on the ground of its usefulness.

In order to become generally accepted this theory has to be
considered from two main points of \'iew. The contention that
species and varieties originate by mutation is essentially experi-
mental in its nature. But the thesis that they cannot be shown
to have ever originated in another way has to be studied in the
field of systematic botany and zoology, and partly in that of
palaeontology. Mutations were well known to Darwin to occur
from time to time, and of late numerous observations of special
cases in animals and plants have been published. A list of them

224 AGE AND AREA [pt. ii

has been prepared by Gates in his new book on Mutations and
Evolution.

In the fruit-fl}^ Drosophila over two hundred instances have
been studied by Morgan and his co-workers, and the evening
primroses, or Oenotheras, have afforded some dozens, many of
which differ more Avidely from their parent form than recog-
nised wild species of this polymorphic genus do among them-
selves. On the other hand, no observations have been adduced
of new forms originating experimentally from fluctuating vari-
ations.

The experimental work has not, however, chosen for its scope
the proof of the reality of mutations, but has preferred other
lines of research. In the studies of Morgan the distribution of
the genes along the chromosomes, as predicted from the prin-
ciple of Pangenesis, has been the main aim. With Oenothera the
prominent question was the search for a method of studying
the internal and external causes, Avhich induce mutations to
occur repeatedly. A thorough knowledge of these causes must,
in the end, enable us to produce artificially distinct changes,
determined beforehand. In other words, it must afford the means
of evolving arbitrarily new useful varieties of chosen qualities,
in agricultural and in horticultural plants.

In systematic studies it is now generally recognised that the
characters used in the diagnostic distinction of related species
are not such as would be expected on the ground of Darwin's
selection theory. As a rule they relate to qualities, which cannot
be explained on the assumjjtion of an origin by the accumulation
of infinitesimal steps on the basis of their usefulness for the
species. They are not observed to increase the chance of success
in the struggle for life. Most forms would thri^•e as well without
their aid. This is especially the case with morphological charac-
ters, whereas adaptation to such environmental conditions as
moisture or drjaiess, shadow or open field, physical and chemical
constitution of the soil, etc., might far more easily be imagined
to evolve slowly. But even here direct jjroofs are wanting.

It is a curious fact that most of the striking instances of
beautiful adaptation to special forms of life are characters of
genera and subgenera, or even of whole families, but not of
single species. Climbing plants and tendrils, insectivorous plants,
desert types of Cactus, Euphorbia, and so many others, sub-
merged water plants, and numerous other instances could be
adduced. Since we do not know when and where and under

CH. XXI] AND THE MUTATION THEORY 225

which external conditions those types have originated, all specu-
lations concerning their evolution on the ground of their uses
must be considered to be more of a poetical than of a really
scientific nature. Wherever striking adaptations to the environ-
ment are met with, we will always have to grant that they did
not originate under the conditions of the locality, where we
observe them, but elsewhere and in long forgotten times, the
environmental conditions of which are necessarily imknown to us.
Hitherto systematic enquiry Avas obviously handicapped by
the weight of such objections, and they were simply left out of
consideration. No principle was known, which would enable us
to decide the question, Avhether advantageousness to their
bearers had played any role in the evolution of new characters.
Later on, after many wanderings of a species into different new
environments, a character might prove to be useful in some of
the new localities, and here induce a rapid multiplication.
Striking adaptations, such as those of desert plants, may be
the consequence. But whether the characters have evolved
under analogous or under quite different conditions, we do not
know.

It is at this point that the theory of Age and Area has come
into the discussion. It showed that the dispersal of species,
especially in the first period after their birth, is independent of
their distinctive morphological characters. This phenomenon
may be studied on a purely statistical basis without the aid of
personal appreciations of biological qualities.

In the first place, the discovery tJiat endemic species are, as
a rule, the youngest in their country, has provided us with a
means of judging the value of their characters in the struggle
for life. But even here such a relation is not observed. The en-
demic forms of Coleus of Ceylon, and numerous other instances,
show their marks to be minute and of subordinate importance,
although they are recognised by the best systematists as having
full specific value. Many endemic species are still living in the
same locahty and obviously under at least almost the same con-
ditions as those under Mhich they have originated. But no
relation of their new marks to any use in the struggle for exist-
ence can be pointed out. They have inherited their adaptation to
the environment from their ancestors, but are rarely known to
have increased it. Only in some cases they have succeeded in
spreading rapidly and widely, and then, of course, an improxe-
ment in adaptation may be granted. But even here there is

226 AGE AND AREA [pt. ii

nothing to show that the e^okition of the character was due to
this cause.

The conchision obviously is, that specific characters have
evolved without any relation to their possible significance in
the struggle for life. The facts are contrary to the main principle
of the selection theory of Darwin. Moreover, intermediate steps
betAveen the endemic species and their parents, in the midst of
which they are ordinarily still living, are wanting, and therefore
must be assumed never to have existed. Endemic species must
have appeared at once, by means of one or a few distinct steps,
which embrace their whole differentiation from the parent type.
Considered in this way, it is evident that their origin is in full
accord with the principles of the mutation theory, and has to
be considered as one of the best proofs of its applicability to
evolution in general.

Starting from the endemic species, Willis has worked out his
statistical methods of studying the relation of age to dispersal
for larger and larger groups. Everywhere this relation is shown
to be, in the main, independent of the specific characters. It
obeys the same laws in widely different genera and famiUes.
Dispersal is not due to special adaptation, and often, as in the
Podostemonaceae, the most beautifully adapted forms are the
local ones, whereas the universally spread species of the same
group show the smallest degree of specialisation.

In other words, the area occupied in a country by any gi\en
species depends upon the age of that species in that country,
and not upon special characters. Of course this law applies to
the common type of species, and exceptions may be expected to
occur. For this reason the species are not studied singly, but in
small groups of twenty or so, and on this basis the law has been
found to be everywhere the same in the animal and in the vege-
table kingdom.

Leaving the appreciation of the importance of this principle
for pure systematic studies and for the construction of family
pedigrees to other judges, I might here point out its bearing on
the mutation theory. It affords a full proof that cA-ervAvhere in
nature, in geological periods as well as at present, the morpho-
logical characters of newly originated types have no special
significance in the struggle for life. They are not known to aid
them in their initial dispersal. The}' may afterwards prove to be
useful or useless, but this has no influence upon their evolution.
Obvious instances of usefulness occur, as a rule, only at much

CH. XXI] AND THE MUTATION THEORY 227

later periods during the wandering of the new forms, when un-
expectedly they arrive in environments specially fitted for them.

The usual phrase, that species are adapted to their environ-
ment, should therefore be read inversely, stating that most
species are now found to live under conditions fit for them. The
adaptation is not on the side of the species, but on that of the
environment. In a popular way we could say that in the long
run species choose their best environment. Favourable local
conditions induce a rapid multiplication, whereas elsewhere the
forms remain rare, or are seen to disappear slowly.

The general belief in adaptation as one of the chief causes of
the evolution of specific characters is thus directly contradicted
by the statistical studies of Willis, which are independent of all
personal appreciation or estimation of a supposed value. This
result must be considered as the one great proof, which the
mutation theory still wanted for its acceptance in the field of
systematic zoology and botany.

15—2

CHAPTER XXII

GEOGRAPHICAL DISTRIBUTION: GENERAL

Our general outlook upon biological problems has been, and to
a great extent still is, principally governed by the theory of
natural selection — the mechanism by whose invention, and by
virtue of whose a priori reasonableness, Darwin Avas able to
render the immortal service of establishing the theory of evo-
lution. Few people nowadays would be found to give a complete
assent to the doctrine of natural selection, but though the pre-
mises are therefore weakened or destroyed, the conclusions
draw^n from them are still accepted with little or no question.
Somewhat to my surprise I have found many who no longer
accept natural selection as operative in evolution in a positive
(rather than negative) manner, but who are prepared to fight to
the death for conclusions that are essentially based upon it, such
as that species of small area are usually relics.

When one comes to look at the history of the subject of geo-
graphical distribution, one soon reahses that since the impulse
which was first given to it by the acceptance of the theory of
natural selection has spent its force, little work of any import-
ance^ dealing with the broad general distribution of plants about
the world (as distinguished from their local distribution into
societies and associations occupying various types of habitat)
has been carried on. The limiting factor in progress at the present
time is the lack of a proper theoretical background from which
fruitful hypotheses may be derived. The facts of distribution
remain an insoluble problem so long as one cndeaA'ours to explain
them by the theory of natural selection, and the more that the
attempt is made, the greater is found to be the incompatibility
between theory and practice. The serious study of geographical
distribution has consequently been more and more neglected,
A\'hilst at the same time it has been admitted in a vague theoretical
way that no theory of evolution can stand wh'ch will not explain
the facts of dispersal.

Chief among the deductions — consciously or unconsciously

1 The last important work was probably that of Gi.ppy (44, 46, 47), and
it is to be noted that this work has led him to conclusions (expressed in his
Theory of Differentiation) diametrically opposed to the theory of Darwin.

PT. II, CH. XXII] GEOGRAPHICAL DISTRIBUTION 229

made— from the theory of natural selection, which are to-day
strenuously supported, and the belief in which seems to me the
chief preventive to further progress in the study of distribution,
are perhaps the following:

(1) That distribution of species about the world has in general
been rapid.

(2) That the present distribution of species and genera about
the world represents the maximum possible to those species and
genera, and that distribution is consequently a closed chapter.

(3) That species and genera now existing occupy, as a rule,
just those places to which they are suited.

(4) That species and genera occupying small areas are as a
general rule species and genera that are dying out (relics).
Natural selection could not produce them upon areas so small
as are occupied by a great many. It also demands that there
shall be a good many moribund forms; and therefore these
localised forms are assumed to be dying out.

(5) That on the whole, in the same way, small genera (with
few species) are to be regarded as relics, and as in process of
dying out.

As regards the first two of these, we have seen in Chapters ii
to V that there is no reason to suppose that as a general rule
dispersal in nature is anything but extraordinarily slow, the
ground being usually fully occupied by societies or associations
of plants, into Avhich entry will be difficult or even impossible.
This is confirmed by ordinary oljservation, for if one remember
the position of various clumps of plants from one's childliood.
one soon realises that if man have made no alterations in the
neighbourhood they will be found in the same places, without
having extended their area except in very rare instances. Dis-
persal may be rapid if there be (which is very rarely the case)
virgin soil available, or if man or other cause have made some
great alteration in conditions, but usually it will be a matter of
the most extreme slowness. The figures for areal distribution
that have been given above, showing that the "hollow curve"
is apparently' a universal rule, not only for totals, but for indi-
vidual families and genera, show clearly that dispersal follows a
largely "mechanical" course, and that if a species now occupy
a small area, it is in most cases because it has not had time to
occupy a larger one. If the areas occuj^ied had been determined
by natural selection, it is inconceivable that they should have
been thus graduated in sizes from many small to few large, with

230 GEOGRAPHICAL DISTRIBUTION [pt. ii

no breaks in the continuity of the figures, and that not only on
the totals, but in individual families and genera. We have also
seen that there is no need for rapid dispersal, when the time
available is considered (cf. p. 33).

