Non-UK authors omitted
THE DEVELOPMENT OF DARWIN'S GENERAL BIOLOGICAL THEORIZING M.J.S. HODGE
Charles Darwin (1809C1882) was only twenty seven when he returned, in 1836, from the five-year voyage of HMS Beagle.
What is more, within the next five years he reached all those conclusions in general biological theory that his books were to expound three decades later.
For, by 1841, he had worked out not only his theory of the origin of species, natural selection, but also, it seems, his theory of generation (or reproduction, including heredity, variation and so on), pangenesis.
This chapter surveys the development of his general biological theorizing over that remarkable early period.
The analysis draws throughout on the work done in the last decade by Gruber (1974), Herbert (1974, 1977), Ghiselin (1975), Ruse (1975a, b; 1979), Schweber (1977, 1980), Kottler (1978), Manier (1978), Sulloway (1979, 1982a, b), Kohn (1980), Ospovat (1981), and Sloan (1983a, b) and is derived from studies by the present writer (Hodge, 1982, 1986; Hodge &amp; Kohn, 1986) where full reference is made to the documentary sources and secondary literature.
Such a survey can serve more than mere biographical curiosity, and a final section will suggest how it may clarify some issues of current interest to historians, to philosophers and to biologists.
It can also free us from many mistaken myths about Darwin himself.
These myths mostly trace to his own misleading reminiscences later in life, and have been relentlessly reaffirmed since, at the 1959 centennial symposia for example and in the 1978 BBC-TV series on Darwin; but they are nonetheless discredited by the scholarly industry now grown up around the rich manuscript archive from Darwin's early years (Kohn, 1986).
One is the romantic, really Wordsworthian, individualist myth so dear to the literary guardians of English national cultural stereotypes.
It depicts the young Darwin as a lone, sporting gentleman, an amateur beetle-collector seeing nature as she really is by simply looking with the clear gaze of genius, unimpeded by any scientific training, theological prejudice, professional ambition and so on.
Another is the Whiggish, anachronistic myth that Darwin's general biological thought consists of a molecule comprising just two atoms: the idea of evolution and the idea of natural selection.
It depicts his early intellectual development as reducing to two moments of discovery, whereby he moves from having no coherent ideas to having just those ideas.
Fortunately, there is a single antidote effective against both these myths;, and that is to start all over again with the most decisive source of Darwin's new identity, on the voyage, as a committed man of science: his zealous discipleship of Charles Lyell's (1797C1875) views in geology (including biogeography and ecology).
This antidote is effective against the romantic-individualist myth, because, as a protg of Lyell, the young Darwin of the Beagle is at once invested with all the intellectual and institutional context that that myth would suppress.
It is effective against the evolution-plus-natural-selection myth, because it forces us to reconstruct the narrative of his subsequent theorizing not as so many unknowing steps towards his final positions, but as so many deliberate departures from positions initially shared with his mentor.
Darwin as a protg of Lyell (1834C7)
Darwin's acceptance of Lyell's views was complete by mid-1834 when he sailed around the Horn.
Having now studied all three volumes of the Principles of Geology (1830C3), he was applying Lyell's entire system, physical and organic worlds alike, to South America and beyond that the whole earth.
Henceforth this system provided the framework for his preoccupation with the problems of the extinction and origin of species.
All the causes of change are presumed, in Lyell's system, to persist undiminished into the present, the human period, and on into the future.
Now, as at all times, habitable dry land is being destroyed by subsidence and erosion in some regions, while it is being produced by sediment consolidation, lava eruption and elevationary earthquake action in others.
Equally, Lyell has the long succession of faunas and floras brought about by a continual, one-by-one extinction of species and their replacement by new ones.
The epistemological rationale for his presumption, of the persistence of all such causes of change into the present and future, is the ideal of explanation by real or existing causes, verae causae.
Like his friend (and the undergraduate Darwin's scientific hero) the physicist John Herschel (1792C1871), Lyell followed earlier writers, most notably the Scottish philosopher Thomas Reid (1710C1796), who had drawn this moral from the superior evidential credentials of the Newtonian gravitational force over the Cartesian ethereal vortices: any causes invoked in an explanatory theory should, ideally, be known to exist through direct observation independently of the facts they are supposed to explain.
That force, unlike those vortices, was a well-evidenced explanation for the planets' orbits because, the argument went, the orbits themselves were not the sole evidence for its existence.
It was a real not a conjectural cause; for it was known to exist from observations of swinging pendulums and falling stones down here on earth.
Lyell's system was, therefore, to exemplify an epistemological analogy.
In geology, only causes active in the present, human period are accessible in principle, although often not in practice, to direct observation.
So, in this science, the present is to the past as the terrestrial is to the celestial in Newtonian physical astronomy.
Such, then, was the context, at once systematic and epistemological wherein Darwin, from 1834 on, was theorizing about species extinctions and origins.
His own thinking over the next three years was developed, accordingly, through successive disagreements with Lyell's views on the organic world, while he continued to accept his mentor's teaching on the physical world of land, sea and climate changes.
Lyell had made adaptational considerations alone completely decisive in determining the timing and placing of both species extinctions and species origins.
Any species must eventually become extinct; for changes in local conditions will sooner or later allow other species better adapted to the changed conditions to invade and conquer in the struggle for existence.
As for species origins, Lyell had argued against spontaneous generation and against new species arising by the modification (' transmutation ') of older ones.
But, in offering no positive alternative account, he had left the means whereby new species originate quite mysterious if not miraculous.
He had held explicitly, however, that when and where any given species is created is determined by the conditions it needs to flourish.
Conversely, the character and so the supraspecific group membership of the species that have originated in any area is determined by conditions there.
So, on Lyell's account, if two areas are very similar in conditions they will have congeneric or cofamilial endemic species; if they are very dissimilar, then their endemic species will be of distinct families or orders.
Now, by early spring 1837, Darwin had decided that such purely adaptational explanations could not account adequately for the timing and placing of either the extinctions of old species or the origins of new ones.
And he was emphasizing, in his Red Notebook, the parallel in the explanatory inadequacy of adaptation in the two cases (Herbert, 1980).
Already, in 1835, he had concluded that several species of large mammals, formerly flourishing on the eastern plains of South America, had later become extinct while no change in conditions physical or ecological had occurred.
And he had adopted another theory of extinction, discussed but rejected by Lyell, wherein species are like individuals and die of old age.
A species, as a succession of organisms produced sexually, might have only a limited total lifetime, Darwin argued; just as a succession of apple trees propagated by grafts was supposed to last only so long before degenerating and dying as if it were merely the extension of a single limited life.
As for the origin of species, reflection on many palaeontological and biogeographical facts, including those established by the expert judgments of Richard Owen and John Gould on the voyage material, had convinced Darwin, by early spring 1837, that the close similarity seen between any congeneric endemic species in two areas was not always explicable as a common adaptation to common conditions; for often the two sets of species were endemic in areas with very different climate, soil and so on.
However, if more recent species could descend from earlier ones, ancestry could explain what adaptation could not.
Why did the species on younger land  the Galapagos islands or the southern Patagonian plains, for instance  often resemble closely other species on the nearest older land?
Because they were descended from them, many sometimes descending from a single ancestral species.
Thus did Darwin conclude that resemblances between species are often not due to adaptation but to inheritance from common ancestors; while differences are often adaptive and are due to differing, multiple divergences from those common ancestors.
It is not, as he saw it, that species are not exquisitely adapted to their respective places in the economy of nature.
They are and any theory of species origins must explain why.
It is simply that adaptation and ancestry can explain what adaptation alone can not.
On such grounds Darwin had decided, by that spring, for transmutation and common descent.
Always the bold ' philosopher ' as much as the cautious' naturalist ', he soon went far beyond these disagreements with Lyell.
He did so in a new sequence of theoretical notebooks, B-E (1837C9) and M- N (1838C9)  M for ' metaphysics', meaning mind, man, materialism, morals and so on.
(See a note at the end of this chapter on editions of these texts.)
He did so, moreover, in two ways that were consciously conditioned from the start by his immediate context and resources in 1837 as a biological theorist.
First, from way back in his Edinburgh days and his apprenticeship to Robert Grant in invertebrate zoology, he had been much pre-occupied with comprehensive generalizations about sexual and asexual modes of generation (Sloan, personal communication).
His extinction theorizing had thus been developed, since 1835, at the first intersection of that old preoccupation with his newer devotion to Lyellian geology.
Now, in the summer of 1837, he would understand the origins no less than the extinctions of species through appropriate comparisons and contrasts between sexual and asexual generation.
Having read again a book he had admired when studying at Edinburgh, a book much concerned with precisely such comparisons and contrasts  his grandfather Brasmus Darwin's Zo?nomia (1794C6)  he was soon taking its title for the opening heading of his Notebook B, where he was now to pursue his own inquiry into ' the laws of life '.
He was soon making explicit, too, a fundamental teleological analogy, in which changing conditions are to sexual generation as sexual crossing is to asexual generation.
Although it can extend an individual life, continued grafting eventually brings death without issue; and likewise sexual generation in unchanging conditions brings death, extinction of the species, without issue.
With sexual crossing, however, an individual whose life has been extended by grafting, although not enabled to go on itself, can give rise to a new individual with a new lease on life.
Likewise, then, in changing conditions, sexual generation is the providential means whereby a species can change into another, a new species adapted to the new conditions, so avoiding the extinction without issue that would otherwise occur.
Second, Lyell had outlined very fully  and of course rejected  the comprehensive transmutationist system of biology developed by jean Lamarck, praising the French zoologist for his courage in extending his repugnant ideas to our own species, man.
Darwin accordingly opened his Notebook B, in July 1837, with an integrated sequence of entries whose twenty-seven pages of argumentation were to match the structure of Lyell's exposition of Lamarck.
Darwin, like Lamarck  as presented (and misrepresented) by Lyell  would now trace any degrees of difference, no matter how wide, to long-run divergences from common stocks.
And he would trace any higher levels of bodily organization and mental faculties, explicitly including those in man, to long-run progress from remote starting points in the simplest organisms of all, infusorian monads.
With these decisions Darwin became more than the protg of Lyell.
He would be ever hereafter his own successor in developing sequels to the steps he had taken in the spring and summer of 1837.
The new programme pursued (1837C8)
Darwin himself often reflected in his notebooks on his new programme's presuppositions concerning God, nature, man and science.
God was, for Darwin then, still the traditional good and wise creator, but one never working in so many separate acts of miraculous interference, always through the natural consequences of a few initial enactments of general causal laws: as with planetary orbits and the law of gravitation, so, Darwin insisted, with species origins and the laws of generation.
As for man, he is a species produced like any other, lawfully, his mental faculties the causal consequences of his bodily organization and not miraculously superadded; while science is a quest for lawful causes that are evidenced both directly and independently of the many, diverse facts they can explain, and indirectly and dependently, by the very multitude and diversity of those facts.
The place in the programme of his generational theory of species origins Darwin understood through the analogy of a tree of life.
By the autumn of 1837, he had developed this analogy so as to understand all long-run trends in diversification and progress through an arboriform extrapolation, on a changing but stable Lyellian earth's surface, of successive species propagations; these being analogous to the successive bud propagations whereby any tree grows, with many buds ending without branching, in species extinctions, while other buds branch without ending, in species multiplications.
His species propagation theorizing itself was accordingly constructed, from the very opening of Notebook B, as an argument that starts with the sexual generation of one individual organism from another and ends with the propagation of one species from another.
Moreover, this species propagation is ultimately made possible by the two features that Darwin sees distinguishing sexual from asexual generation: namely, maturation in the offspring and the interaction of two parents in their production.
Thanks to the impressionability of immature organization, hereditary adaptive variation accompanies sexual generation in changing conditions.
But how then can any species be constant in character across its entire range?
Because crossing with the blending of parental characters keeps the species constant as long as the conditions are constant overall, only changing temporarily and locally.
Conversely, then, a new variety can be formed if this conservative action of crossing is circumvented by reproductive isolation of a few individuals in new conditions, whether that isolation arises with or without geographical segregation.
And how may this variety formation proceed to new species formation.)
As Darwin knew, the usual criteria for specific rather than mere varietal distinctions were those that Swedes and Italians do not satisfy, but lions and tigers do: namely, true breeding, lack of intermediate forms and unwillingness or inability to produce fertile hybrids on crossing.
So, the final stage in Darwin's argumentation concerns how a species meeting those three criteria would eventually arise with prolonged isolation and divergence.
Over the next year, through the summer of 1838, it is the variety and species formation stages of this argument that Darwin develops most fully and explicitly.
The character gaps between good species he explains by continued divergence and by the extinction of intermediate varieties.
The true breeding he explains by the law that in crossing the characters of an older domestic race dominate those of a younger one.
This dominance shows, Darwin reasons, that older characters are more permanent, more deeply embedded in the hereditary constitution and so more resistant to the influence either of mates in crossing or of changing external conditions.
Such permanent divergences in character have often arisen in the distinct races of a domesticated species, and in the wild they would be accompanied by an unwillingness or inability to interbreed; for there, Darwin argues, the reproductive system, with the associated instincts, is not disrupted as it is in domesticated species.
So, for Darwin, in the early summer of 1838, the races of domesticated species are providing positive analogies, especially for the final, species formation, steps in his argument.
But at the other, the opening end of that argument, concerning the initiation and transmission of adaptive change, they are a source of contrasts and not of comparison.
Thus he says that many varieties of domesticated species, fancy pigeon breeds for example, are monstrous not adaptive; they can only be maintained by artificial feeding and breeding, including selective breeding; they are quite unlike wild, natural and adaptive varieties and even more unlike wild species.
Now, as Darwin sees it, at this time, for a change to be an adaptation it must be more than merely hereditary and advantageous; it must be necessary rather than accidental, elicited, that is, by the very conditions that make it advantageous, as albinism sometimes seems to be by cold.
And it must be acquired by the whole of the race faced with those conditions, so as not to be lost in crossing.
Then, as with reproductive isolation, changed habits can initiate permanent adaptive changes in structure, especially in higher animals by entailing changes in the conditions of foetal maturation.
All these developments, in Darwin's comprehensive species propagation argumentation, are reinforced and not rejected in further developments, consummated in mid-September, 1838, in his Notebook D conjectures about sexual and asexual generation, more particularly about sexual buds (or ova) and asexual buds.
A sexual bud or ovum has started life, he supposes, as a bud from the mother when she was herself a newly fertilized egg in her mother, the grandmother.
Why then does it not eventually become an exact facsimile of that mother?
Because, he answers, of two lots of differences.
During its maturation it is subject to conditions not exactly like those the mother matured in; and at fertilization it is acted on by a mate with a constitution unlike its own and unlike its mother 's.
The whole object of sex is to have unlike acting on unlike so as to make possible the production of unlike offspring, thus allowing for adaptive change, Darwin argues.
But where does the constitutional unlikeness of the two mates come from?
Well, since each has the same ancestry, their constitutional difference must trace to the cumulative influence of different conditions in the two lines of descent from those ancestors.
So, sexual generation is the means whereby the past influences of changing conditions can be accumulated and combined with present ones to ensure continued variability.
If conditions change only locally and temporarily then only individual differences will result, but, Darwin concludes, if conditions change overall and permanently then a new species will eventually be formed.
Darwin is pleased with this novel analysis of how sex enables new species to arise from old.
But he admits that it leaves as mysterious as in 1837, how the maturation of the individual fertilized ovum ensures that the changes be adaptive in each generation.
He can only conjecture that additional maturational innovations will not become hereditary unless they harmonize with the previous ones that are already being recapitulated in maturation.
Adaptive innovations could thus be separated, as the ones that are hereditarily transmitted, from the maladaptive, as the ones that are not; although Darwin notes that hereditary diseases show that this separation is often fallible.
Such is Darwin's species origin theorizing in mid-September 1838.
It compounds still further those two legacies so actively conjoined since the previous summer of 1837: the historical, biogeographical (including ecological) concerns that he had inherited from Lyell, and the generational concerns deriving from his study with Grant and subsequent reading in Erasmus Darwin.
Consider, then, what geography and generation together have done for Darwin's understanding of the problems of organic diversity and the origin of species.
First, his explanation of resemblances and differences among species contrasts directly with any developmental stage theorizing as found most famously in Robert Chambers, Herbert Spencer or Karl Marx.
A developmental stage theory refers similarities and differences to more-or-less-equal advances made from the lowest point on a universal scale of advance.
Thus Chambers explains the similarities between Old and New World monkeys as due to life having advanced to the same level of organization from quite independent origins of life in the two hemispheres (Hodge, 1972).
By contrast, for Darwin as a genealogical (ancestry and descent) theorist of historical biogeography, resemblances and differences are not traced to developmental advances, but to ramifying migrations and adaptive divergences from common ancestors that are more or less remote from their diverse descendants in time, place and character.
Now, Darwin at this time is explicitly taking each organism's ontogeny to recapitulate its phylogeny.
But his very assumption that many descendant species may diverge from a single common ancestry precludes his construing descendants as grown-up ancestors and ancestors as descendants in embryo; for any given immature, embryonic life already has but one determinate mature adult future; if a puppy as a dog, if a tadpole as a frog.
It is, then, because he is explaining differences and resemblances as he is that Darwin, in 1838, needs a theory of purely opportunistic adaptive change in changing conditions, a theory making no developmentalist assumption as to a preferred direction that life will take provided it can go on at all.
And his thinking is indeed knowingly premised on the assumption that in the complete absence of change in conditions there would be no changes in organization; so that whatever different changes in organization have occurred in the many lines descending from some common stock are due to differences in the conditions in those lines.
Darwin, as a genealogical biogeographer, thinking horizontally as well as vertically, to use Ernst Mayr's (1982) terms, would explain change generationally but not developmentally.
This integration of horizontal, genealogical, geographical and generational constraints is also the source for what Ernst Mayr (1982) calls Darwin's populationist thinking about species.
For it has led Darwin to think of each species as spreading out into varying conditions, over a range, over time.
And only the extension of a species, not its intension, as philosophers say  only the collective membership of a species, not the properties earning the members their membership in it  can have a geographical range and a historical career.
Now, recall his account of individual adaptations to conditions as arising in individual maturations in sexual generation; and recall his account of hereditary variation as embedded constitutionally through successive individual matings.
And then it is clear how Darwin has come to be thinking of a species as a population of individuals, with each member differing from every other because arising in a unique sequence of influences exerted by a unique succession of conditions and mates.
The origins of natural selection and pangenesis (1838C42)
Darwin's mid-September 1838 theorizing, as developed in his Notebook D, provided the immediate context for his most celebrated innovation of all: the theory of natural selection.
For  contrary to the legend that it was all thought through in a day  this theory was worked out in three main stages over the next half year.
The first stage involved only the opening steps in Darwin's overall argumentation from individual generation to species formation.
For it sees him changing his mind about the adaptiveness of structural change rather than about species formation as such.
This first stage did come on reading Thomas Malthus's (1766  1834) Essay on the Principle of Population, on 28 September 1838; after a generational conjecture  that higher animal foetuses are initially hermaphrodite  had led him to Adolphe Quetelet's findings on the sex ratio at birth and so, it seems, to Malthus as an author linked by a reviewer with the Belgian social statistician (Schweber, 1977).
Now, what Darwin's reflections on Malthus did was to move him away from a prenatal, maturational sorting of adaptive from maladaptive variation.
For he moved at once to a post-natal, ecological sorting wherein individual adaptive variants are retained, while the maladaptive are eliminated in the Malthusian crush of population.
These initial reflections did not include, then, any analogy between artificially and naturally selective breeding.
But Darwin did draw an explicit analogy between this teleology  of population pressure and sorting, as ensuring adaptation of plant and animal structure to changing conditions  and Malthus's theistic teleology of superfecundity, as ensuing the energetic dispersal of ancient tribes beyond the original Asian seat of the human species.
Man, naturally slothful according to Malthus, only spread into more adverse climes thanks to the local scarcities of food entailed by his excess fertility; with later settlers always eventually victorious because made doubly energetic in struggling with both rigorous conditions and previous settlers.
Alien species beating natives on their home ground had been decisive for Lyell's species extinction theorising, where the analogy with the European human conquests over America Indians was explicit.
Colonizing species of alien genera doing likewise had long been decisive for Darwin's species origin theorizing, where the analogy with the European human conquests over Australasian natives was hardly less explicit.
Throughout 1838, Darwin had been allowing for Lyellian competitive defeats to extinguish some species before their predetermined ageing overcame them.
Now, his Malthusian reflections prompted both a new ecological understanding of adaptive sorting and a final return to Lyellian ecology for all extinctions.
Thus did ecological explanations regain ground earlier lost to generational ones.
The second stage began around the end of November 1838.
For Darwin then drew his first explicit analogies between adaptation in wild species and the fitness for human ends of domestic races, as both due to selective breeding.
To make this new comparison Darwin had to drop his old contrast between monstrous, selected, domestic races and adaptive, not selected, wild species.
He dropped it first, it seems, on considering sporting-dog breeds, notably greyhounds; for these, although formed by the human artifice of selective breeding, had been given both structures and instincts useful in the wild.
(Hence the carnivorous canines In Darwin's later thought-experiments on natural selection.)
Up to now, domestic race formation had provided Darwin only with analogies for the formation of wild species as ancient, true breeding and intersterile races.
With this new selective breeding analogy, domestic races, as adaptations, are also providing analogies for the formation of wild species as ancient and perfectly adapted races.
The third stage came seemingly early in 1839, when this new analogy prompted a further revision to the opening steps of the overall argumentation.
In artificial selection, the chosen end, for which the race is being fitted, does not elicit the variation being selected.
So, Darwin reasoned, natural selection likewise could work with variation that is accidentally rather than necessarily adaptive.
If changes in conditions disrupt the precise replication of parental characters so as to yield hereditary variation, then, providing only that some of it happens to be adaptive, this will suffice in the long run for selection as a cause of adaptive species formations.
With these developments beyond his mid-September 1838 positions, Darwin had reached the theory of natural selection much as he would publish it later.
In March 1839, he even outlined the argumentative structure that opens his Sketch of 1842 (F. Darwin, 1909), the first, manuscript, version of The Origin of Species (1859).
By 1841, he had very probably worked out, also, his later theory of individual organism generation: pangenesis.