It is thus fairly clear that the existing distribution of species
and genera, in probably the great majority of cases, represents
only the dispersal possible in the time that has elapsed since their
evolution. If one could return to the world after ten thousand
years, one might find an appreciable extension of their area l^y
existing species, but to expect it in a short time is more sanguine
than reasonable.

The fact that the composition and distribution of the floras of
the outlying islands of New Zealand can to a large extent be
predicted from a knowledge of the distribution in New Zealand
of the New Zealand flora (pp. 66-75) is a very strong argument
indeed in favour of the view that dispersal depends chiefly upon
age, i.e. that it is determined by various factors which when one
deals with long periods are found to act at a more or less uniform
speed, and that consequently the existing dispersal of species
does not represent the end of the chapter, but only the point
which has so far been reached.

If one accept the two suppositions under discussion, it is quite
impossible to explain numerous facts in distribution which are
easily explained by aid of Age and Area, for instance, the fact
that the Auckland Islands have -i5 per cent, of their flora mono-
cotyledonous, the Chathams 31 per cent., and the Kermadecs
only 21 per cent.; or that the plants of the floras of these out-
l}'ing islands (p. 67) are unusually widespread in New Zealand,
and those of the Chathams much more so than those of the
Aucklands and the Kermadecs. It is impossible with these sup-
positions to do any prediction about distribution at ah, whereas
nearly a hundred predictions have already been successfully
made with the assistance of Age and Area, and have added con-
siderably to our knowledge of the distribution of plants in the
New Zealand area.

In regard to the third hypothesis (p. 229), the supposition that
species and genera occupy just those places to which they are
suited has usually been taken for granted, and a vast amount of
energy has been devoted to the problem of finding out why they
are suited. But, as has just been pointed out, we can no longer
safely draw this conclusion. If a species is not suited to its loca-
tion, it will probably die out, as is apparently happening with

CH. XXII] GEOGRAPHICAL DISTRIBUTION 231

Cupressus macrocarpa at Monterey (p. 88), though this species
IS admirably suited to life in a climate a little damper. But it is
stretchmg our imaginations somewhat to imagine that most
localised species are suited only to the places in which they occur
Conditions change so much from year to year that unless a
species IS suited to a considerable range, it will not be able to
survive at all. It would not obtain a greater change by moving
to another locality not too far away. It is probable that the
slow acclimatisation practised by nature will ultimately ac-
custom species to widely different conditions, but loner" time
must be allowed. "^

The arithmetical facts disclosed in this book are much opposed
to any such supposition. It is almost impossible to suggest con-
ditions to which the overlapping species in the map on^^p. 56, or
the grouped species of varying size of area on p. 156, can be
suited. The point of view usually taken up on this matter has
been very well put by Huxley (59, p. 123), who says:

"We are very much in the habit of tacitly assuming that
because certain plants and certain animals exist only under cer-
tain chmatal conditions, there is something in what' we vaguely
call the 'constitution' of the plant or animal which binds them
to these conditions, and renders it impossible for them to live
elsewhere. I wish we could get rid of this word 'constitution';
for I take it to be one of the many verbal anodynes by which the
discomfort of ignorance is dulled."

The arrangement of species in areas that are concentrated
about particular points, as is shown in the curves and maps on
pp. 79, 80, 153, 156. goes to show that local adaptation has had
little to do with the dispersal. If not locally adapted, the species
Mould die out Avithout spreading at all; but once established
they begin to spread, at an average rate determined by the
various factors that act upon them. The fact that the northern
invasion of New Zealand (cf. table on p. 77, and curves on pp. 79,
80) does not show any increase of local species at the region
where the southern invasion shows its maximum, and vice versa,
is a strong proof against local conditions having anything serious
to do with multiplication of species.

The fourth and fifth suppositions, that species of small area,
and genera of one or few species, are dying out, are those most
strenuously adhered to, but in view of the facts set forth in tliis
book seem to form a very difficult position to uphold. It need
not be entirely abandoned, but in place of suj^posing fnost such

232 GEOGRAPHICAL DISTRIBUTION [pt. ii

species and genera to come under this head, one must be satisfied
with a small number; to the great bulk the contention is not
applicable. We have seen, and seen it so strikingly in numerous
instances that there can be no doubt that it is a general rule,
that the species in a gi^en countr}^ endemic or not, are grouped
there (according to the areas that they occupy) in a perfectly
definite manner, Avhich is always the same. The wides are found
(when there are also endemics) with many in the class of largest
area, and numbers decreasing downwards, the endemics arranged
in the reverse direction. This regular arrangement is completely
opposed to the idea of relic nature, for how could there be many
at the last stage of relicdom, fewer at the last but one, still fewer
at the last but two, and so on? It is equally opposed to the idea
of local adaptation, it may be Avorth while to point out, for why
should there be many adapted to the smallest areas, with num-
bers steadily decreasing upwards. Still more difficult is it to
explain, upon either of these suppositions, why the wides (if
endemics occur also) should be arranged in the reverse direction^.
If there be special local adaptation, then the wides must be
much better suited to the country than the locally evolved
forms !

Inasmuch as all families and genera, of reasonable size, agree
in arrangement, some mechanical explanation is needed to ac-
count for the mechanical regularity, and the only reasonable one
suggested is age (for youth cf. pp. 89, 92). Age in itself, as already
explained, does nothing, but it allows time for the active factors
in distribution to produce their effect. To accept age as a mechani-
cal explanation simply means that we regard these factors as
producing a resultant or total effect which goes on at an average
speed, so that age becomes a measure of dispersal. The dispersal
is of course stopped sooner or later by barriers, physical or eco-
logical, including the barrier imposed by the fact that a species
has reached the extreme of temperature, dryness, etc., that it
can -vAithstand. The real difference between the old view of dis-
persal and that given by Age and Area is that imder the latter
we regard almost all species as in process of extending their areas
of dispersal, not some as extending their areas and as many or
more contracting theirs (cf. footnote on p. 174). The exceptions
to this — the real relics — are comparati\'ely few and far between,

'^ When, as in Britain, there are no endemics, the wides diminish upwards,
but show considerable mmibers in the most widely dispersed classes, owing
to accumulation there of species that could not rise higher.

CH. xxir] GEOGRAPHICAL DISTRIBUTION 233

forming perhaps 1-2 per cent, of the total of species of very
restricted area.

Very many arguments against the old position have been
brought up above, e.g. on pp. 58, 81, 88-94, 141, 164-6, and 179.
No one has yet attempted to reply to any of these, which have
mostly been already published, but the position is obstinately
held, and the facts brought out by the study of Tertiary floras
are especially appealed to. These show that there are without
doubt, in the north temperate zone, a number of forms, perhaps
even as many as 600 to 1000, rather widely separated from their
nearest allies (when they have any such), and probably Tertiary
relics ; but it is not properly realised that these are a mere trifle
when compared to the local species that occur south of the tropic
of Cancer. Brazil alone has about 12,000 endemic species, usually
well localised; even the little island of Ceylon has nearly 250
species of the most localised distribution possible, almost half of
them occiu-ring each on one mountain top only, and it has nearly
800 whose area does not exceed 4000 square miles (63 x 63 m,).

In view of the facts that have been brought up abo\'e, showing
the Avay in Avhich not onh' the areas occupied by endemics, but
those occupied by other species, are arranged in hollow curves,
and showing that this same type of arrangement also occurs in
the grouping of genera and families into sizes, the idea of relic
nature, or of special local adaptation (except in so far as this is
needful for all species, if they are to survive), must, it seems to
me, be abandoned for the great majority of cases, and the
mechanical explanation adopted in its stead, that area occupied
goes with age. Nearly all forms are to be looked upon as in-
creasing their area, and only a few, not most, as moribund.

That this view is in all probability the right one to take of
the phenomena of dispersal is shown very clearly by the way in
which, accepting it, predictions as to distribution may be made,
and have as yet been uniformly successful (in almost a hundred
instances). Very strong evidence, and evidence based upon
definite facts, not upon a priori reasoning, is now required to
show that the hypothesis of Age and Area is unsound.

But not only have we seen reason to accept Age and Area, but
also to accept the similarly "mechanical" hypothesis of Size and
Space (Chapter xii, p. 113), which asserts that Avhen one deals
with groups of allied genera the size of a genus depends largely
upon the area that it covers, i.e. ultimately upon its age. This
follows almost of itself when one has once accepted Age and

234 GEOGRAPHICAL DISTRIBUTION [pt. ii

Area with its implication that all, or nearly all, species are in
process of enlarging their area of dispersal, not some enlarging
and some contracting it. There is no need to quote the evidence
a second time (cf. Chapter xii, and pp. 132, 16-i, 165, 174, 178,
188, 190, 197).

But if these new views be accepted, it is clear that a good
many changes must take place in our mode of viewing the
problems of distribution, which it must not be forgotten ha\e
hitherto been regarded as insoluble. One of the chief among
these is the problem of Invasions of plants from other countries.
If it be supposed that the dispersal of a species depends simply
upon its age (representing the average effect of the active factors)
and the barriers that it meets, and that when once it is estab-
lished in any place it will rarely die out there except as the result
of rather sudden or violent changes of conditions^, and further
that only when these changes attack it at the margin of its area
will they cause any diminution of total area "occupied," then
it is clear that the problem of invasions can be studied with some
hope of obtaining results. This has been illustrated in Chapter
VIII, which deals with the invasions of New Zealand. It was there
shown that by taking the places at which the maxima of species,
endemic and wide, occur, one may get a cl\ie to the different
invasions that have reached the country, and the directions from
which they came. But in a coimtry Avithout any endemics at all.
the same princi])lcs may be applied to its " wides." This has lately
been done for Britain by Mr J. R. Matthew, whose work (74)
gives great promise for the future (and cf. p. 114). Careful account
must be taken of the conclusions of geology, but if we get rid
of the ideas that (proportionately) mamj species are necessarily
dying out, and that most have reached their possible limits of
dispersal, we can study invasion and spread with some hope of
arriving at definite results, a proceeding wliich has been im-
possible under the older views of these matters.

If genera give rise to others in a casual way, and at more or
less casual spots (as the way in which the endemic genera in any
country occur at scattered points would seem to indicate), then
it is clear that in any part of the world one must expect to find
a casual mixture of genera of different sizes, made up in much
the same way as is the entire flora of the world, or one of the

1 E.g. the oncoming of excessive cold, heat, dryness, dampness ; clearance,
fire, submergence, etc.

CH. xxii] GEOGRAPHICAL DISTRIBUTION 235

families of which it is composed (fig. on p. 187). A very
httle examination of local floras suffices to show that this fs
indeed the case.

If, for example, one take the fiora of Britain (37), one finds
that the families, by numbers of genera, are arranged in regular
order, diminishing as the number of genera increases, thusTss/l
(33 of 1 genus), 17/2, 9/3, 6/4, 3/5, 2/6, 2/7, 2/8, 1/9, 2/10, and
so on in scattered numbers to 46. The genera by numbers of
their species in Britain are 223/1, 90/2, 35/3, 32/4, 16/5, 15/6,
and so on. Until the numbers become small there is no break in
the regularity. The first two or three numbers contain the great
bulk of the total; 50 families out of 92 contain one or two genera,
and 313 genera out of 512 contain one or two species. This will
be found upon examination to be a general rule for all floras. In
New Zealand, for example, one finds the genera (total 329) to
be 155/1, 54/2, 29/3, 17/4, 12/5, 11/6, 11/7, 5/8. 5/9, 4/10, and so
on. In Ceylon (total 1027) one finds 573/1, 176/2, 85/3, 49/4,
36/5, 20/6, 19/7, and so on. In Vol. i (only) of the Flora of
British India one finds 173/1, 70/2, 33/3, 19/5, 7/10, and so on.
All form markedly hollow curves, with the great bulk of the
genera in the first tAvo figures, so that there is a very steep drop
until the third or fourth figure is reached, and then a gradual
tapering away to the larger genera. The larger the country, on
the whole, the larger the size of the biggest genera.

One may push this type of distribution, shown in the hollow
curve, into yet more detail, and find that not only the whole
local flora of a country, say, for example, Britain, shoAvs this
curve, but also portions of that flora. The same curve is shoAvn
by the Monocotyledons and Dicotyledons of the British flora,
and even by the individual families, Avhcn of reasonable size, the
grasses for instance shoAving 24/1, 13/2, 1/3, 4/4, 1/5, 2/6, and 8,
11, 13. The line is Avavy, but the numbers are small, and there is
no doubt about the shape of the curve.

Or one may take portions of the country inhabited by more
or less definite associations, or groups of associations, of plants,
and find the same thing. Thus if avc take Cambridgeshire, the
Wisbech division of the county (fen), and the very local Wickcn
fen, from Babington's Flora, of Cambridgeshire, Ave get the same
type of curves (cf. curve 6 in fig. on p. 237). One might
expect certain genera to prove unusually suitable, and to be
disproportionately represented, but this does not seem to be the
case.

236 GEOGRAPHICAL DISTRIBUTION [pt. ir

In the Mixed curves on p, 237, the fourth curve shows the Avhole
flora of Ceylon arranged in order, beginning with 573 genera of
one species each, and forming the usual hollow curve; the 6th
curve shows the flora of Cambridgeshire (Babington), the 9th
the flora of Italy. All the floras so far examined give similar
results, and the same is the case in local faunas, as the 10th
curve (Birds of British India) and the 15th (British Echinoderms)
illustrate. The curve is exactly like the curve given by other
combinations of animals or plants, as maj^ be seen by comparing
them with the other curves in the same figure, e.g. for the Com-
positae or the Chrysomelid beetles, the endemics of islands, or
those of Brazil. The tails in the figure are of course cut short:
their length depends in general upon the size of the flora; the
larger it is, the larger size, as a rule, do its genera reach to.

Or if one take the flora of Ireland, one finds it to be, except
for a few Iberian plants in the south-west, a reduced copy of
that of Britain, and the Avay in which age alone has been the
chief determinant of what species shall occur there is very
strikingly shown by the following figures, extracted from ]\Ioore's
Cyhele Hibernica.

The ]3lants of Britain in the Cyhele Britannica are grouped in
hundreds according to degree of frequency in Britain {i.e. the
number of Watson's vice-counties in which they occur). Of the
first hundred all occur in Ireland, of the second and third hun-
dreds all, of the fourth 98, fifth 97, sixth 93, seventh 84, eighth
74, ninth 63, tenth 66 (the only exception), eleventh 43, twelfth
26, thirteenth 16, and fourteenth 8, a steady diminution from
top to bottom.

But if size also depends upon age, then it is clear that in any
local flora the genera, Avhich as a rule will not be endemic, should
be arranged in the same way. The genera arriving for example
in Britain will not all arrive simifltaneously, but some will arrive
sooner than others, and these will tend to be the larger genera of
the nearest source of supply, for the larger genera will usually be
the more widespread. The ultimate result will tend to be that
these genera will not only arrive first, but will tend to be repre-
sented by more species, so that one will expect the most Avidely
distributed species in the large genera {i.e. large for the country
in question, being represented there by many species) to be
more widely dispersed than those of the small. This we have
already seen to be the case in the most striking way (p. 114).
But one may push this arithmetical regularity further yet. If

CH. XXII] GEOGRAPHICAL DISTRIBUTION 237

one take the number of species per family in the British flora,
one finds it to increase steadily with the number of genera; there
are no breaks, as one would be inclined to expect. The families

Monospccifi c Genera at th 13 end 0 f curve

Number of species (or size of area]

Mixed curves, to show the close agreement of the hollow curves, whether
derived from families of plants grouped by sizes of genera (Compositae,
Hymenomycetineae, Simarubaceae), families of animals (Chrysomelidae,
Amphipodous Crustacea, Lizards), endemic genera grouped by sizes
(Islands, Brazil, New Caledonia), local floras grouped by (local) sizes of
genera (Ceylon, Cambridgeshire, Italy), local faunas (Birds of British
India, British Kcliinoderms), Tertiary fossils by sizes of genera, or
Endemic Compositae of the Galapagos by area. [By courtesy of the
Editor of Nature.]

with one genus show an average of 2-2 species per family, those
with two an average of 8-3, with three of 10-7, with four of 12-3,
with five, six or seven genera of 15, with eight, nine, or ten of

238 GEOGRAPHICAL DISTRIBUTION [pt. ii

40, and with more genera than ten of 73. The numbers increase
regularly with the number of genera.

One may even find, here as elsewhere, that (as a general rule)
the small famihes, which, as already explained under Size and
Space in Chapter xii, will tend to be the latest arrivals, have
fewer species per genus. While the families of one genus in
Britain have 2-2 species per genus, those with more than one
genus have a generic average of 3-3 species. If one take New
Zealand, one finds the 34 families of one genus to average 2-8
species per genus, those with more 4-3.

One may even take the families of one genus in a country, and
find that they are arranged in arithmetical order. In Britain
there are 20 of these with one species, 7 with two, and six more
with larger numbers. In New Zealand there are 18/1, 6/2, 3/3,
and seven more. And this rule appears to hold e%'erywhere. If
one take the British families of two genera, one finds 12 genera
with one species, 7 with two, 3 with three, and 12 others. In
Ceylon the bi-generic families show 26/1, 8/2, 3/3, 2/4, 1/5, 1/6
and 9, 19, and 20 species. Everywhere the arrangement of genera
by species follows this simple arithmetical rule, forming hollow
curves. Even the proportions of families and genera of different
sizes in a country show some resemblance. In Britain 35 per cent,
of the families are monogeneric, in New Zealand 37 per cent., in
Ceylon 44 per cent. In larger and less isolated areas the pro-
portions are smaller, and in the world they are only IS per cent.

Another matter upon which it becomes needful to adopt a
somewhat different view-point is the Struggle for Existence.
We have seen that it can no longer be regarded as an important
determining cause in evolution, and that it is most strenuous for
the individuals of new species that are just commencing. If
they cannot succeed in this first struggle, they will simply die
out and leave no trace, but if they do succeed, they may be
looked upon as having passed through the sieve of natural
selection, and being, so to speak, certified as fit for existence in
the region where they arose. Until they have spread to some little
distance, however, they can hardly be looked upon as established,
for they M'ill be very liable to sudden extermination, whether
ideally or badly equipped for life. A fire on the tiny summit of
Nillowe-kanda in Ceylon (p. 55), for example, would probably
exterminate the three species that are confined to it (and cf.
Didymocarpus and Christisonia on p. 151). Once established on

CH.xxii] GEOGRAPHICAL DISTRIBUTION 239

a reasonable area, only individuals, and not the species, will
usually be affected by the struggle for existence. Only very
rarely will a new form overtake its parent over the whole or the
greater part of its range, and destroy it. We are no longer
obliged to regard a new species as coming into existence at the
expense of its ancestors.

Another important general result of the work upon Age and
Area outlined above is to show that in any given countrv, and
therefore in the world in general, the "wides," which occupy
the largest areas (on the average), are the oldest forms, i.e. that
they were the first to appear. The facts set forth showing the
distribution of the various classes are indisputable at the stage
that the work has now reached, and they are wonderfully con-
cordant from one country to another. No one has attempted to
contradict them, but there has been much a priori reasoning to
the effect that this or that has not been allowed for, that it is
obvious that so-and-so must produce great effects, etc. None of
this reasoning, however, has attempted to explain the facts,
which are so striking and so consistent that they must have an
explanation, and that a mechanical one, on account of their
mechanical regularity. The only reasonable one is, as frequently
]Dointed out, that the factors acting upon dispersal produce in
the long run a very uniform effect, so that age forms a measure
of dispersal.

But if this be so, tliere is no possible and reasonable explana-
tion of the endemics, which in general are younger than the
wides, and occur beside or near them, except that they are
descended from the wides, directly or through other endemics.
But when a new endemic arises in this way, unless it is much
better suited to a variety of conditions than its parent, it will
never overtake the latter, and we have seen that there is little
reason to supi^ose a combat a Voutrance between them. The
parent will most often, probably, surA'ive beside the child. At
times it is possible that it may survive only beyond it; but the
distribution, for example, of the Ranunculi of New Zealand,
where the parental wides are just as common in the region where
the crowd of endemics occurs, as in the far north where there are
none (cf. map on p. 156), gives little evidence in favour of this
latter supposition. In my various papers I have assumed that
the wides give rise to the endemics, and have made nearly a
hundred predictions upon this basis. As these predictions have

240 GEOGRAPHICAL DISTRIBUTION [pt. ii

always been successful, the assumption is therefore probably
correct.

The endemics, then, of course with a good many exceptions,
are in general to be regarded as derived from the wides that
occur among them. In many cases, as we have seen, and those
most often cases in which there is reason to suspect greater age
than usual, a genus in any single country may have only endemic
species (cf. pp. 95, 155), sometimes only one, sometimes more,
and in these cases we may suppose some mutation, perhaps at
once, in the first wide to arrive, or perhaps subsequently and
en masse.

But now, if, in general, the appearance of a new form does not
imply, as it did under the theory of natural selection, the dis-
appearance of its ancestral forms, there seems little reason why
both should not survive upon the earth, or, in other words, why
the whole, or great part, of the tree of a family or genus should
not survive. I have already worked out this question in regard
to the Dilleniaceae (120), suggesting that Tetracera, a wide-
spread and very simple genus, may have been the ancestral
form from which the family was derived. In the same waj' the
Polemoniaceae (p. 171) might have been derived from Pole-
monium, the INIenispermaceae from Cocculus and Cissampelos,
whilst in Cissampelos itself, C Pareira (p. 159) might have been
the parent of the other species, directlj- or indirectly. In Doona
(p. 152), D. zeylanica may in the same way be looked upon as the
probable parent, direct or indirect, of the other species, and

It is clear that -when once the general principle of Age and
Area is established — and already the e^-idence in its favour is
\exy strong — it may be called into service in the study of
phylogeny. But if it be accepted, it is clear that Guppy's
Theory of Dijferentiation (p. 221) must almost necessarily be
accepted also. This subject will be dealt with in a later book
upon Evolution generally, and can only be mentioned here.