The theory, as eventually set forth (1868), was constituted by two theses: that the generative material comes from all over the parent body or bodies, and that it consists of minute ' gemmules' budded off from every part.
Darwin's preoccupation with generation went back to Edinburgh and to consequent inquiry, when on the voyage, into modes of reproduction common to various invertebrate groups (Sloan, personal communication).
In early 1837, he had supported the senescence analogy between sexual and asexual generation by interpreting all generation as division, whether artificial or natural, complete or incomplete, simultaneous or successive.
This thesis, that all generation is division, continued throughout the B, C and D notebooks and was reinforced by an explicit equating, in September 1838, of division and gemmation or budding.
The equation made all generation, sexual or otherwise, a form of budding.
However, in his theorizing about sexual generation, as being unique in allowing variation and adaptation, Darwin made a fundamental distinction between sexual buds, such as ova, and asexual buds.
According to this distinction, all and only sexual buds are involved in maturation and fertilization.
Consequently all and only sexual buds are impressionable, whether by the action of changed conditions or by a sexual element from a mate of unlike constitution.
Conversely, while lacking those powers an asexual bud, or even a severed flatworm fragment, can do what an unfertilized ovum can not: namely, produce a whole organism without interactive collaboration with any other part; this power being credited by Darwin to the presence in such a bud or fragment, indeed in any healing flesh, of material determining growth for all the parts of the whole organism.
Switching now from 1838 to 1868, one sees, in direct contrast with this earlier view, that a principal object of pangenesis is to explain how in all generation, sexual and asexual, the powers are the same and so, too, the material.
To this end, Darwin adduces, most especially, those phenomena that would be anomalous for any exclusive correlation of maturation, fertilization and impressionability with sexual rather than asexual modes of generation.
Thus aphid parthenogenesis shows an unfertilized ovum producing a maturing organism with no prior interaction with a male element; again, so-called graft hybrids and the effects of pollen on non-germinal tissue in a female plant both show impressionability without fertilization and maturation; while sporting and reversion in asexual plant buds show variation without fertilization or maturation.
So, as contrasted with his September 1838 position, Darwin sees pangenesis as an evening up, on both sides, of all the powers of the sexual and asexual parts of any organism.
Pangenesis is itself presented as a theory of how this identity of powers arises in development.
In a developing organism, starting as a fertilized egg, each cell or tissue, throughout its own maturation, is budding off miniature facsimiles of itself, the ' gemmules'.
The whole organism can then have, as an adult, the same powers in its asexual buds and its sexual organs, because the same material is there: namely, ' gemmules' collected from all over the body.
These gemmules have, accordingly, been invested by Darwin with two sorts of properties: those credited to every asexual part of the body, in 1838, to explain its generative and regenerative powers, and those invoked then to explain the impressionability and variability of immature ova.
So, pangenesis could have been derived from the 1838 position, by pandynamic extension to the ova of powers previously denied to them, and by a panovulational extension to all other parts of powers and matters formerly reserved for the ovary.
But is that, in fact, how Darwin arrived at pangenesis?
No known document confirms that it was.
But all the indirect evidence, including the records of his reading in such writers on generation as Erasmus Darwin, Johannes Mller and Giorgio Gallesio, makes it most probable that he came to pangenesis in such a revision of his 1838 position, and that he did so in the years 1840C1.
However and whenever it was first formulated, this pangenetic reduction of every mode of generation to micro-ovulo-gemmation could take inheritance, in so far as it was completely conservative, to be effected by an exact replication of a whole in all its parts; so that variation, reversion and so on are explicable as disturbances, suspensions and complications of that fundamental replicative tendency.
Thus could pangenesis unify all Darwin's generation theorizing from gemmules to organisms and on to species, and beyond them the whole tree of life.
Pangenesis was never, however, to meet the vera causa evidential ideal, as Darwin himself was keenly aware, and he published it only after natural selection had been launched unencumbered by any such conjectural causation for generation.
The argument of the 1842 Sketch (and so, too, of the Origin) was knowingly structured to accord with that vera causa ideal (Hodge, 1977).
The existence of natural selection as a causal agency is evidenced first; then its adequacy to produce new species from old is argued by analogy from the ability of artificial selection, although much less precise and prolonged, to produce domestic races.
Finally, Darwin displays the indirect evidence for the theory: the explanations it provides for a wide array of facts in biogeography, geology, embryology and so on.
With the 1842 Sketch written, Darwin's most creative period was ended.
That year he moved out of London to the Kent countryside and was henceforth mainly writing books, raising children and nursing his health.
He never had another fundamentally novel idea in general biological theory.
But then he had already had enough to keep him and many others occupied for a very long time.
The nature of Darwin's science
Even this brief analysis of Darwin's biological theorizing in these early years suggests the following conclusions about the nature of his science in its formative phase.
The social context.
The traditional way to connect Darwin's science with his society is through Malthus's laissez faire political economy.
But this can not be adequate.
A capitalism connection is relevant in the 1840s and 1850s when Darwin is applying division of labour theory to the problems of divergence (Schweber, 1980; Ospovat, 1981).
But in 1838 it was Malthus's theodicy of ancient empires, not his political economy of the modern state that bore decisively on Darwin's biogeography and ecology (Bowler, 1976).
More generally, any laissez faire connection helps very little in understanding how Darwin came to take up the problems his theorizing was to solve.
Here it is his family, especially his grandfather, and his mentor Lyell, that indicate connections with movements of thought directly linked with fundamental social change.
Erasmus Darwin, a Birmingham Lunar Society man, belonged to a provincial movement of ' radical ' dissent from national and metropolitan orthodoxies in politics and religion.
Lyell was a ' liberal ' Scots Whig very much in the Edinburgh Review tradition of John Playfair and his own father-in-law Leonard Horner, a tradition bent on using electoral and educational, including university, reforms to break the national and metropolitan hegemony of the Tory, Oxonian, Anglican establishment.
It is in such mediating contexts, rather than in any direct tie to capital through laissez faire, that the social history of Darwin's science should be sought.
Laws, causes and chances.
To Darwin, natural selection, as a causal agency and lawful process, was akin, in its vera causa credentials, to gravitational force in celestial mechanics.
But his earliest critics often judged his theorizing not to match the standards set by Newtonian physics (Hull, 1973).
Was Darwin, then, mistaken in so relating his own science to Newton 's?
He was, in so far as he underestimated the implications of one major disanalogy: he had no law that was to natural selection as the Newtonian inverse square law (with proportionality to mass products) was to gravitational attraction.
For this disanalogy arose from another: Darwin's natural selection as a lawn process was complex, being compounded from heredity, variation and superfecundity, each of those processes having its own laws; while Newton's gravitational force was not compound and had a single law of its own.
Now, from these disanalogies arose a further one.
In simple cases, the consequences in certain conditions of Newtonian gravitational attraction could be deduced and the adequacy of this cause for certain phenomena, most notably elliptical planetary orbits, thereby established.
By contrast, the consequences and so competences, and hence the adequacies or inadequacies, of natural selection, especially for long-run effects, were practically impossible to decide, at least for a finite intellect; although the young Darwin himself could consistently suppose that God in choosing this means for adapting life to a changing earth had foreseen all its consequences.
However, although these disanalogies were fundamental, Darwin's theorizing had not taken him out of the causal, lawful, deterministic Newtonian universe, into one as irreducibly acausal and absolutely probabilistic as is sometimes thought implicit in quantum mechanics.
Any natural selection involves differential reproductive success that is nonfortuitous because determined by the way variant organisms are interacting with an environment that is causally sensitive to those particular physical differences in the organisms.
If more red than green members of a species of moth living on green foliage are being killed by predators, that may be selection; but only if the predators are not colour-blind; if they are it is random sampling error.
But even when killed by colour-blind predators the moths are not victims of any capricious cosmic indeterminacy; for each of their deaths is a causally determined event.
It is only as deaths of red or green moths in that environment that these are fortuitous events; this fortuitousness being, as philosophers say, description relative.
Again, as Darwin saw it, the chanciness of hereditary variation is relative, not absolute.
The processes generating this variation he supposed to be lawful, causal and so determinate; but not causally sensitive to environmental conditions in such a way that any particular change in conditions elicits an increased supply of those variants that are adaptive in the new conditions.
So, the distinctions decisive for Darwin's account of the generation and the fate of variation are distinctions drawn within the presuppositions of a deterministic universe.
In its dependence on those presuppositions his biology was more like statistical than either celestial or quantum mechanics (Hull, 1974).
The non-tautologousness of natural selection.
Darwin's vera causa argumentation shows that natural selection is not tautologous for one reason, ultimately.
The definitional question, of what natural selection is, can be answered by specifying necessary and sufficient conditions for its occurrence, namely hereditary variation that is causally relevant to reproductive success thanks to organism-environment interactions; it can, then, be answered without begging in advance of empirical inquiry all those further questions as to whether these conditions are ever met: whether, that is, any natural selection exists; and, if so, how it is distributed, what it can do now and what it has been responsible for in the past.
It is, then, a mistake to defend natural selection against the tautology objection by proposing criteria of fitness independent of reproductive success.
Fitness differences are best understood as reproductive expectancy differences analogous to normalized life expectancy differences.
And as such they have no causal or explanatory power of their own.
If Jones has outlived Smith this can not be explained by showing that he earlier had the higher life expectancy, and then arguing that this duly caused him to live the longer life.
It can be explained by citing earlier physical differences; perhaps Jones jogged while Smith smoked.
Likewise, in natural selection it is physical differences, not the differences in reproductive expectancies estimated from them, that can cause and can explain subsequent reproductive performance differences.
As differing expectancies, fitnesses can neither cause reproductive performance differences nor be definitionally equated with them.
So, natural selection is no untestable tautology, but not because fitness measurements, as expectancy estimates, can sometimes be falsified by later performance measurements.
Natural selection is no tautology, because there is no a priori proof, from its definition alone, for its existence, nor then for its prevalence, or its adequacy or its responsibility for evolution.
If there were, then the last decade and a half of selectionist-neutralist controversies over all these different non-definitional issues could have been settled in advance without recourse to empirical data, in an armchair with a scientist's glossary and a logician's truth table.
As Michael Ruse observes (1981) if selectionism is a tautology, neutralism is a contradiction, Darwin's strategy in structuring his argumentation to conform to the vera causa ideal shows why it is not.
Evolution, cytology, genetics and the unification of nineteenth-century biological theory.
Early in 1900, the doyen of Columbia University biology, E. B. Wilson, introduced the new second edition of his treatise, The Cell, by arguing that the main challenge then facing biological science was to integrate its two greatest achievements in the past century: the theory of evolution and the theory of cells.
And he explained why he saw August Weismann's theory of the continuity of the germ plasm as the most promising foundation for any such integration.
Wilson's position makes sense of a great deal in the history of general biological theory before and since 1900.
It has often been said that the decisive development since Darwin was a new synthesis, in the 1920s and 1930s, of Mendel on heredity and Darwin on selection.
Wilson brings out the importance of that earlier and no less fundamental post-Darwinian synthesis, which this later one presupposed: the synthesis of evolution and cytology.
Since Wilson's teachings were a principal inspiration for the Morgan school, the twentieth century science of genetics may be said to have arisen within his Weismannist programme for the integration of evolutionary and cytological theorizing (Mayr, 1982).