Just as the endemics of small area are to be looked upon as
descended from species of larger area, so also we have seen that
the monotypic genera are to be looked upon in general as
descended from larger genera. The Avay in which the nimibers of
genera, not only in the total, but family by family, are arranged
(cf. p. 187) in hollow cur\-es, with a great preponderance of mono-
types and steady decrease to a few of large numbers, shows that

GEOGRAPHICAL DISTRIBUTION

241

there is a definite mechanical relationship between them. If we
imagine existing genera to give rise to new genera, as they give
rise to new species, by mutations at intervals, we shall then
expect that genera as a whole will follow the law of compound
interest. But if this be the case, then it follows that whilst the
number of genera plotted to the numbers of s{>ecies that they
contain vnW give a hollow curve like those on p. 237, the loga-
rithm of the number of genera plotted to the logarithm of the
numbers of species that they contain will give a straight line^.

Number of species
I 5 10 20 30 40 50 100 200

10 12 1-4

log (N9 of species)

Logarithm curve for all Flowering Plants (from Willis, Diciionary).
(By courtesy of the Editor of Nature.)

That this is in fact very close to the actual truth when con-
siderable numbers are dealt with is shown by the figures on
pp. 241, 242, which give the logarithmic curves for all flowering
plants, for the Rubiaceae, and for the Chrysomelid beetles. This
subject must also be left for further consideration in a later book.
Suffice it to say for the present that the evidence is decidedly in
favour of the origin of new species and genera from old by
mutation, which in the long run has followed a very definite

1 For this deduction I am indebted to my friend Mr G. Udny Yul
C.B.E., F.R.S.

W.A. 16

242

GEOGRAPHICAL DISTRIBUTION

,

N2 of

1

species

0

30

100

0 ^

^..,

"^

^x:^.

^6^ ^

''■^.N

\

>

^^^\

\

s...

2 4 5 8 10 1-2 14 16

log (N9of species)
Logarithm curve for Rubiaceae (from Willis, Dictionary).
(By courtesy of the Editor of Nature.)

\

1

\

^^ i

^ C

%

^^^"

^'Vq

"'^

^

\

^

^

\

D

' -6
5

•fi

log(

{■

Number 0

ic

: 1
f species)

2 ' 1-
20

* ' 1
30

5 it

2
100

Number of spscies

Logarithm curve for Chrysomelid beetles (from old Catalogue).
(By courtesy of the Editor of Nature.)

CH. xxii] GEOGRAPHICAL DISTRIBUTION 243

plan, new mutations being cast on the average at a fairly definite
speed, differing of course for different classes of animals and
plants.

The acceptance of the view that B is the direct descendant of
A, another living species, instead of both being the descendants
of some hypothetical a (an ancestor which by the way has never
been found in the fossil state, so far as I know, though on the
current theory there should be hundreds of thousands of them)
will make the work of tracing phylogenies easier, though if
mutations may be of large size, this will not always be easy.

Except in cases where we have geological evidence of former
greater spread, when of course the "fossil" area must be added
to that occupied by the living plants, we may leave out of
accovmt the more local genera in tracing phylogenies, and it is
clear that species or genera that are widely separated in space,
and in whose case no fossils can be found filling up the spacial
gap, cannot, without great risk of error, be looked upon as
necessarily closely related, however much alike they may be
(cf. 130, p. 346). Their resemblance may be due to parallel
mutation, and their ancestors may have been more widely
separated than they themselves are.

In the same way, no fossil that is not of wide dispersal can
safely be regarded as an immediate ancestor for anything that
is of equal or nearly equal dispersal, and still less if it be of
greater. Nor must widely separated fossils be regarded as nearly
related Avithout links. Nor is it safe to regard two layers as of
the same horizon without a number of fossils in common; and

Age and Area also throws light upon the question of Floral
Regions, which are usually defined as marked out by containing
large numbers and proportions of endemic forms, and as being
the better marked and more natural the higher the rank of these.
Great difficulty has always been encountered in defining such
regions; and to make them agree with those of the zoologists is
usually regarded as hopeless. In the accepted grouping of them,
the southern regions are very much smaller than the northern,
owing to the fact that endemics increase in number and propor-
tion towards the south (p. 149). Thus south-west South Africa
is regarded as a region equivalent to the Mediterranean region,
which includes all the land around that sea as far as Beluehistan;
and even to the whole of tropical America, including the ^Vcst

16—2

244 GEOGRAPHICAL DISTRIBUTION [pt. ii

Indies, a region which contains colossal numbers of endemic
forms. Other regions of absurdly small size are the Galapagos,
Juan Fernandez, the Hawaiian Islands, Kerguelen, etc.

It is difficult to understand why so much energy and labour
has been applied to the problem of differentiating floral regions,
for one fails to perceive any object which is gained by defining
them, for example, any progress in the study of geographical
distribution. The term floral region may, it seems to us, be added
to "constitution" in the extract from Huxley given on p. 231,
as one of the verbal anodynes by which the discomfort of igno-
rance is dulled. When we say that Lactoris fernandeziana (which
is now usually regarded as of family rank) is characteristic of
the Juan Fernandez floral region, it sounds as if we knew more
about it than if we simply stated the bald truth, that it occurs
upon the island of Juan Fernandez. In plain fact it is no more
specially characteristic of that island than Centaurodendron
dracaenoides, an endemic genus of Compositae, or Spergularia
rubra, which is of enormously wide distribution.

What it reallj^ comes to is that as, on the whole, in recent periods
of the world's history, migration of plants has been largely
southwards (owing to the cooling of the north), and the subse-
quent northward migration has not yet had time to show very
obvious results, the southern regions contain greater proportions
of endemics. In the same way, the islands being at the edge of
the dispersal that has gone on, where the oldest and most
variable (p. 218) types occur, and being also isolated, show great
numbers of them. But, as pointed out on p. 170, it must not be
forgotten that the larger regions of the world have greater pro-
portions than the small.

Ver}' little consideration is required to show that these divisions
or floral regions are very arbitrary, but very little trial of the
actual facts is needed to prove to an enquirer that it is a matter
of extraordinary difficulty to improve upon them. The islands,
by being clearly cut off from the rest of the world, are evident
divisions to make, but to divide the continents, except to cut
off a few such obvious regions as South Africa or ^Vest Australia,
is quite a different matter.

The one thing that comes clearly out is that endemics are not
a good test to apply, and with the new light that is thrown above
upon the question of endemism, it would seem probable that
this test vAW no longer be used. It gives a much greater pro-
portionate value to small areas in the south or upon islands than

CH.xxii] GEOGRAPHICAL DISTRIBUTION 245

they haA^e any real right to possess. Of the 32 floral regions
accepted in the latest work, 9 are upon islands, and 7 upon small
southern areas, and 12 in all are in the south, against only 1 i for
the very much larger land masses of the north!

The work upon Age and Area described above makes it much
more clear why these difficulties arise. No two genera, in all
hkelihood, will spread about the world at the same rate, so that
it is evident that what may be a marked floral region for one
genus of plants (or animals) will not be so for another, unless the
region has been well isolated for a long time, when it will, as in
the case of many islands, contain many endemics of many
different families. The whole subject requires a complete re-
consideration in the light of the results provided by Age and
Area, before it Avill be safe to try to divide up the world In this
manner. All that can be safely said at present is that regions
with great numbers of endemics in many families can be regarded

as regions tha"- ^ '' '" ' •■ • •

tive isolation.

as reo-ions that have existed for a long time, perhaps in compara-

Another thing that seems indicated by the work outlined above
is that in general the floras of the world, including those of most
of the islands, must have reached their present positions over
land or narrow straits which would not seriously interfere with
the passage of species. The arithmetical, systematic, and other
relationships between them, are too complex, and too evident,
to ha^e resulted from transport over wide stretches of sea, a
process which would sift out a Axry few from a comparatively
large flora.

A way in which Age and Area may proA'e incidentally useful
has been indicated above, and in a number of papers (126-134).
For example, in New Zealand (127, p. 452) a number of widely
distributed species, many more than would be expected, were
found in the class of smaHest area. On examination, they proved
to be, so to speak, the leavings of the flora. Twelve of 21 were
Monocotjdedons, four were from the neighbourhood of Kaitaia,
and so on; it was clear that many of them, though they perhaps
appeared to be really native, were in fact introductions to
the countr}^ Pomaderris apctala pro\-ed to be a xcry marked
exception among the plants of the Chathams in regard to its
distribution in New Zealand (129, p. 332), and therefore was
probably an introduction. The doubtful natives of Jamaica were

246 GEOGRAPHICAL DISTRIBUTION [pt. ii, ch. xxii

picked out in the same way, through showing irregularities in
regard to their distribution, judged by Age and Area (130,
p. 337), and so on. Whenever a species is found whose distribu-
tion is markedly different from what one would expect under
this hypothesis, that species is nearly always found to be an
introduction, or of doubtful identification, or in some way
irregular.

Sufficient has been said in Chapter xviii about the Hollow
Curve of Distribution, and both this subject and that of Evolu-
tion will be treated of in fuller detail in a later book. It is clear
that Age and Area becomes sim]:)ly a corollary of the larger law
that was indicated in what was said about Evolution.

There are many other directions in Avhich Age and Area may
prove to be a very useful hypothesis in dealing with problems
of distribution, but in the present somewhat controversial stage
in which the matter remains, it is better not to attempt too
closely to define, or even to outline, new positions. The fact
remains that Age and Area (with its subsidiary hypothesis of
Size and Space) is strongly supported bj^ very numerous facts
which demand an explanation that is largely mechanical, and
that the more inasmuch as the same type of facts is exhibited
both by animals and by plants. It is also clear that in dealing
with questions of Geographical Botany, the statistical method,
which has remained almost untouched since Hooker long ago
(p. 104) pointed out its usefulness, will probably plaj^ an
important part.

LIST OF LITERATURE

(1) Andrews, E. C. The Development and Distribution of tlje Legu-

minosae. Proc. R. S., N.S.W., xlviii, 1914, p. 333.

(2) The Development of the Myrtaceae. Proc. Linn. Soc, N.S.W.,

XXXVIII, 1913, p. 529.

(3) Arber, a. On the Law of Age and Area, in relation to the extinction

of Species. Ann. of Bot. xxxiii, 1919, p. 211.

(4) Wafer Plants. Cambridge, 1920.

(5) Bartlett, H. H. Mass Mutation. Bot. Gaz. lx, 1915, p. 425, and

Amer. Nat. xxxix, 1915, p. 129.

(6) Bateson, W. Presidential Address, Brit. Assoc., Australia, 1914.

(7) Bentiiam, G. Notes on the Classification, History and Geographical

Distribution of the Compositae. Linn. Soc. Jl. xm, 1873, p. 335.

(8) Bergson, H. Creative Evolution (Engl, trans.). London, 1914.

(9) Berry, E. W. A Note on the Age and Area hvpothesis. Science. xl\i,

1917, p. 539.

(10) Beyer, R. ...Ueberpflanzen ausserhalb der Tropen. Abh. B. I'.

Brandenb. xxxvii, p. 105.

(11) Black, J. .AL The NatnraUscd Flora of South Australia. Adelaide,

1909.

(12) Blaringhem, L. ct Viguier, P. Une nouvelle espece... Capsella

Mguieri...nee par mutation. Comptes Rendus, cl, 1910, p. 988.

(13) Breakwell, E. The Gramineae and the Age and Area hypothesis.

Proc. Linn. Soc, N.S.W., xlii, 1917, p. 303.

(14) Burtt-Davy, J. Alien Plants... Transvaal. Rep. S. Afr. .4ss. Adv. Sci.

1904, p. 252.

(15) Chandler, M. E. J. The arctic Flora of the Cam Valley at Barnwell,

Cambridge. Q. J. G. .S". lxxvii, 1921.

(16) Clements, F. E. Plant Succession . Washington, 1916.

(17) CoPELAND, E. B. Natural Selection and the Dispersal of Species.

Phil. Jl. Sci. XI, 1916, j). 147.

(18) — - — The Comparative Ecology of the San Ramon Polypodiaceae.

Ibid. 11,1907, p. J.

(19) Cockayne, L. Observations concerning Evolution.... Trans. X. Z.

Inst, xnv, 1911, p. 1.

(20) Coulter, M. C. Reviews in Bol. Gaz. Lxm, 1917, p. 419, and lxv,

1918, p. 116.

(21) Darwin, C. Animals and Plants under Dojnestication, 2nd ed. London,

1890.

(22) Life and Letters, ILd. by F.DaTv;in. London, 1888.

(23) Origin of Species. 6th cd. London, 1872.

(24) De Candolle, A. Geographic Botanique. Geneva, 1855.

(25) De Candolle, C. Monographiae Phanerogamarum. Paris, 1878.

(26) DE Yries, H. Die ciuleniisciic I'flanzcn von Ceylon und die niulierende

Oenotheren. Biol. Centr. xxxvi, 1916, p. 1.

(27) L'evolution des etres organises par sautes brusques. Scientia,

XIX, 1916, No. 1.

(28) The Distribution of endemic spp. in New Zealand. Science, xt,y,

1917, p. 641.

(29) The Mutation Theory (Engl, trans.). London, 1910.

248 LIST OF LITERATURE

(30) DE Vries, H. The Origin by Mutation of the endemic Plants of

Ceylon. Science, xliii, 1916, p. 785.

(31) The relative age of endemic species. Ibid, xlvii, 1918, p. 629.

(32) Van Amoebe tot Mensch (with Engl, trans.). Utrecht, 1918.

(33) Drude, O. Manuel de Geographic Bolanique (French trans.). Paris,

1897.

(34) Ernst, A. The new Flora of the Volcanic Island of Krakatau (Engl.

trans.). Cambridge, 1908.

(35) EwART, A. J. The Weeds... of Victoria. ^Melbourne, 1909.

(36) Farrow, E. P. The Ecology of Breckland. Journ. Ecol. 1916, et seq.

(effects of rabbits, 1917, p. 1).

(37) Floras employed include Babington's Cambridgeshire, Bolle's Canaries,

Britton and Millspaugh's Bahamas, Cheeseman's New Zealand,
Christ's Canaries, Englers Natiirlichen Pflanzenfamilien and
Pflanzenreich, Hillebrand's Hawaiian Islands, Hooker's British
India, London Catalogue, 8th ed., McNeill's Colonsay, Praeger's
Clare Island, Reiche's Chile, Spence's Orkneys, Thwaites' and
Trimen's Ceylon, and others.

(38) Gates, F. C. The revegetation of Taal volcano. Plant World, xx,

1917, p. 195.

(39) Gates, R. R. The Mulafion Factor in Evolution. London, 1915.

(40) Grisebach, a. H. R. Die Vegetation der Erde. 2 Aufl. Leipzig, 1884.

(41) GuppY, H. B. America's Contribution to the Story of the Plant World.

Journ. Ecol. ix, 1921, p. 90.
(4.2) Distribution of Plants and Animals. Peterm. Mitt. 1910, Heft 2,

(43) Fossil Botany in the Western World. .4mer. Jl. Sci. xlix, 1920,

p. 372.

(44) — — Observations of a Naturalist in the Pacific. London, 1906.

(45) Plant Distribution from an old Standpoint. Trans. Vict. Inst.,

April, 1907.

(46) ■ Plant Distribution from the Standpoint of an Idealist. Linn.

Soc. Journ. xliv, 1919. p. 439.

(47) Plants, Seeds, and Currents in the West Indies and Azores.

London, 1917.

(48) The Dispersal of Plants as illustrated by. ..Keeling or Cocos

Islands. 1'ict. Inst. 1890.

(49) The Island and the Continent. Journ. Ecol. vii, 1919, p. 1.

(50) The Testimony of the Endemic Species of the Canary Islands

in favour of the Age and Area Theory of Willis. Ann. of Bol.
XXXV, 1921, p. 513.

(51) Hemsley, W. B. In Biologia Centrali-Americana, London, 1879-88.

(52) Insular Floras. Sci. Progr. 1894, p. 27.

(53) Insular Floras in Rep. Bot. Challenger Exp. I, 1885.

(54) Hildebrand, F. Die Verbreitungsmiltel der Pflanzen. Leipzig, 1873.

(55) Hooker, J. D. Botany of the voyage..." Erebus'" and "Terror."''

(a) II. Flora Nov. Zeal. 1853; (b) III. Flora Tasmanieae. 1860.

(56) In Biologia Centr.-Amer. p. Ixii.

(57) Life and Letters. London, 1918.

(58) Outlines of the Distribution of Arctic Plants. Trans. Linn. Soc.

XXIII, 1800, p. 251.

(59) Huxley, T. H. The Gentians.... Jowr». Linn. Soc. xxiv, 1888, p. 101.

(60) Jessex, KxuD. Moserundersogelser i det nordostlige Sjaelland. Dan-

marks geol. undersog. ii Raekke, no. 34.

(61) JoHANxsEN, W. Elemente der exacten Erblichkeitslehre (Germ, trans.).

Jena, 1909.

LIST OF LITERATURE 249

(62) Jordan, A. Diagnoses d'espdces nouvelles ou miconnues. Paris, 1864.

(63) Kryshtofovich, A. A new fossil Palm and some other Plants of the

Tertiary Flora of Japan. Journ. Geol. Soc. Tokyo, xxvii, 1920.

(64) Leavitt, R. G. The Geographical Distribution of nearly related

Species. Amer. Nat. xli, 1907, p. 207.

(65) LoTSY, J. P. Die endem. Pflanzen von Ceylon und die Mutations-

hypothese. Biol. Centr. xxxvi, 1916, p.' 207.

(66) Evolution by means of Hybridisation. Tlie Hague, 1910.

(67) On the Origin of Species. Proc. Linn. Soc. 1914, p. 73.

(68) Lyell, Sir C. Antiquity of Man. 4th ed. London, 1873.

(69) Principles of Geology. 9th ed. London, 1853.

(70) MacDougall, D. T. Review of Age and Area. Plaiit World, 1916,

p. 79.

(71) MacLeod, J. List of literature relating to seed-dispersal. Bot.Jahrb.

Gent. III. 1891, p. 192. And cf. Small, no. 103, p. 182.

(72) Mark, J. E. An arctic Flora in Pleistocene beds, Barnwell, Cambridge.

Geol. Mag. 1916.

(73) Massee, G. a revision of...Cordyceps. Ann. of Hot. ix, 1895, p. 1.

(74) Matthew, J. R. The Distribution of certain elements in the British

Flora. Brit. Assoc. 1921 ; abstract in Journ. Bot. Jan. 1922, p. 26.

(75) Millspaugh, C. F. Flora of the Alacran shoal. Publ. Field Mus.,

Bot. II, 1916, p. 421.

(76) Morgan, T.H. Contr. to the Genetics of Drosophila. Washington, 1919.

(77) Evolution and Adaptation. New York, 1903.

(78) Nathorst, a. G. Contr. a la Flore Fossile du Japon. Acad. Roy. du

Sc. SuMe. Stockholm, 1883.

(79) Zur fossilen Flora Japons. Palaeont. Abh.l,\>ax\. 3. Berlin, 1888.

(80) Oswald, F. The Sudden Origin of new Types. Sci. Progr. xix, 1911,

p. 396.

(81) Pearson, H. H. W. The Botany of the Ceylon Patanas. Linn. Soc.

Journ. XXXIV, 1900, p. 300.^

(82) Reid, C. The Plants of the late glacial deposits of the Lea Valley.

Q. J. G. S. Lxxi, 1910.

(83) Reid, C. and E. M. The pre-glacial Flora of Britain. Journ. Linn.

Soc. xxxviii, 1908.

(84) The fossil Flora of Tegelen-sur-Meuse. Verb. k.Akad. Wctens.,

Amsterdam, xxii, no. 6, 1907.

(85) A further investigation of the fossil Flora of Tegelen-sur-Meuse.

Ibid. Versl. Afd. Natuurk. pt. xix, 1910.

(86) The Pliocene Floras of the Dutch-Prussian border. Med. Rijks-

opsporing v. Dclfstoffen, No. 6, 1915.

(87) Reid, E. M. Two pre-glacial Floras from Castle Eden. Q. J. G. S.

Lxxvi, 1920.

(88) A comparative review of Pliocene Floras. Jbid.

(89) Rech. sur quelques graines Pliocenes du Pont-de-Gail (Cantal).

Bull. Soc. Giol. France, xx, 1920.

(90) Ridley, H. N. Endemism and the Mutation Theory. Ann. of Bot.

XXX, 1916, p. 551.

(91) On the Dispersal of Seeds by Mammals. Journ. Straits As. Soc.

1894.

(92) On the Dispersal of Seeds by Wind. Ann. of Bot. xix, 1905,

p. 351.

(93) Samuelsson, G. Ueb. d. Verbreitung ein. endem. Pflanzen. .IrA-. /.

Bot. Stockholm, ix, 1910, No. 12.

(94) Seward, A. C. Darwin and Modern Science. Cambridge, 1909.

16—5

250 LIST OF LITERATURE

(95) SiNNOTT, E. W. Comparative Rapidity of Evolution in various Plant

Types. Amer. Nat. i, 1916, p. 466.

(96) The Age and Area Hypothesis and the Problem of Endemism.

Ann. of Bot. xxxi, 1917, p. 209.

(97) The Age and Area Hypothesis of Willis. Science, xlvi, 1917,

p. 457.

(98) SiNNOTT, E. VV. and Bailey, I. W. Foliar Evidence as to the Ancestry

and Climatic Environment of the Angiosperms. Amer. Journ.
Bot. II, 1915, p. 1.

(99) The Origin and Dispersal of Herbaceous Angiosperms.

Ann. of Bot. xxviii, 1914, p. 547.

(100) Small, J. Geographical Distribution of the Compositae. Rep. Brit.

Assoc. 1916, p. 509.

(101) Modern Theories of Evolution. Pharm. Journ. xcvii, 1916,

p. 612; xcviii, 1917, p. 3.