Darwin's integration of evolutionary and physiological biology had been attempted, in the 1840s, through pangenesis.
But pangenesis was not reconcilable with the cytological consensus just emerging when Darwin published it a quarter of a century later.
Physiologists were then increasingly agreeing that every cellular organism is either a single cell or a cell colony arising from the successive divisions of a single cell, and that two cells come together to form one at fertilization, each having arisen by the division of one cell in the respective parent body.
In Darwin's version pangenesis could not be squared with these cytological generalizations; for, if each of the two masses of gemmules coming together at fertilization is taken to be one cell, then it has not arisen in the division of one cell in that parent; while, if each is taken to be a myriad of cells, then far too many are coming together at fertilization.
The problem of transforming Darwin's pangenetic theory to square it with cytology in general, and with Weismann's theory in particular, was taken up most systematically by De Vries; and later developments leading to the theory of the gene were to owe much to his solution: ' intracellular pangenesis'.
So, Darwin's attempt to integrate evolutionary and physiological biology contributed indirectly to the unification that was called for in Wilson's 1900 programme and that genetics was to secure.
That Darwin's ideas could have such manifold influence throughout the entire structure of modern biological theory should not now be surprising.
Compulsive, selfconscious intellectual, ' philosopher ' no less than ' naturalist ', he had worked from his earliest years on a very broad canvas indeed.
A DARWINIAN PLANT ECOLOGY John L. HARPER
'... on a piece of ground 3 feet long and 2 feet wide, dug and cleared, and where there could be no choking from other plants, I marked all the seedlings of our native weeds as they came up, and out of 357 no less than 295 were destroyed, chiefly by slugs and insects.
If turf which has long been mown, and the case would be the same with turf closely browsed by quadrupeds, be let to grow, the more vigorous plants gradually kill the less vigorous, though fully grown plants; thus out of 20 species growing on a little plot of mown turf (3 feet by 4 feet) 9 species perished, from the other species being allowed to grow up freely ' (The Origin of Species).
It is difficult to detect any direct influence of Darwin's writings on the development of the main stream of plant ecology.
The extreme reductionist approach that is represented in the above quotation, and is apparent again and again in his writings, not only in The Origin of Species but in many of his later books, is conspicuous by its absence from early plant ecological texts and is barely represented in the ecological literature until towards the middle of the twentieth century.
The approach that involved marking individual plants or seedlings in the field, tracing the fate of individual leaves as they are pulled down earthworm burrows, the behaviour of tendrils as they touch a support, the fate of insects as they land on a Drosera leaf, or recording the number of seeds at the bottom of an earthworm burrow, represented a reductionist level of concentrated observation that contrasted with the geographical view of vegetation with which Warming and others set the early direction of plant ecology.
There is no way in which Darwin can be regarded as a parent of the science of plant ecology.
Nothing illustrates this point so clearly as the fate of a paper published in 1874 by C. N?geli entitled Verdr?ngung der Pflanzenformen durch ihre Mitbewerber.
This paper was directly stimulated by The Origin of Species and in it N?geli developed mathematical models that describe the interaction between populations of two species of plants.
His models attempt to describe in formal mathematical terms the replacement of a population of one species by another and also situations in which pairs of species persist together as stable, mixed populations.
N?geli's models included density-dependent and frequency-dependent situations and came very close to providing a formal description of the niche.
It might have been expected that such a paper from N?geli, who was one of the most distinguished botanists in Europe at that time, would have had immense impact on the early development of plant ecology.
Instead, it appears to have been wholly ignored for 60 years until it was mentioned briefly by Gauze (1934) in his book The Struggle for Existence.
Plant ecology developed not as a study of the factors affecting the lives and deaths of individual plants and their parts but as a study of the distribution of vegetation types and of particular species.
It included also the description of those specialized features of morphology and physiology that distinguish species and might (often by more or less inspired guesswork) be said to account for the differences in their distribution.
That these features were often called ' adaptations' did nothing to explain them.
Much of the early science of plant ecology sought for correlations between vegetation and physical, not biotic, factors in the environment, most particularly temperature, water supply and soil types.
These forces have been described as' Wallacian ' (Harper, 1977), because they represent those agents of natural selection that were of more concern to Wallace than to Darwin in accounting for how organisms are as they are and behave as they do.
The role of biotic (Darwinian) forces in determining the distribution and abundance of species was largely neglected except by token reference to grazing animals.
The essentially Darwinian forces of struggle for existence, involving competitive interactions between members of the same species and between different species, played a negligible part in the interpretation of natural vegetation.
The Darwinian approach, involving reductionist concentration on individual plants and the hazards that they experience, particularly the interference from their neighbours, entered the science of plant ecology in the late nineteen twenties in simple experiments involving mixed populations of two or more species and I have described elsewhere (Harper, 1967) the curious piece of history in which three leading ecologists, Sukatschev in Russia, Clements in the United States and Tansley in Britain, all made simple competition experiments involving deliberately-sown plant populations.
None of these authors continued with this type of study and all returned to essentially descriptive studies of vegetation.
It was as if they had found it too difficult to bridge the gap between simple experimental systems and the complexity of nature  as if the reductionism of the experimental method lost the holist qualities of the integrated complex whole that these distinguished ecologists saw in natural vegetation.
Both Clements' Community as an Organism and Tansley's Ecosystem can be represented as a retreat to community holism after a brief flirtation with an organismal, Darwinian ecology.
A piece of ecological history that remains to be fully researched was the decision by a number of individuals, many apparently working in isolation from each other, to establish, like Darwin, permanent plots within which the fate of individual plants could be recorded over time.
The great classic amongst such observations is the work of Carl Olaf Tamm (1948, 19721 who marked out permanent quadrats in woodland and grassland near his country home and as a hobby mapped, remapped and continues to map the plants within them.
Tamm attributes his decision to embark on this long-term programme to stimulus from Romell, a distinguished Swedish ecologist and soil scientist.
Other permanent quadrats were set up by Forrest Shreve (1915) at the Desert Laboratory of the Carnegie Institute of Washington at Tucson, Arizona, and it appears to be through a colleague of Shreve, W. A. Cannon, that T. G. B. Osborn was stimulated in 1926 to set up permanent quadrats in heavily used shrub land and a reserve released from grazing at Koonamore in S. Australia (Osborn, Wood &amp; Partridge, 1935; O. B. Williams &amp; Mott, 1981).
There is also a European tradition that precedes the work of Tamm, associated with the name of Bogdanovskaya-Gienef (1926) who seems to have been a pupil of Sukatschev and to have influenced the reductionist approach to ecology later developed by Rabotnov, Uranov and their pupils in Russia.
She reported her two-year study of four 20  20 cm quadrats.
Her publications list appears to consist of only two papers, but she may have had a greater influence than this suggests (J. White, personal communication).
The peculiar quality of Tamm's work was that he recorded individual plants, shoots or rosettes, within his populations and this enabled him to follow the fates of individual plant units (often tillers or ramets) rather than to study the grosser vegetational change that was the aim of many others who set up permanent quadrats.
The data obtained from Tamm's studies revealed, for the first time, the magnitude of the flux that underlay, the apparent stability of many plant communities and (after Darwin) gave the first real insight into plant community dynamics dominated by establishment and deaths.
Vegetation is composed of the few plants that survive and grow: to explain that vegetation it may be more important to study the many that die.
In many of Tamm's populations it can be shown (Harper, 1967) that the plants originally present were progressively lost and that the rate of loss was remarkably constant.
Half-lives could be calculated for the populations  differing from species to species but constant over the years.
Such data made it clear that (i) it was realistic and profitable to study individual species, even within a complex community and (ii) that the dynamics of the populations appeared to be largely independent of year-to-year fluctuations in climate.
Later studies of permanent quadrats in grassland by Sagar, Sarukh?n, Hawthorn and others (see Harper, 1977) extended over fewer years but involved repeated observations within each year.
They showed that the death risks to plants in populations within years were commonly greatest when the survivors were growing fastest  not during the periods when the anthropomorphically-minded botanist would regard the physical environment as' harsh '.
The likely explanation is that it is biotic pressure from competing neighbours rather than ' harshness' of the physical environment that is the prime cause of death of most plants and perhaps of natural selection.
Most of these detailed demographic studies have been made with pasture or woodland systems in northern temperate regions and it could well be that in arid zones, and some other extreme environments, biotic pressures are less dominant and then climatic factors may play the major role in killing plants and in natural selection.
At a Symposium organized to pay tribute to Darwin a hundred years after his death, it is perhaps permitted to expand on one tiny fragment of ecological research that may truly be said to have been directly stimulated by Darwin's own writings, in particular by the paragraph quoted as heading to this paper.
Early in the 1960s the decision was made to concentrate a number of intensive ecological studies based at Bangor, North Wales, on a single, small (1 ha) field of permanent grassland, part of the College Farm at Henfaes, Aber, near Bangor.
The field was chosen because it was superficially very dull  lacking any obvious heterogeneity of contour and relatively homogeneous in soil properties.
The field had not been ploughed for at least 80 years (probably not for more than 150 years) and there is no record of fertilizer or herbicide ever having been applied.
A general description of the field is given in Turkington &amp; Harper (1979a) and more detailed analysis of the soil and flora in Turkington (1975).
Many graduate students and overseas visitors have worked on aspects of the ecology of this field.
In many cases the studies have concentrated on populations of a single species within 1 m&amp;sup2; quadrats, though sometimes the reductionist level of study has been yet smaller  a scale of 1 cm&amp;sup2; (Thorhallsdottir, 1983).
The concentration of effort within one small field, and on small quadrats within it, gives the studies high precision and high relevance but with an absolute sacrifice of generality.
We do not know whether most of what we have observed in this field can be generalized to other fields or, indeed, to less intensively studied parts of the same field.
The vegetation of the field was analysed by ordination and correlation techniques which showed that only a minor part of the variation in species distribution could be accounted for by underlying edaphic factors, though in the peripheral areas of the pasture the presence of hedges and trees accounted for significant changes in the vegetation  e.g. Dadtylis glomerata occurred mainly in or close to the shade of the trees.
Two common species, Lolium perenne and Trifolium repens were usually positively associated in their microdistribution and negatively associated with other species.
The structure of the vegetation was interpreted as determined by regeneration cycles directed by T. repens and L. perenne  the species themselves appeared to be the prime determinants of each others distribution (Turkington and Harper, 1979a, b).
The most intensive studies on the field have been made on populations of three species of Ranunculus and on Trifolium repens.
Some of the work is still in midstream and consequently some material referred to in this paper represents an interim report.
Two of the species present in the field are of particular interest, both to population biologists interested in the manner in which the numbers of plants are regulated and to evolutionists concerned with the extent and significance of natural variation.
The two species are Ranunculus repens and Trifolium repens.
Both possess the property of clonal growth by which the product of a single zygote forms a spreading clone of rooted nodes capable of vegetative extension through the sward.
Such plants have the potential for a single clone or genet (product of a single zygote) to dominate large areas of the vegetation.
Indeed, it would be theoretically possible for one genet of Trifolium repens or Ranunculus repens to have occupied the whole of the Henfaes field.
In other plants with such clonal growth it is known that the product of a single zygote may indeed occupy considerable areas of land.
An extreme example is bracken, Pteridium aquilinum; Oinonen (1967) showed that individual clones in Finland were up to 1440 years old and one old clone extended over an area of 474  292 m.
There are other cases in which clone-forming species are known to produce large areas of genetic monotony, e.g. Festuca rubra (Harberd, 1961) and Holcus mollis (Harberd 1967).
It might be expected that where such clonal growth is possible, the struggle for existence over long periods of stable management would lead to the local dominance of single clones  those that had succeeded in a struggle for existence with others.
The population dynamics and genetics of these two species in the permanent pasture at Henfaes seemed to over the opportunity to study natural selection in action  Hutchinson's ' ecological theater and evolutionary play '.
Population dynamics of Ranunculus repens
The detailed demographic studies of Ranunculus repens made by Sarukhn in the field at Henfaes and followed by unpublished studies by Soane made it possible to give quantitative measures to the population dynamics of this species.
These are summarized in the life-cycle diagram of Fig. 16.1.
Populations of this species showed vigorous vegetative growth, and most new rosettes recruited to the population appeared as ramets (clonal replicates) from existing rosettes.
The input of new seedlings was small and the number eventually contributing to mature rosettes in the pasture seemed insignificant in comparison with the contribution from clonal growth.
In comparison with the other two species of Ranunculus, R. acris and R. bulbosus present in the pasture, seed production and the numbers of seedlings observed was very small.
Plants of Ranunculus repens produced on average less than one seed per rosette (cf. ca 10 for R. acris, ca.
15 for R. bulbosus).
In the studied quadrats, 25 seedlings of Ranunculus repens emerged per metre square from the long-lived bank of seeds in the soil, in contrast with 176 of Ranunculus acris and 95 of Ranunculus bulbosus.
The dynamics of Ranunculus repens populations was followed over four years during which new seedlings that died and those that survived to form rosettes were recorded.
The development of clones from the rosettes present at the beginning of the study was also recorded in detail so that the clonal parent was known for almost every rosette present in the quadrats at the end of the study.
These data allowed Soane &amp; Watkinson (1979) to build a computer simulation model to examine the relationship between the flux of ramets, the recruitment of seedlings and the diversity of genets within the populations.
The computer models simulated the actual flux of ramets in each of eight studied populations and followed the fates of ramets and families of ramets assuming no selection between families.
The real and simulated changes in the populations are shown in Fig. 16.2.
Agreement is extremely close and provides little evidence for selection between families or against new seedling recruits.
In the simulation model, in the absence of selection, the number of original families present in the population declined at an approximately exponential rate.
With the passage of time the contribution that these families made to the total genetic diversity of the population became subordinate to the seedling recruits: although the number of seedlings appears to be very small, their contribution to the total number of genetic individuals in the population is clearly significant in determining the number of clones or genets that are present.
In the absence of selection, the observed small numbers of seedling recruits would apparently be sufficient to maintain potentially high genet diversity within such a vigorously clonal plant population.
Such genetic diversity was indeed present, because there was visible genetic polymorphism within the populations and it was shown (M. J. Lawrence, personal communication) that there was considerable genetic variability both in quantitative characters and in polymorphism at two enzyme loci in populations immediately adjacent to the permanent quadrats.
Population dynamics and genetic variation in Trifolium repens
Analysis of the population dynamics and associated genetics of Trifolium repens in the same pasture is even more revealing.
Trifolium repens is unusual in that much of the genetic variation present within the populations is easily recognised in phenotypic differences that are visible or easily determined on plants in the field.
Leaf mark polymorphism is one such property.
A variety of white leaf marks is found in natural populations and these are represented by multiple alleles at a single locus.
In mid-summer, when the marks are most fully expressed, the genotypes of most plants can be identified in the field and an estimate of the number of clones present within an area can be made: it will, of course, be a minimal estimate.
Cahn &amp; Harper (1976a) determined the number of clones present within 10  10 cm quadrats and found, to considerable surprise, that between 3 and 4 clones per quadrat was the most common situation in the field.
This level of genetic diversity at a fine scale was confirmed in other permanent grasslands in Britain including Port Meadow, Oxford, which is at least 896 years old!
Clearly, single clones did not dominate patches even at this very fine scale.
More recently, Trathan has identified genetic individuals with more precision using isoenzyme analysis.
He finds 48C50 distinct genotypes present per metre square.
The various clones weave amongst and intermingle with each other and amongst grasses and associated herbs.
The degree of intermingling may itself reflect the growth form of stoloniferous species.
Both Ranunculus repens and Trifolium repens have ' guerrilla ' growth forms in contrast to the predominantly ' phalanx ' forms of the associated grass species.
By ' guerrilla ' growth form is implied one that is continually wandering amidst associated vegetation, creeping into new and escaping from old patches in the community.
In contrast, ' phalanx ' growth forms develop a structure of tightly packed shoots (most of the pasture grasses and some pasture dicots such as Bellis perennis) (Lovett Doust, 1981).
Not only has Trifolium repens a guerrilla growth form but its guerrilla character is exaggerated when it is growing with grasses; its branching then tends to be reduced and growth is concentrated in linear extension.
Instead of a genet locally consolidating its occupancy of a site, individual stolons wander as linear extensions of the genet into surrounding vegetation.
This growth form itself maximises the chance that genets will intermingle: it maximizes the role of interspecific contacts in the life of the genet.
Darwin commented on the growth of such plants and the ways in which they penetrate amongst other vegetation.
He describes his own (very Darwinian!) experiment in which he allowed the stolons of Saxifraga sarmentosa (a classic ' guerrilla ' growth form) to encounter an artificial vegetation that he had constructed: ' Many long pins were next driven rather close together into the sand, so as to form a crowd in front of... two thin lateral branches; but these easily wound their way through the crowd.
A thick stolon was much delayed in its passage; at one place it was forced to turn at right angles to its former course; at another place it could not pass through the pins, and the hinder part became bowed; it then curved upwards and passed through an opening between the upper part of some pins which happen to diverge; it then descended and finally emerged through the crowd ' (Darwin, 1880).
Another approach to the study of variation within populations of white clover in the Henfaes field was made by Burdon (1980).
He sampled 50 white clover clones from a grid covering the whole field.
He multiplied the clones in the glasshouse and screened them for a variety of characters (Fig. 16.3).
These included a number with simple Mendelian inheritance and a number of characters of agronomic importance with polygenically controlled expression.
He was able to use these characters to produce identity diagrams that distinguished and ' finger-printed ' each clone.
The 50 clones differed on average from one another in 3. 3 vegetative characters.
If floral characters were included in the comparison the average difference between clones was 5.4 characters.
(One pair of clones differed in 13 statistically significant and apparently independent respects!)
Many of the characters considered had been shown by other workers to be of selective importance in white clover or another species of Trifolium (e.g. Cahn &amp; Harper (1976b) had presented evidence suggesting that sheep selected between leaf marks; Dirzo &amp; Harper (1982a) and others have shown that slugs select between cyanogenic and acyanogenic forms; Black (1960) had shown the selective value of long petioles).
The polymorphisms in the Henfaes population were not exhausted in Burdon's study.
It has now been shown that variation in relative growth rates can be added to the list (Burdon &amp; Harper, 1980).
Natural populations of white clover are polymorphic for incompatibility alleles and for the ability to form nitrogen fixing symbioses with strains of Rhizobium but these have not been looked at at Henfaes.
Berrington (personal communication) has recently shown that clones within the field differ in their ability to form endotrophic mycorrhizal associations.
The polymorphism that is found within white clover populations in the field represents therefore a subtle (or? random) variety of unique associations of apparently selectively important properties.
The imagination of the most extreme selectionist is stretched to breaking point by such a situation.
Evidence for local specialization within white clover
Plants that are capable of clonal growth offer peculiar opportunities for testing the extent to which particular genets are locally specialized.
A plant may be sampled from the field, multiplied clonally and the clonal products (genetic identities) can then be reinserted into the field, both in the places from which the clones originally came and into other places.
Plants of the same genotype can then be tested in different environments.
Turkington made such an experiment with white clover in the field at Henfaes (Turkington &amp; Harper, 1979b).
Clones of white clover were sampled from within patches in the field dominated by each of four common grasses, Lolium perenne, Holcus lanatus, Agrostis tenuis and Cynosurus cristatus.
The clones were multiplied in the glasshouse and then transplanted back into patches of the field dominated by the four grass species.
The performance of the transplants was measured by vegetative growth expressed as dry weight at harvest after twelve months.
The results are shown in Fig. 16.4.
Over the whole experiment clones of clover that were returned to their original grass associate made more growth than those introduced to alien sites (significant at P &lt; 0.001).
The clover clones had also been introduced into sites from which the existing vegetation had been denuded by treatment with the paraquat herbicide.
Some of the ' principal diagonal effect ' remained, though the difference in yield between clones returned to their site and those returned to alien sites was now significant only at P &lt; 0.05.
Turkington made a further experiment in which he introduced the four clone types of white clover into pure swards of the four grass species that had been sown on soil sampled from the experimental field.
This part of the experiment was designed to remove possible place-to-place variations in soil conditions in the studied field from the comparison.
The yield of clones grown with the grass species from which they had originally been sampled again exceeded that made when they were grown in an ' alien ' sward, P &lt; 0.00001.
It is difficult to interpret the results of this experiment as representing anything but evidence of precise specialization of clover clones in their ability to grow in association with particular grass neighbours.
It suggests that, within the pasture, strains of white clover have been selected by competitive interaction with associated grasses and that different species of grass contribute to the diversifying or disruptive selection operating upon the population of white clover.
Some, at least, of the variation within the white clover populations thus appears to be directly interpretable in terms of attributes contributing to present fitness.
It is a very Darwinian interpretation to suggest that the grass neighbours may be primary forces selecting and diversifying the clover populations.
Hill (1976) grew a single clone of white clover with a variety of clones of Lolium perenne.
Quite distinct phenotypic modifications were elicited from the clone by the different ryegrass strains.
If different strains of Lolium perenne produce different phenotypes from the same clone of white clover, it is difficult to escape the conclusion that different species of grass are even more likely to exert different selective pressures within populations of white clover.
It is not easy to measure and describe just how the different clones of white clover differ in their reaction to different neighbouring grass species or forms.
Survivorship and dry matter production are very gross measures of a plant's reaction to different types of neighbour.
We have, at present, no real indication of the manner in which ecological compatibility (ecological combining ability) between particular strains of white clover and particular pasture grasses is accomplished.
It may represent subtle differences in growth cycle or growth form or more complex interactions involving the soil microflora, perhaps the mycorrhizae.
There may be subtleties of interaction below the soil surface of which we know little or nothing.
The hazards in the life of a plant in the field are not only those of competition from neighbours, though it may be these that are the most relentless.
Hazards in the life of a plant in the field
Some of the hazards to the life of a plant in the field can be measured by studying the fate of individual leaves or flowers or seeds.
An attempt to catalogue and quantify these hazards in the field at Henfaes was made by Peters (1980) who included white clover amongst the species that he studied.
He marked young leaves as they began to expand and then followed their fate by repeated observation.
From this he could obtain survivorship curves for cohorts of leaves born in the same time period and record some of the causes of death or damage within the populations (Fig. 16.5).
Some leaf predators leave tell-tale records of their activity.