(102) The Age and Area Law. Science Progr. xii, 1918, p. 439.

(103) The Origin and Development of the Compositae. New Phylol.

Hepr. XI. London, 1919. Cf. lists of literature on pp. 182 (seed
dispersal), 214 (geographical distribution), and 242 (evolution).

(104) Solms-Laubach, Graf zu. Cruciferen-Studien. Capsella Ileegeri.

Bot. Zeit. 1900, p. 107.

(105) Tavlor, N. Endemism in the Bahama Flora. Ann. of Bot. xxxv,

1921, p. 523.

(106) Endemism in the Flora of the Vicinity of New York. Torrexja,

XVI, 1916, p. 18.

(107) Thellung, a. Flore Adventive de MonipcUier. Cherbourg, 1912.

(108) TowNSEND,F. Contr.. .. Flora. ..Scilly Is. Jo?<r«. Bo<. ii, 1864, p. 102.

(109) Treub, M. Notice sur la nouv. Flore de Krakatau. Ann. Buitenz.

VII, 1888.

(110) Wallace, A. R. Daricinistn. London, 1890.

(111) Island Life. 2nd ed. London, 1892.

(112) Went, F. A. F. C. Ueb. Zwecklosigkeit in der lebenden Natur. BioL

Centr. xxvii, 1907, p. 257.

(113) Willis, .7. C. A Dictionary of the Fhnvcring Plants and Ferns. 4thed.

Cambridge, 1919.

(114) Agriculture in the Tropics. 2nd ed. Cambridge, 1914.

(115) c7ilalogue of Ceylon Plants. London, 1911.

(116) Studies... Podostemaceae. ^??». Pcrarf. i, 1902, p. 267.

(117) The Flora of Ritigala, a study in Endemism. Ibid, in, 1906,

p. 271.

(118) Some Evidence against the Theory of... Nat. Selection of

Infinitesimal Variations.... Ibid, iv, 1907, p. 1.

(119) Further Evidence. Ibid. p. 17.

(120) The Geographical Distribution of the Dilleniaceae, as illus-
trating the Treatment of this subject on the Theory of Mutation.
Ibid. p. 69.

(121) Hill-Top Floras of Ceylon. I6id. p. 131.

(122) The Flora of Naminakuli-kanda. Ibid, v, 1911, p. 217.

(123) The Endemic Flora of Ceylon.... Phil. Trans. B. ccvi, 1915,

p. 307, and correction in Proc. li. S. B, Lxxxix, 1916.

(124) On the Lack of -Vdaptation in the Trist. and Podostem. Proc.

R. S. B, Lxxxvii, 1914, p. 532.

(125) The Origin of the Trist. and Podostem. Ann. of Bot. xxix,

1915, p. 299.

(126) The Evolution of Species in Ceylon, iftirf. xxx, 1916, p. 1.

LIST OF LITERATURE 251

(127) Willis, J. C. Distribution of Species in New Zealand. Ann. of Dot

XXX, 1916, p. 437.

(128) Relative Age of Endemic Species. Ibid, xxxi, 1917, p. 189.

(129) Distribution of Plants of the Outlying Islands. Ibid, xxxi

p. 327. '

(130) Further Evndence for Age and Area. Ibid, xxxi, p. 335.

(131) Sources and Distribution of the New Zealand Flora. Ibid.

xxxii, 1918, p. 339.

(132) The Flora of Stewart Island. Ibid, xxxiii, 1919, p. 23.

(133) The Floras ofthe Outlying Islands of New Zealand. Ibid. p. 479.

(134) Plant Invasions of New Zealand.... /6trf. xxxiv, 1920, p. 471.

(135) Endemic Genera of Plants. Ibid, xxxv, 1921, p. 493,

(136) The Age and Area Hypothesis. Science, xlvii, 1918, p. 626.

(137) Willis, J. C. and Burkill, I. H. The Flora of the Pollard Willows

near Cambridge. Proc. Catub. Phil. Soc. viii, 1893. p. 82.

(138) Willis, J. C. and Gardiner, J. Stanlev. The Botany ofthe Maldive

Islands. Ann. Pcrad. i, 1901, p. 45.

(139) Bews, J. W. Plant Succession and Plant Distribution in S. Africa.

Ann. of Bat. xxxiv, 1920, p. 287.

(140) — — Some general Principles of Plant Distribution, as illustrated by

the S. African Flora. Ann. of Bat. xxxv, 1921, p. 1.

(141) DE Vries, H. The present position of the Mutation Theory. Nature,

CIV, 1919, p. 213.

(142) Hooker, J. D. On Insular Floras. Journ. Bot. 1867, p. 23.

(143) Hayward, I. and Druce, G. C. The Advenlive Flora of Ticeedside.

Arbroath, 1919.

(144) Lang, W. D. Old Age and Extinction in Fossils. Proc. Geol. Ass.

XXX, 1919, p. 102.

(145) Willis, J. C. and Yule, G. U. Some Statistics of Evolution and

Geographical Distribution in Plants and Animals, and their
Significance. Xoture, 109, Feb. 9, 1922, p. 177.

INDEX

Acclimatisation, 29, 45. And cf.
Introduction

Acrotrerna dissectum, witii inter-
mediate stages, 219

Adam's Peak, endemics of, 55

Adaptation, 19, 55, 57, 59, 87, 148,
209, 210, 220, 224, 227, 229, 230,
231. And cf. Correlated varia-
tion. Endemic genera and species,
Evolution, Intermediates, Local
adaptation. Natural selection.
Species, Struggle for existence

Advantage as a guide in evolution,
189. 212, 215, 221, 225; ruled out,
215, 219, 221

Africa, endemic genera, 170, 178;
monotypes, 188, 189. And cf.
Canary Islands, Madagascar, Mas-
carene Islands, St Helena

Age, effects of, :3, 4, 5, 6, 61, 62, 85,
99, 174 71., 196: the most powerful
factor, 6, 197, 198

Age and Area, 4, 5, 6, 54, 61, 63
(statement of rule), 85, 103, 147,
150, 189, 206, 221, 225; applica-
tion to single species, 84, 85 ; to
animals, 200; to unallied forms,
86; and ecological results, 98; and
palaeobotanical study, 137, 147;
and the mutation theory, 222;
confirmation by prediction, 66,
76, 87, 230; in Australia, 64; in
Ceylon, 54; in Compositae, 119;
in'^New Zealand, 64; invasions,
76, 139, 234; objections to, 70,
84; position of the theory, 101;
reservations, 63, 70

Aim of nature, 205, 215

Alacran reef, endemics of, 152, 212

Allied forms only comparable under

age and area, 62, 85
America. See North America, South
America, Tropical America. And
cf. Alacran, Andes, Argentina,
Bahamas, Brazil, Chinese, Eu-
genia, Galapagos, Jamaica, .Juan
Fernandez, Lacioris, Mexico, New
York, Rio de Janeiro, Sequoia
Andes, endemics of, 176
Animals, applicability of age and

area to, 200
Anodynes, verbal, 231, 244
Anthemideae, 127, 135, 136
Antiquity and amplitude, 116
Anuradhapura, climate of, 43

Arber, Mrs A., 92

Arctotidcae, 126, 127, 136

Area, broken, 89; differences in,
occupied, 2, 33, 55; early species
gain in, on later, 34; fossil, 243;
graduation of, 58, 170, 229; in-
creasing more rapidly with age,
33; increasing with size of genus,
114—18; large, due to spreading,
11; minimum, occupied by com-
mencing species, 206; necessary
for origin, 10; of endemic genera,
171 ; of endemic species, 56, 150;
occupied, 2, 11, 33, 55, 115, 150,
151, 170, 191; phenomena matched
by those of size, 175; plants of
smaller, the younger, 206 ; possible
of occupation, 49; restricted, 57;
to which suited, 11. And cf.
Age and Area, Dispersal, En-
demic, etc.

Argentina, spread of introductions
in, 26

Argument of Part 1,7 ; of Partll, 107

Asia, tropical, endemic genera, 178.
And cf. Ceylon, Chinese, India,
Krakatau, Maldives, Monsoons,
Taal

Associations of plants, 20, 25, 30, 35,
50, 51, 229

Astereae, 129, 130, 131, 134, 135

Auckland Islands, 66-74, 230

Australia, age and area in, 64;
endemic genera of, 170, 190;
endemic species of, 150; grasses
of, 64. And cf. West .Australia,
New Caledonia

Average generic area, 126 ; necessary
in age and area, 61

Axioms of the systematists, 105, 217

Bahama Islands, endemics of, 64,

150, 210
Barriers to spread, 12, 13, 16, 20, 21,

36; especially Chap, v
Bateson, W.. 217
Beetles, hollow curve in, 202, 203,

236 ; logarithmic curve, 241
Birds, dispersal by, 12-18
Boraginaceae, distribution in New

Zealand, 161
Brazil, endemics of, 150, 170, 190,

233; endemic genera, 170, 176;

sizes of genera, 190. And cf.

Eugenia

254

INDEX

Breakwell, E., 64

Britain and Ireland, and outlying
islands, distribution in, 70, 73;
flora of genera and families by
sizes, 235; Hieracia in, 160; in-
vasions of, 234, 236; species per
family, 237. And cf. Cambridge-
shire, Clare, Colonsay, Ireland,
Orkneys, Scilly, Wicken, Willow,
etc.

Broken areas, 89

Burkill, I. H., 12

Cakile alacranensis, distribution of,
152

Calcutta climate, 43

Calenduleae, 126, 127, 134

Callitriche, distribution of, 92

Colli fris, distribution in Australia,
157, 64

Cambridgeshire flora, 235, 237

Campanula Vidalii, distribution of,
152

Canary Islands, flora of, 27, 88

Causes favouring or hindering dis-
persal, 32

Cenchrus insular ia, distribution of,
152

Ceylon, age and area in , 54 ; catalogue
of flora, 59; distribution of classes
in flora by area, 59; distribution
of species in, 59; dry and wet
zones, 14; endemics, 54, 55, 56,
150, 152, 233; endemics mainly in
big genera, 165; endemic genera,
170; genera by sizes, 235; spread
of introductions in, 25— 6. And cf.
Adam's Peak, .Anuradhapura,
Christisonia, Coleus, Doona, Eu-
genia, Hakgala, Hinidun-kanda,
HorUmia, Kandy, Ritigala, Schu-
machcria, Thwaites, Titlumia,
Trimen

Change in living world, rate of, 3;
of conditions, 39-44, 51 ; of tem-
perature, 44; greater in greater
time, 144

Characters, family and generic, 209;
intermediate usually not possible,
209, 211, 219; often appear late,
210; rudimentary beginnings of,
209 ; usually evolved withotit refer-
ence to usefulness, 226; usuallv
indifferent, 152, 209, 224, 225.
And cf. Intermediates,Species,etc.