In particular, grazing molluscs leave characteristic erosions from a leaf edge; birds, particularly the wood-pigeon, feed on clover and often leave characteristic beak-marks; weevils remove circles of tissue, often leaving the upper epidermis intact; sheep (causing damage probably indistinguishable from that caused by rabbits) remove whole leaves, leaving torn petioles or leave their bite marks on the leaflets that remain.
Tracking the fate of individual leaves immediately reveals a number of other hazards.
Leaves may be submerged under a dropping of dung, soused in a downpour of urine, trodden on, pulled into the ground by earthworms (further shades of Darwin) or damaged by frost.
The frequency of these various events in, or ending, the life of a leaf are shown in Table 16.1.
Grazing animals are, however, the major hazards for a leaf though all leaves in a pasture are not equally at risk.
Weevils bite holes in the leaves of white clover and of Ranunculus species but the relative severity of attacks on Trifolium and Ranunculus change through the season, Trifolium being more attacked in early, and Ranunculus in late, summer.
Fig. 16.6 shows the relative proportion of the leaf population that suffered damage from leaf grazers in populations of T. repens and R. repens.
Surprisingly, the leaves of white clover suffered proportionately much less from grazing by sheep than did those of R. repens, but leaves of T. repens suffered much more from both slugs and weevils (except towards the end of the growing season).
Grazing by molluscs figured so strongly among the hazards to a clover leaf that it seems reasonable to expect that, in those years when slugs or snails were abundant, they may act as important selective forces within clover populations.
White clover is polymorphic for the presence or absence of cyanogenic glucosinolates and is also polymorphic for the beta glucosidases that release HCN from glucosinolates when a leaf is damaged.
There is considerable variation in the extent to which cyanogenic properties are expressed and there is some seasonal and perhaps other causes of variation in expression.
Nevertheless, it is possible to classify plants in the field by taking leaf samples and performing appropriate tests (see Dirzo &amp; Harper, 1982a) and to categorize the plants into four groups Ac Li, Ac li, ac Li, ac li.
After a period of exposure to grazing one can then test for the frequency of the various types of damage to clover leaves with the different genotypes and afterwards reconfirm the cyanogenic or acyanogenic status of the plant (Fig. 16.7).
There have been many studies of selection by molluscs on cyanogenic and acyanogenic morphs of white clover under laboratory or other controlled conditions.
It seemed that it might be possible to relate the variation in the polymorphism to slug density in the Henfaes field.
The density of slugs was determined at sites arranged on a grid across the field at Henfaes.
Dirzo (1982b) used trend-surface analysis to draw contours of mollusc density (see Fig. 16.8) and added some further information from visual inspection of the field.
Most notably, he added one big and two small islands (shown in Fig. 16.8 as having high and very high slug densities) which were patches of nettle, Urtica dioica.
Slugs tend to concentrate in such refuges, presumably because they give protection from desiccation.
He then determined the cyanogenesis category of the clover plant nearest to each grid intersection.
Table 16.2 shows the distribution of the different glucosinolate morphs of T. repens between areas with different densities of active molluscs.
There is a clear excess over expectation of cyanogenic forms in the areas of high and very high mollusc density and of acyanogenic forms in the areas of very low mollusc density.
It appears that some at least of the variation in the cyanogenesis polymorphism over the field can be explained as local micro-evolution in response to locally patchy selection.
pp. 337C338: FIGURES
Two strands of evidence, that from the reciprocal transplanting of clover into the neighbourhood of different grasses and that from the study of the distribution of cyanogenic properties and the distribution of slugs, allow an easy interpretation in terms of immediate and present selective forces.
How far this interpretation can be extended to the whole gamut of characters for which Burdon had shown the populations to be polymorphic must be very doubtful.
Certainly almost all of the characters that Burdon lists have been shown to be of selective importance either in clover or in some related species, though sometimes under very specialized circumstances.
The scientific method may sometimes mislead.
We commonly test for the selective value of a particular feature by holding background variation of both genotype and environment at a minimum.
We thereby maximise our chance of demonstrating what we are looking for.
The real measure should be whether selection is significant against normal levels of background variation.
This is why it was important to test the effects of cyanogenesis and reaction to neighbours in the field.
It is difficult to believe that any of the characters examined by Burdon could be selectively neutral.
However, the contribution of each property to fitness must vary dramatically from year to year as well as from place to place within the field.
Most winters at Henfaes (only 400 m from the sea) are mild and frost is rare.
Occasionally, as in the winter of 1982, there is severe frost.
Populations of molluscs fluctuate wildly from year to year in North Wales.
In some years spring growth of the sward is vigorous and exceeds the capacity of sheep and the other grazing animals to keep it fully grazed.
In other years, as in the spring of 1982, a protracted spring drought slowed the growth of the sward and it became tightly graze.
A severe drought is not a common feature on the field but when it occurs it may be in any month from April to September.
During the period of our observations we have detected three significant leaf pathogens on white clover in the field, Uromyces trifolii, Cymadothea trifolii and Pseudopeziza trifolii.
It seems unlikely that these three diseases attack with equal intensity in all seasons and in all years.
In a field that is patchy in space and time, be it ever so small, we may expect that the populations of a species such as white clover will, at any time, reflect selective forces from its past.
The genotypic composition of the population may in some cases dimly reflect forces that operated twenty or thirty years ago.
Other selective forces may have operated quite recently and left a strong memory or image in the structure of the population's genetics.
If this is the case, we would expect to find only a few of the many polymorphisms readily interpretable as responsive to present proximal selective forces.
Much of the polymorphism could be transient and, without an even more detailed history of the field, uninterpretable.
It is doubtful whether such an explanation of naturally occurring polymorphism could be tested without long-term, detailed recording, not only of the variety of genetic changes occurring within clover populations, but at the same time of a detailed recording of the known hazards in the life of the clover plant over the seasons and the years.
Conclusions
The studies that I have described, concentrated in the field at Henfaes, are now being extended by deliberate experimentation within the field.
We are destroying the site as a long-term study on a supposedly stable system by introducing a variety of perturbations such as transplant experiments, the creation of islands for invasion and further perturbations are planned.
The study has involved a curious concentration of effort in one very specialized environment.
The type of observations that have been made have been quintessentially Darwinian.
Another great naturalist, Thoreau, has focussed attention at the same scale: ' Nature will bear the closest inspection.
She invites us to lay our eye level with her smallest leaf, and take an insect-view of its plain'.
If we are to see evolutionary processes in action in plant communities and the proximal events determining their character we must focus our attention away from an anthropomorphic scale of acres or square metres and onto a scale appropriate to the organisms with which we are concerned.
The appropriate scale is determined by the organism and not by us.
It will be different for different species.
We ask for a plant's eye view of life and death in a sward and hope ultimately to be able to collect these reductionist observations into statements about the population, the species or even possibly the community.
I doubt if it is possible to hold the view of Margalef (1968) that ' Relevant evidence does not consist of a massive accumulation of trivia ' and reconcile it with his' Ecology... is the study of systems at a level at which individuals or whole organisms may be considered elements of interaction... '.
It was, indeed, from the massive accumulation of trivia and tiny details, that Darwin assembled the evidence for The Origin of Species.
In a volume commemorating Darwin's death, I have tried to show how his way of looking at the behaviour of individual plants in nature can be extended.
A hundred years after his death his approach seems more relevant to botanical studies than it has ever been.
This part of his intellectual legacy has not yet been fully invested.
A part of the legacy, however, ceases to bear interest.
He was writing in the Origin for readers most of whom were steeped in Victorian optimism, religion and the romantic movement.
It was necessary in 1859 to write about the process of evolution as if it produced the best of all possible worlds, a substitute for the finger of the Almighty at work.
If the process of evolution had not been presented in this way (though with careful caveats) it is very questionable whether it could have been accepted so rapidly by Victorian society.
It was then appropriate to show how '... from so simple a beginning endless forms most beautiful and most wonderful have been, and are being evolved '.
But ' beauty ' and ' wonder ' are in the eye of the beholder and that eye has itself evolved.
The teleology of evolution as a goal-seeking activity persists in indefensible form a hundred years later in the writing of biologists.
This particular heritage may be a millstone around the neck of scientific natural history.
Most particularly, it harms biology as a means of teaching science to the young (Harper, 1982).
Looking back at the variety of studies conducted on plants in the little field at Henfaes, I feel little temptation to explain the behaviour of organisms within it as perfectly fitted for adaptive optima in an ideally evolved ecosystem.
Rather, I see the population of plants trapped in narrowly constrained evolved ruts, bearing the battered imprint of recent and not-so-recent selective and other forces.
' I returned, and saw under the sun, that the race is not to the swift, nor the battle to the strong, neither yet bread to the wise, nor yet riches to men of understanding, not yet favour to men of skill; but time and chance happeneth to them all ' (Ecclesiastes).
PLATE TECTONICS AND EVOLUTION A. HALLAM
In his famous concluding paragraph of The Origin of Species Darwin expressed wonder that the diversity of the organic world can have been produced from one or a few ancestors by the operation of several natural laws  ' there is grandeur in this view of life... '.
The preferential replication of genes by means of natural selection may well be a necessary condition for evolution to take place, but it is hardly a sufficient explanation for how the enormous diversity of life in space and time has come about.
That there are, for instance, kangaroos in Australia and lemurs in Madagascar, and that antelopes rather than dinosaurs currently roam the plains of Africa, is among other things consequent upon a whole series of historical contingencies.
The study of palaeontology in conjunction with geology ought to throw light on the interaction of historical events and the evolution of a substantial sample of the organisms that have inhabited this planet.
Darwin viewed biotic interactions  the struggle for existence  as being the major promoter of evolution.
This is clearly indicated in the following passage from The Origin of Species.
' As species are produced and exterminated by slowly acting causes, and not by miraculous acts of creation; and as the most important of all causes of organic change is one which is almost independent of altered and perhaps suddenly altered physical conditions, namely, the mutual relation of organism to organism  the improvement of one organism entailing the improvement or the extermination of others'.
Such a view would imply, for instance, that the mammals progressively outcompeted the dinosaurs in the late Mesozoic to become the dominant terrestrial vertebrates in the early Tertiary.
We have known for some time that this can not have been the case.
For many millions of years, through the Jurassic and Cretaceous periods, primitive mammals coexisted with dinosaurs, but remained low in diversity and small in size.
Not until after the dinosaurs finally became extinct at the end of the Cretaceous did the mammals radiate explosively into a great diversity of forms such as we see today, to occupy an even wider range of ecological niches than those vacated by the dinosaurs.
It is quite likely that the early mammals were nocturnal in habit and thereby avoided direct competition with the smaller dinosaurs.
With hindsight we can envisage them as biding their time, as it were, until their reptilian competitors disappeared.
This pattern of change is by no means exceptional in the fossil record (see, for example, Gould &amp; Calloway, 1980).
Indeed it seems to be rather characteristic, as is the close coincidence in time of episodes of mass extinction and radiation of a wide variety of animal and plant groups, both terrestrial and marine.
There is a clear implication that such significant evolutionary episodes may be the consequence of major physical events in earth history.