Chatham Islands, 66-74, 230; plants
oldest in New Zealand, 67

Cherrapunji climate, 43

Chinese-North- American flora, 88,
140, 144

Christisonia, 151 ; distributionof,159

Cichorieae, 130, 135, 136
Cissampelos, distribution of. 159,

172
Cistaceae, distribution of, 172
Clare Island, distribution of flora, 70
Classes bv area, 60, 61
Clements', Y. E., 20
Climate, 40; climatic boundary, 45,-

changes of, as barriers, 40;

changes of, as directing forces in

migration, 138
Closed plant-societies, 20, 30, 50, 51
Cockayne, L., 26
Coffee, Liberian, acclimatisation of,

29
Coleus, dispersal of Ceylon species,

2, 54; C. elongatus, 2, 54. 151. 152,

208, 219
Colonsay, distribution of flora, 70
Commencing species, 10
Comparison of unallied forms, 86
Compositae. age and area in, 119;

distribution of. 18, 22. 48. 85,

Chap. XIII ; evolutionary history,

132; of Madagascar, 175; size and

space in, 132
Compound interest law, 241
Constitution of plant or animal. 231
Copeland, E. B., 5, 50
Correlated variations, 208, 209
Cosmopolitan genera, 21
Cromerian flora, 137-8
Cupressus macrocarpa, 88, 113
Currents, dispersal by, 14—17
Cyanca, distribution in Hawaii, 161
Cynara, spread of, 27; Cvnareae,

127, 134, 135
Cyrtandra, distribution of, 159; in

Hawaiian Islands, 160

Darwin, C, 4, 10, 204, 217. 222;
Darwinian theory. 102. 204

Destruction in struggle for exist-
ence, 113, 221

De Vries on Age and Area and the
Mutation Theory, 222

Diagnostic characters usuallv indif-
ferent, 152, 224, 225

Didymocarpus Per di la, species of two
individuals onlv, 151

Differentiation, 18, 103, 105, 221,
228 »., 240

Dillenin, distribution, 159, 219;
Dilleniaceae, family tree, 240

Dipterocarpaceae, distribution of,
19, 36, 85, 152

Dispersal of plants, 10, 12, 24, 32,
85, 101, 228; a measure of age. 61,
197, 230; by birds, explosions,
mammals, sea, vegetative repro-
duction. Mind, etc., 12-1 9; average

INDEX

255

rate probably very uniform, 99;
causes favouring or hindering, 32 ;
due to youth, 89, 92; into virgin
areas, 12, 14, 15, 19-21; niainlv
by land, 21, 182, 245; mainly
conditioned by barriers, 48;
mechanical, 21," 22, 3G, 61, 229;
methods, 12-22; of introductions,
24-27; of unrelated forms, 30;
oldest types at edge of, 218;
present, the maximum possible,
11, 197, 229, 230; rapid, 11, 19,
20, 26, 94, 229; rapid, not neces-
sary, 33; reached possible limit,
11, 27, 229; regular and irregular,
12; slow, 20, 30,33-52; to short
distance only, 14, 20, 32, 35;
withoutalteration of conditions, 25
Distribution, 101, 228; a closed
chapter, 11, 27, 230; determined
largely by time, 179; discontin-
uous, 11; general, 228; largelv
mechanical, 6, 61, 191, 203, 200,
214, 229; due to interaction of
many factors, 1, 2, 5, 85, 87;
limited, 3, 4; of endemics, 54, 58,
163, 166; of rainfall and moisture,
43-4; of wides greater than of
endemics, 60; to what due, 1,5;
vital factors in, 2. And cf. Dis-
persal, Endemics, Species, and
the various coimtries
Ditypes, 185

Doona in Ceylon, 94, 152, 161
Drosopfnhi, mutation in, 224
Dving out, 1 , 4, 10, 58, 81, 88, 89. 90,
"91, 93, 142, 144, 148, 165, 180.
192-3, 229-34. And cf. Inter-
mediates, Killing out. Relics, etc.

Early species gain upon later, 34
jBberrwfl/crrt, distribution in India, 163
Ecologv, 20; ecological barriers or
aids "to spread, 46, 99, 100; re-
sults, and Age and Area, 98
Effects of barriers, 12, 13, and

Chap. V
Elodea, dispersal, and spread, 17, 26,

27, 51
Endemic genera, 169; areas oc-
cupied, 170; as local adaptations,
57-8, 87, 166, 179; as relics, 166,
179, 182: as young beginners, 166,
179, 183; belong chieily to larger
families, 182; distribution of, 169;
explanations of, 179: of contin-
ents, 177-8; of islands, 17.5-83;
number increases with various
factors, 169; and with increasing
area, 169; phenomena paralleled
by genera of larger area, 176;

with small areas, 171. And cf.
Endemic species. Evolution, and
the various countries

Endemic species, areas occupied,
150; as local adaptations, 54, 87,
160, 186; as relics, 58, 81, 88, 93,
141, 148, 166, 186; as young be-
ginners, 16.'>-7; belong chiefly to
large genera, 91, 105; commoner
in the south, etc., 218; descended
from wides, 61, 74, 77, 86, 153,
221, 239, 240; explanations, 218;
families and genera to which be-
long, 164; increase to southwards,
etc., 149; localities in which occur,
149; of Bahamas. 64; Ceylon, 54;
Galapagos, 150; Hawaiian Islands,
150, 164 (including of endemic and
non-endemic genera, 163), India,
164; mountain tops, 54, 55, 58;
New York, 64; New Zealand,
or New Zealand and outlving
islands, 64, 66-74, 164: North
America, 86; resemblance to non-
endemics, 161 ; relation to wides,
61-, 77, 86, 198: their distribution
a special case, 163; tyi)es of dis-
tribution same as wides, 161 ; un-
related to wides, 86 H.; with
maximum numbers at certain
spots, 77-8; without differences
in conditions, 88; younger than
wides, 89, 221. And cf. Age and
Area, Dispersal, Distril)ution,
Local adaptation, Relics. Species

Endemism, 17, 54, 148, 106; a sign
of age, 84; and distribution,
(species) 148, (genera) 169: and
isolation, 148-9: especially to the
southwards, 149, 170; explana-
tions of, 166: on continents. 149:
on mountains. 149. 1.50. .And cf.
above, and Killing out. Local
adaptation. Relics, Wides, etc.

Epilobium, distribution in New
Zealand, 155

Eugenia, in Brazil, 157, 165; in
Cevlon. 58, 115, 157, 165

Eupatorieae, 126. 127, 134, 136

Europe, endemics of, 149. And cf.
Britain, Italy

Europeans in the Tropics, locations
.)f, 24

Euryale, formerly of great extension,
141

Evolution, bv infinitesimal varia-
tion, 2, 10. 207, 213; by differen-
tiation, cf. Differentiation: by
mutation, cf. Mutation : guided by
advantage or natural selection,
189, 212, 214, 215: mechanical,

256

INDEX

203, 205-6, 211, 214; predeter-
mined, 215; theory, 204; tree of,
surviving to present time, 221 ;
type of, 101; without adaptation,
57,58,224-6. And cf. Adaptation,
Natural selection, etc.
Exacum, distribution in India, 163
Exceptions to Age and Area, 67, 68,

84; exceptional species, 245
Expectation of life, 5
Explosions, dispersal by, 16
Extermination, 140, 238; regional,
140; specific, 142. Cf. Killing out
and Dying out
Extinction of species, 142-4. Cf.
Killing out and Dying out

Factors in dispersal, etc. Cf.

Causes, Dispersal, etc.
Families to which endemics belong,

182; sizes of, in hollow curves,

186. And cf. Genera
Faunas, local, 203
Festuca on different types of soil 38 ;

on the downs, 51
Fixity of vegetation at a given spot,

20
Fleshy fruits, 13
Floral regions, 243
Floras. Cf. under countries, etc.;

due to land connections, 182, 245
Fluctuating variation, 207, 211,213,

222
Foreign species, introduction and

spread of, 24
Forest, 42, 47, 51
Fossil areas, 243

Galapagos, endemics of, 150
Gardiner, J. Stanley, 14, 200, 202
Genera, as local adaptations, 189;
as relics, 189; ditypic, etc., 185;
endemic, 109; form more species
with increasing area, 117; formed
in a casual way, 234; grouped by
number of species, 186; largest in
largest families, 187; monotypic,
185; number of species in, related
to variety of conditions, 115; of
endemics only, 95, 18, 155; of few
species usually relics (?), 229; of
one or more species, 185; of Old
and New Worlds, 21 ; on both
sides of a barrier, 39; percentage
confined to various areas, 189,
190; possible size increasing with
increasing area, 178; sizes of, in
hollow curves. 174, 178, 186
Geographical distribution. Cf. Dis-
tribution, Limiting factor, Pro-
gress

Geological changes, 1, 52, and

Chap. XIV
Glacial period, 2, 172 ^?., 199
Gnaphalieae, 126, 128, 134, 135
Goats, effect of, upon vegetation, 26
Graduation, of areas of endemics

and wides, 00, 61 ; of areas of

genera from small to large, 170;

of areas of species from small to

large, 58
Gramineae of Australia, 64
Great Britain. Cf. Britain
Guunera, distribution of, in New

Zealand, 155
Guppy, H. B., 17, 49, 95, 101, 117,

130; his theory of differentiation,

18, 103, 221, 228, 240
Gymnema, distribution of, 159

Haastia, distribution in New Zea-
land, 153

Habit, types of, effects upon dis-
persal, 49

Hakgala (Ceylon), 151

Hawaiian Islands, age and area in,
64; Cyanea in, 161; Cyrtandra in,
160; endemics of, 150, 163, 164,
170; endemics of endemic and
non-endemie genera compared,
163; genera above average world
size, 164; Pelea in, 161. And cf.
Waialeale

Helenieae, 126. 131, 184, 136

Heliantheae, 128, 134, 135

Hclol)icae, size and space in, 116

Herbs, shrubs, and trees, 46; her-
baceous vegetation and drier
climate, 42: advantages of, 48;
younger than forest, 46

llicracia in Britain, distribution of,
160

Hindrances to dispersal, 32-53; to
progress, 228-9

Ilinidun-kanda species, 55

Hollow curves, 155, 161, 163, 166,
171, 174, 176, 180, 185, 186, 187,
188, 195 (Chapter), 199, 202, 205,
211, 214, 229, 235, 236-7, 240

Hooked truit, 12

Hooker, Sir J. D., on age and area,
4; axioms, 217: on Botanical
Geography, 6, 104; on dying out,
4; on natural selection, 205; on
proportion of mono- to di-coty-
ledons, 22; on general perman-
ence of species, 207

Horioiiia in Cevlon, distribution of,
159

Huxley, T. H., 2, 231

Hydrocotyle, 46; acclimatisation to
different climates, 30

INDEX

257

Increase of area occupied, 33
India, endemic genera of, 170;

genera by sizes, 235 ; genera above

average world size, 104
Infinitesimal variation, evolution by,

2,207,211,213,214,222
Interaction of factors in dispersal, 1,

2,5
Intermediates, between diagnostic

characters usually impossible,

209, 21 1 , 219 ; between genera and

species, not found, 214, 226; in

Acrotrema, 219 ; no need for them

to die out, 218
Introduction of foreign species, 24;

on continental areas, 25
Inuleae, 126, 134, 135; diphyletic

origin, 126; limited, 126, 127, 136
Invasions, 20, 234; of New Zealand,

76, 139
Ireland, flora of, 236
Irregular dispersal, 12-16
Islands and endemics, 148-50, 175-