Following the earth sciences revolution within the last couple of decades (Hallam, 1973) it is natural to investigate what relationship, if any, exists between major biogeographic, radiation and extinction episodes and plate tectonics, which appear to have controlled first order events in the physical environment for at least as long as a diverse metazoan fauna became established in the Cambrian, nearly six hundred million years ago.
Because of my limited space, I can do no more, of course, than outline some of the more significant features as they are currently understood.
Physical effects of plate movements
The theory of plate tectonics states that the outer layer of the earth, the lithosphere (C 100 km thick) comprises a small number of (relatively) rigid plates which are separated by narrow zones along which most tectonic, seismic and volcanic activity is concentrated.
These plate margins are of three types: (1) divergent, where crustal material moves apart, under the oceans by a process known as sea-floor spreading; (2) convergent, where one plate plunges down into the underlying mantle (also known as subduction zones); (3) transform faults, where one plate slides laterally with respect to its neighbour, crust being here neither created (1) nor destroyed (2) (Fig. 18.1).
The most obvious effect of plate tectonics is that continents can be split and their components driven apart if a divergent plate margin becomes established beneath them, and can be caused to collide with each other along the lines of subduction zones, where mountain belts such as the Himalayas may thereby be generated.
Other mountain belts such as the Andes are also produced by subduction but at the boundary of continent and ocean floor, which have very different geological character.
Plate tectonics is not the same as continental drift.
Continents are carried passively on moving sectors of plates which also embrace ocean floor; they do not ' drift ' across the latter (Fig. 18.2).
Migrating continents have obvious implications for biogeography and evolution, but there are other consequences of plate movements which may alter the physical environment in such a way as to affect the biosphere just as profoundly.
Firstly, it is widely accepted, though admittedly not conclusively established, that major ice ages may result from the siting of large continental masses in the polar regions, because only in such circumstances can extensive ice sheets become established, with significant consequences for world climate (Frakes, 1979).
In Phanerozoic time (the time which has elapsed since the beginning of the Cambrian) there have been three such ice ages, separated by intervals of a few hundred million years when the world enjoyed a more equable climate and lacked extensive polar ice caps.
Such a gross cyclicity is not, of course, to be confused with the much shorter-phase climatic cycles such as within the most recent, Pleistocene, ice age, which appear to be the consequence of an interaction of several astronomical variables (Imbrie &amp; Imbrie, 1979).
The short-term climatic fluctuations of the Pleistocene have resulted in rapid world-wide (or eustatic) falls and rises of sea level as polar ice has alternately frozen and melted.
The stratigraphic record indicates, however, that eustatic changes have also occurred during the long equable periods when the available evidence suggests that polar ice was absent.
These sea-level fluctuations must therefore have been produced by changes in the cubic capacity of the ocean basins.
By plotting the areal distribution of marine strata and making reasonable inferences from facies distributions about the location of former shorelines, it has been established that for long periods of time in the Phanerozoic, especially in the early Palaeozoic and late Cretaceous, the continents have been inundated by up to two thirds of the present area by epicontinental seas.
The present shelf, or neritic zone, is in comparison a mere pericontinental fringe (Hallam, 1981c).
p 370: FIGURE
By far the most effective way of changing the cubic capacity of the ocean basins is to vary the volume of the mid-ocean ridges; increase in volume will cause a displacement of sea water on to the continents and vice versa.
One popular hypothesis has it that variations in rates of sea-floor spreading are the controlling factor (Hays &amp; Pitman, 1973).
Since ocean-floor basalt subsides as it cools while migrating away from the spreading axis, a faster-spreading ridge will be hotter and more buoyant over a larger area and hence will cause more displacement of sea water over the continents.
The rate of eustatic change producible by this process has been estimated to be about three orders of magnitude slower than that produced by the climatically induced fluctuations of the Pleistocene (Pitman, 1978).
Alternatively, eustatic changes in non-glacial periods can be produced by changes in the cumulative length of active spreading ridges, which will obviously relate to changes in plate patterns (Hallam, 1977).
Epicontinental seas have been biologically significant for several reasons.
They were the site of habitation of a very large proportion of all adult aquatic organisms, and, because of their overall shallowness  generally less than (and often much less than) 200 m  quite modest eustatic and other physical changes may have had disproportionately large environmental consequences compared with the open ocean.
Furthermore, extensive spreads of such seas can as effectively isolate pieces of emergent continent as spreading ocean floor, thereby creating barriers to migration of terrestrial organisms, and should also promote equability of the continental climate.
It has recently been proposed that variations in albedo with respect to latitude (with considerable consequences for global climate) are a result of both the changing distribution of continents and sea-level oscillations.
The latter, causing a change in land-sea proportions, is apparently the more important (Barron, Sloan &amp; Harrison, 1980).
The organic response
For the purposes of a general survey, changes in various biological groups through time are most usefully analysed in terms of diversity (or, more strictly, taxonomic richness), and changes in spatial distribution by the amount of provinciality or endemism.
Obviously, significant radiations will appear as marked increases, and extinctions as decreases of diversity, while the degree to which free migration is inhibited by geographic barriers, thereby promoting genetic isolation according to the classic model (Mayr, 1963), should be expressed by the ratio of endemic to pandemic organisms.
Pioneer attempts to relate temporal diversity changes to plate tectonics were undertaken by Valentine &amp; Moores (1972) and Flessa &amp; Imbrie (1973), while the relationship of changing patterns of endemism through time in relation to plate movements was outlined by Hallam (1974).
For more up-to-date palaeobiogeographic reviews see Gray &amp; Boucot (1979) and Hallam (1981a, b).
The organic response to the changes in the physical environment induced by plate tectonics can be considered under three headings.
Migrating continents
I have proposed four simple distributional patterns for both marine and terrestrial animals which involve changes in time.
Convergence (not to be confused with phyletic convergence) refers to the degree of resemblance of faunas in different regions increasing from an earlier to a later period, and divergence refers to the reverse phenomenon (Fig. 18. 3).
Disjunct endemism refers to a type of regionally restricted distribution of a fossil taxon in which two or more component parts are separated by a major physical barrier and hence is not readily explicable in terms of present-day geography.
Complementarity in the distributional changes of contiguous marine and terrestrial animals is recognisable when one group exhibits convergence and the other divergence (Fig. 18.4).
This happens, for example, when a land connection is created between two hitherto isolated areas of continent, so allowing convergence of the terrestrial faunas to take place, while severing of a once-continuous landmass gives rise to divergence as a result of genetic isolation.
Pliocene uplift of the isthmus linking North and South America, apparently related to movement of the Cocos Plate in the East Pacific, is a classic example of complementarity and is also significant from a Darwinian point of view.
Substantial cross-migration of terrestrial mammal faunas coincided with mass extinction of the endemic South American mammals which had been isolated by sea through Tertiary times.
This has generally been assumed to have been the result of competition from the adaptively superior North American mammals, but Marshall (1981) has recently argued that at least some of the extinction was caused by changes in the physical environment.
In contrast, the Australian marsupial fauna has remained isolated by sea and not until late Tertiary times, when Australia-New Guinea had moved close to the Indonesian islands, was even limited and chance colonisation possible of Asian placentals across small water barriers (Whitmore, 1981).
More generally, the substantial diversity increase of faunas through the Mesozoic and Cainozoic, accelerating from mid-Cretaceous times onwards, which has recently been confirmed statistically (Sepkoski, Bambach, Raup &amp; Valentine, 198 1), is evidently related in substantial part to the progressive break-up of the late Palaeozoic-early Mesozoic supercontinent known as Pangaea, with the consequent increase in endemism of both terrestrial and neritic groups (Valentine, 1973).
Sea-level changes
Among the multiplicity of causes proposed to account for mass extinction events, both terrestrial and extra-terrestrial, strong supporting evidence for a succession of such events is available for only one, related to eustatic changes of sea level (Fig. 18.5).
Whether the extinctions among neritic organisms were the consequence of regression of epicontinental seas (Newell, 1967) or the widespread bottom-water anoxia characteristic of the initial phase of subsequent transgression (Hallam, 1981c) a significant reduction of habitat area and hence deterioration of the environment would have been produced by either phenomenon.
By far the most important episode of mass extinction took place at the end of the Permian period (Sepkoski et al.,
1981), when it is estimated that as many as 96% of all marine species died out (Raup, 1979).
Schopf (1974) and Simberloff (1974) have demonstrated that diversity changes of marine invertebrates across the Permian-Triassic boundary correlate closely with areas of epicontinental sea, in fact seem to obey rather well the ecologists' well known ' species-area ' relationship.
The even more familiar end-Cretaceous mass extinction event is also associated with substantial regression (Fig. 18.5), as are the equally striking events in the marine realm at or near the end of the Ordovician and Triassic, and to a lesser extent the Devonian (Raup &amp; Sepkoski, 1982, and Fig. 18.6).
It is by no means clear, however, with what particular plate tectonic events they may have been associated, either directly or indirectly.
The end-Ordovician regression is indeed widely thought to be the result of the growth of a Saharan ice sheet.
The question may be raised: to what extent were the extinctions selective or random?
Computer mode fling exercises have persuaded Raup (1981) that for the important group of trilobites it was more a matter of ' bad genes' than ' bad luck '.
On the other hand, a species extinction rate for the end of the Permian as high as 96% implies that the role of chance may in extreme cases have been dominant (Raup, 1979).
Generally speaking, organisms in warm, shallow seas that either build or are closely associated with reefs have been relatively vulnerable to extinction, as have planktonic foraminifera and ammonites, which have undergone a succession of ' boom and bust ' cycles (Vermeij, 1978; Haliam, 1981c).
Among terrestrial vertebrates (as well as at least some marine invertebrates) large organisms have been more vulnerable (Bakker, 1977).
With regard to phases of spectacular radiation, the two most important in the Phanerozoic record, affecting a large variety of organisms, correlate closely with major physical events that appear ultimately to be bound up with plate tectonics.
The radiation of marine faunas following the massive end-Palaeozoic extinctions was a long-continuing, progressive phenomenon, but was marked by a pulse of acceleration and replacement in the mid-Cretaceous, with diversity increase continuing into the Cainozoic (Fig. 18.6).
The better known end-Cretaceous extinction event can in many respects be considered as a mere temporary setback in this very dramatic faunal change.
The more-or-less coincident, spectacular mid-Cretaceous radiations included teleost fish, infaunal veneroid bivalves, carnivorous neogastropods and crabs (Vermeij, 1977; Stanley, 1977).
There were contemporary radiations of plankton, including coccolithophores, foraminifera, diatoms and dinoflagellates (Lipps, 1970) and deep-sea ichnofauna (Frey 7and; Seilacher, 19801, while on land the angiosperms rapidly replaced the gymnosperms as the dominant plants (Doyle, 1977).
It can hardly be coincidental that these remarkable evolutionary events, taking place within only a few million years, correspond so closely in time with an episode of exceptional igneous and urogenic activity (Larsen &amp; Pitman, 1972), the rapid disintegration of Pangaea (Hallam, 1980) and the biggest marine transgression since the mid-Palaeozoic, apparently produced either by a phase of accelerated sea-floor spreading or by a dramatic increase in the length of the ocean ridge system.
In particular, plankton evolution could perhaps have been stimulated by a drastic change in ocean current systems consequent upon Pangaea breakup, but the sea-level rise might have been just as significant.