83; monotypes, 188-9
Isolation, 17, 148, 169, 170
Italy, flora of, 236

Jamaica, age and area in, 64
Jordanian species, 215-21
Juan Fernandez, endemic genera of,
169-70, 244

Kandy climate, 43
Kermadec Islands, 66-74, 230
KiUing out, 1, 157, 135, 142, 144
Krakatau, flora of, 15

Lactoris, distribution of, 244

Land connections, 21, 182, 245

Landslips, 37, 48

Large families and genera the suc-
cessful ones, 113

Larger genera, 117, 185; on larger
areas, 178

Largest families in the world, 21

Light, effects of, 45

Light seeds, 13

Limit of distribution, 45

Limiting factor in progress , 3 ,205 ,228

Linnean species, and splitting, 98,
216, 218, 221

Literature, 247

Local adaptation, 54, 57, 58, 87, 148,
216, 231; species, 50, 151, 217
(and cf. Endemic) ; distribution, 3
4; faunas and hollow curve, 202
floras and hollow curve, 236
migration, 20, 32, 35

Lofgren, A., 206

Logarithmic curves, 241

Lyell, Sir C, 3, 20, 219

Madagascar, endemic genera of,
175, 178; sizes of endemic genera.
190 ^

Maldive Islands, flora of, 14

Mammals, dispersal by, 19

Man, action of, 52

Mascarene Islands, endemic genera
of, 169, 170

Matthew, J. R., 234

Mechanical explanations necessary,
89, 183, 206, 232, 233

Mechanisms for dispersal, 12; not
imperative, 34

Menispermaceae, distribution of, 172

Mesophytic plants, dispersal of, 49;
adaptation rare in, 210

Mexico, endemics of, 150

Meyrick, E., 200

Microspecies, 98, 216

Migration, 138

Mogi flora (Japan), 145-6

Moisture of air, distribution of, 43

Monimiaceae, distribution of, 174

Monocotyledons in islands off New
Zealand, 230

Monotypic genera, 185; areas oc-
cupied by, 191; as relics, 186,
191-3; as special adaptations,
186, 191-2; descended from larger
genera, 240; explanations of, 192;
greatest proportion in largest
families, 192; increase south-
wards and outwards, 193

Monsoons, 14, 41

Moribund species. Cf. Relics, and 148

Mountains, as agents facilitating
migration, 37; as barriers, 36, 40;
as last resorts, 58; and climate,
40-42; and endemics, 55, 92, 149;
endemics as reUcs, 92; endemic
genera of, 176

Multiple origin, 11, 47, 105

Mutation, 208, 211-21, 222 (de
Vries), 223; causes of, 213; large,
216; Lyell on, 219; parallel, 243;
several, not necessary for forma-
tion of species, 218; size of, 215;
small, 216; theory and age and
area, 222
Mutisieae, 126, 127, 131, 135

Najas, distribution of, 159; fossil
record of, 143

Natural selection, and explanations
based upon it, 10, 58, 61, 104, 148,
188, 198, 199, 204, 206, 208-14,
220, 229; a destructive and
negative agent, 220; governing
general outlook upon biological
problems, 228

Nest making, 12, 13

258

INDEX

New and Old World genera, 190
New Caledonia, endemic genera of,

169, 170
New forms at commencement of
life, 212, 213; most frequent at
edges of dispersal, 218 ; range from
small to large, 220
New species, formation of, 34-5
New York, endemics of, 04
New Zealand, age and area in, 64;
endemics of, 150; of, and islands,
69; endemics belong to large
genera, 165; endemic genera, 170,
171 ; flora of outlying islands, 66,
72; genera above average world
size, 164; genera by sizes, 235;
invasions of, 76, 139; predictions
about flora of New Zealand and
islands, 66-74; Ranunculus in,
153; species per family, 238;
spread of introductions in, 26; the
most irregular curve of all, 196.
And cf. Auck lands, Chathams,
Epilobium, Gunnera, Ilaasfia,
Kermadec, Monocotyledon , Olear-
ia. Outlying, Ranunculus, Stewart,
Vegetation, etc.
North America, spread of introduc-
tions in, 26; endemics of, 86;
monotypes, 188, 189

Objections to hypothesis of age and

area, 70, 84
Oenothera, mutation in, 224
Oldest and most variable types at

edge of dispersal, 218; living

species, 143
0/earia, distributi.in in New Zealand,

101
Open plant societies, 20, 27, 50,

213
Origin of species, 10, 204
Orkneys, distribution of flora, 70
Outlying islands of New Zealand,

flora of, 66

Pacific Islands, plants of, 17
Palaeobotanical study and age and

area, 137
Palaeotropical genera, 190
Pangenesis, 222
Parent and child occur together,

219, 220, 221
Pelea, distribution in Hawaiian

Islands, 161
Permanence of species, 207
Phylogeny, 240
Physical barriers, 36
Plant migration, 137
Plant societies or associations, 20,

50

Podostemaceae, 4; distribution, 57,

92; characters, 210
Polemoniaceae, distribution of, 171
Pollard WiUow flora, 12
Polyphyly, 11, 47, 105
Pomaderris apetala, distribution, 67
Pont-de-Gail flora, 137, 143, 146
Prediction, 66, 76, 87, 230
Progress in knowledge of geographical

distribution, 3, 228, 229. Cf.

Limiting factor

Rainfall, 41-4; distribution of, 43
Rank and range, 105, 118, 130
Ranunculus, 153; distribution in

New Zealand, 1.53, 163, 216, 220,

239
Rapid spread of introductions, 24,

25, 94
Raylcigh, Lord, 33, 145, 152, 212
Regional extermination, 140
Regression, 207
Regular mechanisms for dispersal,

12-19
Reid, Mrs E. M., 82, 137
Relics, 86, 88, 93, 186, 192-3, 199,

216, 229, 231-3; explanation of

endemism, 58-9. And cf. Djing

out, KiUing out
Reservations in regard to age and

area, 63, 70
Reversion, 207
Ridley, H. N., 18, 151
Rio de Janeiro climate, 43
Ritigala, and flora of, 14. 54, 55
Rivers as barriers, 37
Rubiaceae, logarithmic curve, 241

St Helena, endemic flora, 150;

spread of introductions in, 26
Salsola Kali, distribution of, 49
Schumacheria in Cevlon, distribution

of, 159
Scillv Islands, distribution of flora, 70
Scott, H., 202
Sea, dispersal by, 14-17; as barrier,

36
Seed, quantity of, necessary for

transport to a distance, 32
Senecioneae, 126, 128, 134
Sequoia, formerly of great dispersal,

141
Sinnott, E. W., 95; and Bailev, I.

W., 40
Size and space, 71, 74, 113 (chapter),

115, 171-2, 174, 178, 185, 188,

190, 197, 233; in Britain, 113; in

Compositae, 132 ; in Helobieae, 110
Sizes of families in hollow curves,

186; of genera in hollow curves,

174, 178; of mutations, 215

INDEX

259

Small, J., 18, 119

Societies, plant, 20, 27, 32, 50-2,

229
Soil as barrier or assistance to

spread, 38
South America, endemics of, 190;
monotypes and larger tfenera,
188-90

Sparliiia, spread of, 26

Specialisation of plants, 49 (twice),
50

Species, best limited when of com-
plex floral structure, 217-8 ; causes
favouring or hindering dispersal,
32; commencing life, 36; diagnostic
characters usually indifferent,
224-6; dispersal, cf. Dispersal;
early gaining on late, 34; endem-
ism and distribution, 148; foreign,
introduction and spread, 24;
general permanence, 207; going
under, cf. Relics; least complex
that are most widely distributed,
218; local, 50, 216, and cf. En-
demics; occupying just those
places to which suited, 229, 230;
occupying overlapping areas, 57;
of large genera often resemble
varieties, 217; on smaller areas in
general younger, 206; per family
or genus in local floras, 237-8;
that vary most, 217. And cf.
Endemic species. Evolution, Local
adaptation. Relics, etc.

Specific extermination, 135

Splitting of Linnean species, 98

Sports, 211

Spread of introductions, 24; with
alteration of conditions, 25-6;
often rapid, 27

Statistical treatment of geographical
distribution, 6. 246

Stewart Island, 71, 72

Strut lot es, succession of species, 143

Struggle for existence, 50, 148, 206,
210, 213, 220-1, 238

Successful and unsuccessful species,
55

Succession, 20, 51, 138

Survival of species, 142

Swamping, 95, 18

Systematist, the, 101, 105, 217

Taal volcano, revegetation of, 16
Taylor, N., 64

Temperature changes as barriers, 44
Tertiary flora, 2, 34, 49, 88, 137, 233
Theorv'of differentiation, 18, 103,

105, 221, 228 n., 240
Thiselton Dyer, W. T., 49

Thwaites, G. H. K., 151

Time available for evolution and

dispersal, 33, 145, 152, 212
Tithonia, dispersal of, 17, 26
Tree, ancestral, of genus or family,

surviving, 20, 243 21, 24
Trees, of multiple origin, 47. And

cf. Herbs
Trees, shrubs, and herbs, 46
Treub, M., 15
Tribulus alacranetisis, distribution

of, 152, 212
Trimen, H., 54, 56
Tristichaceae, dispersal of, 92
Tritypes, 185
Tropical America, endemic genera of,

190; Asia, endemic genera of, 190
Type of vegetation, as barrier, 50-1

Unallied forms not comparable

under age and area, 63, 85, 86
Useless characters, 209

Variation a centrifugal force, 105;
variations, correlated, 208; most
common in genera simplest in
structure, in species of larger
genera, and in wide-ranging
species, 217

Vegetation, of northern type in New
Zealand, 40; type of, 'as barrier,
50-1

Vegetative reproduction, 16

Verbal anodynes, 231, 244

Vernonieae, 126, 127, 136

Virgin soil, dispersal into, 12. 14, 15,
19-21

Mtal factors in distribution, 2, 4

Waialeale, climate, 43

Water-plants, dispersal of, 49

Went, F. A. F. C, 205

West Australia, endemism in, 149,
169, 170

Wicken Fen flora, 235

Wides, 59; endemics descended
from, 61, 74, 167,221,239; first
to appear, 239; most widely dis-
tributed in a country, 60; of
wide dispersal, 84; oldest forms,
61,72, 239

Widespread genera, 21

Willow, pollard, flora, 12

Wind, as barrier, etc., 45; dispersal
by, 13-17

World, endemic genera of, 178

Youth, greater distribution due to,

89, 92
Yule, G. U., 241

PRINTED IN ENGLAND

AT THE CAMBRIDGE UNIVERSITY PRESS

BY J. B. PEACE, M.A,

This book is a presen'ation facsimile.

It is made in compliance with copyright law

and produced on acid-free archival

60# book weight paper

which meets the requirements of

ANSI/NISO Z39.48-1992 (permanence of paper)

Preserv'ation facsimile printing and bindinj

by

Acme Bookbinding

Charlestown, Massachusetts

2006