Thus Hart (1980) has put forward a factually well-supported model relating the explosive diversification of late Cretaceous planktonic foraminifera to increasing depth of epicontinental seas.
This is a surprising conclusion with intriguing implications, bearing in mind the problem of near total extinction of this group at the end of the Cretaceous, which has appeared to be an even greater enigma than the contemporary extinction of the dinosaurs.
The latter might well have suffered from an increase in continentality of climate following regression, but it has not unreasonably been assumed by most palaeontologists that a planktonic group such as the globigerinid foraminifera should have been indifferent to what was happening to epicontinental seas.
Darwin was troubled by the sudden appearance of a wide diversity of fossils in the Cambrian, and tentatively suggested that there might have been a long interval prior to this period when no stratal record was preserved on the continents.
For a variety of reasons this type of explanation must be rejected today and the explosive diversification of Metazoa across the Precambrian-Cambrian boundary, as recorded in the strata, is now generally accepted as being a true reflection of what actually happened (Stanley, 1976; Seilacher, 1977).
This diversification has been shown to correspond closely to a simple exponential growth model.
As the number of taxa increased the rate of diversification seems to have become diversity-dependent (Sepkoski, 1978).
The close correspondence of diversity increase with rise of sea level in the Cambrian suggests the possibility of a causal correlation (Brasier, 1979).
We lack the information necessary to relate it with confidence to specific plate tectonic events, as is possible for the Cretaceous, but the Cambrian sea-level rise might well have been a consequence of opening of the Iapetus Ocean, with the growth of a spreading ridge (Anderton, 1980, 1982).
Climatic change
The pronounced increase in latitudinal temperature zonation through the course of the Cainozoic, as the world altered progressively from its Mesozoic condition of equability, must have had the effect of creating a great number of ecological niches.
This is seen as a major factor contributing towards the marked increase in faunal diversity (Valentine, 1973).
Increase in annual temperature range on the continents as a consequence of regression of epicontinental seas might well have played a significant role in the mass extinctions of large reptiles at the end of the Palaeozoic and Mesozoic.
Pangaea at the end of the Palaeozoic must have experienced a climate of extreme continentality, not only because of its coherence (Valentine &amp; Moores, 1972) but because of the high albedo of extensive low-latitude deserts (Barron et al.,
1980).
The extinction of many marine foraminiferal and ostracode species at or close to the Eocene-Oligocene boundary could be bound up with the establishment of the layer of cold, deep water in the oceans known as the psychrosphere.
This in turn relates to the establishment of the Circum-Antarctic current as first Australia and then South America broke away from that continent, and to the formation of south polar sea ice (Hallam, 1981c).
Discussion
I should now like to broaden the scope of this essay by discussing the relative importance of stochastic and deterministic, as well as biotic and physical, factors as promoters of evolution.
In recent years there has been an increasing tendency in some circles to investigate the evolutionary record of fossils in terms of general rules and processes without regard to specific causes operating on specific taxa.
Thus Van Valen (1973) applied the survivorship curve technique of population biologists to the study of extinction rates for numerous fossil taxa, and claimed to demonstrate a general approximation to linearity in his curves, which are cumulative frequency distributions of taxonomic durations with logarithmic ordinates.
This led him to propose a new evolutionary ' law ', which, in brief, states that within a relatively homogeneous higher taxon, subtaxa tend to become extinct at a stochastically constant rate.
In explanation Van Valen put forward what he termed the Red Queen's hypothesis, named after the Lewis Carroll character who found it took all the running one can do to keep in the same place.
It is thoroughly Darwinian in its stress on the paramount importance of biotic interactions.
All species within a given adaptive zone compete intensively.
A successful adaptive response by one species is assumed to occur at the expense of other species, which must either adapt by themselves speciating or become extinct, as the ' quality ' of their environment is reduced.
This phenomenon leads to an endless chain of adaptive responses and in the long run means that fitness and rate of extinction remain constant.
The high rate of diversification and evolutionary turnover in mammals is thought likely to be the result of a variety of factors, such as strong competitive interactions leading to specialisation in feeding methods, limitations on food supply, high mobility and energy use, interspecific aggression and territoriality.
Such factors will conspire to lower the ' resource threshold ' needed to prevent extinction, compared with other animals.
Epistandard rates of evolution are required to make up the loss through extinction.
Though the Red Queen model could conceivably apply to mammals, there are doubts about its more general validity.
Thus Stanley (1973) analyses the effect of competition on evolutionary rates by comparing mammals with bivalve molluscs.
In sharp contrast to mammals, bivalves are nearly all benthic suspension feeders which appear to mind their own business, being characterized by weak interactions with other species, primitive inflexible behaviour, uncrowded, largely sedentary mode of life and generalised feeding habits.
Limits on bivalve populations are imposed more by predation and fluctuations in the physical environment than by food resources, and biotic competition is minimal.
As Stanley remarks laconically, ' Interspecific aggression is not characteristic of bivalve behaviour '.
What is true of bivalves is without much doubt true of the majority of benthic invertebrates.
More general criticisms of the Red Queen hypothesis have been voiced by Foin, Valentine &amp; Ayala (1975), Raup (1975), Salthe (1975) and Sepkoski (1975).
In a nutshell, it is argued that either Van Valen's results show linearity with time, which is held to be biologically without significance, or most do not, in which case Van Valen's ' law ' breaks down.
Other stochastic models have been explored by the computer generation of phylogenetic diagrams (cladograms), with termination and branching events being controlled using random numbers (Raup et al.,
1973; Gould et al.,
1977).
Part of the input dealt with the required establishment of an equilibrium diversity.
The cladograms so produced displayed a variety of patterns, many of which appear to simulate temporal diversity changes in well-studied fossil groups.
It does not, of course, follow that the radiation and extinction of the monophyletic units known as clades do not have deterministic explanations.
Stanley, Signor, Lidgard &amp; Karr (1981) present some cogent criticism of the work of Raup and his associates and establish that chance factors have not played a dominant role in producing dramatic changes in diversity.
Hoffman (1981) gives a sophisticated critique both of stochastic modelling and the application of equilibrium theory in palaeontology.
The most powerful case against randomness is that afforded by mass extinction and radiation events, whereby a wide variety of taxonomic groups with different modes of life, and effective biological independence, have experienced synchronised diversity reduction or increase.
This strongly implies the operation of some form of control by the physical environment, and some likely examples of an ultimate link with plate tectonics have been presented above.
The interesting question arises: were the several widely acknowledged major episodes of mass extinction in the Phanerozoic caused by exceptional events different in kind from what occurred in the much lengthier intervening periods, or are they merely the spectacular end-members of a whole series?
Those who have recourse to the deus ex machina of lethal rays or thunderbolts from outer space may incline to the former view, but only at the expense of disregarding the abundant evidence in the stratigraphic record of a correlation between mass extinctions and physical events on earth (Hallam, 1981c).
Even in the one case where independent evidence is claimed in the form of an abnormal enrichment in thin sedimentary layers of iridium and other platinum group metals, namely the much-publicized asteroid impact hypothesis for the end-Cretaceous extinctions (Alvarez, Alvarez, Asaro &amp; Michel, 1980) serious doubts have been raised by a careful analysis of some key palaeontological evidence (Clemens, Archibald &amp; Hickey, 1981).
There is in fact evidence to indicate that regressions at intermediate and small scales correlate with coordinated extinction events in particular fossil groups (e.g. Hallam, 1978; Williamson, 1981) but much more work is required to establish how generally such a relationship holds.
If it does turn out to be of general significance at a variety of scales it will imply that the increased environmental stress associated in some way with regressions promotes an increase in extinction rate and corresponding vacation of ecological niches, with a consequent opportunity for new species to establish themselves.
Alternatively, or in association, it could be that the pronounced environmental changes associated with the regression events act to destabilize intraspecific selection pressures and hence promote speciation.
The comparatively long time intervals between such environmental vicissitudes may be characterized by stasis in ecosystems as well as the component species.
A brief consideration of the well-documented Pleistocene sea-level oscillations caused by glaciation and deglaciation suggests an apparent problem, because these events do not, by and large, correlate with episodes of pronounced extinction or speciation.
Indeed, the characteristic response of both terrestrial and marine organisms to the pronounced climatic changes of the Pleistocene has been to migrate to ecological refuges, in effect to track their environment.
Without such a phenomenon, in fact, stratigraphic correlation would operate under a crippling handicap.
Why therefore did not organisms respond in a similar way to the much slower changes of sea level in the lengthy periods of climatic equability?
At least two possibilities readily suggest themselves.
Perhaps the Pleistocene sea-level falls, though dramatically rapid in geological terms, were too short-lived to have the kind of environmental impact required to cause extinction.
Or perhaps the increasingly unstable environments of the late Cainozoic associated with climatic deterioration caused a selection for eurytopic organisms well adapted to withstand environmental instability.
In contrast, the comparative stability of, for instance, Mesozoic environments might have allowed the establishment of complex ecosystems characterized by comparatively stenotopic organisms, which would have been vulnerable to even modest environmental vicissitudes.
Such ideas need to be tested, and further studies made on the relative susceptibility of different fossil groups to extinction as a result of particular environmental events.
Furthermore, we need to enquire further into the factors governing the response of organisms to given disturbances of their environment, whether it involves migration to refuges or extinction and speciation.
There is clearly a rich field for future investigation into matters of such major importance.
Though I believe that Darwin laid too much stress on biotic interactions as a promoter of evolution and extinction, to the extent of substantially dismissing changes in the physical environment, there is no justification for going to the other extreme, as several examples will illustrate.
That the South American mammals might have gone extinct in the early Pleistocene as a consequence of competition from the North American invaders has already been noted, though we should take due account of the caveats of Marshall (1981).
With regard to marine benthic communities, the expansion of mobile, infaunal deposit- and suspension-feeding populations in the Mesozoic correlates with a decline of immobile suspension feeders on soft substrates (Thayer, 1979).
Some of the groups in decline, such as stalked crinoids, cidaroid echinoids (Kier, 1974) and brachiopods, apparently sought refuge in the deep sea.
The late Mesozoic radiation of predatory crabs, neogastropods and teleosts correlates with an increase in resistance to destruction of the shells of their molluscan prey (Vermeij, 1977; Ward, 1981).
Further likely examples of coevolution concern the rise of the angiosperms.
From the late Cretaceous onwards the new sea-grass communities transformed the shallow neritic environment and seem to have promoted the contemporary radiation of deposit-feeding and epiphytic gastropods and miliolid foraminifera (Brasier, 1975).
Although the fossil record is much poorer, coevolution with the terrestrial angiosperms probably accounts for a major component of the radiation of insects and birds.
In all these cases, however, the initial trigger to change might have been physical rather than biotic.
A final intriguing example involves a subtle interplay between biotic and physical factors, and concerns the common phyletic trend towards size increase known as Cope's Rule, which I suspect may be the only important gradualistic exception to punctuated equilibria (Hallam, 1978).
There is no shortage of adaptive explanations for size increase, but Darwin is one of the few to have pointed out that large organisms such as mastodonts and dinosaurs would have been more vulnerable to extinction because of the limitations of food resources.
I have proposed that, on the reasonable assumption that resources remained more or less constant for the time in question, the price exacted for (phyletically) growing larger was to become rarer, thereby increasing the probability of extinction (Hallam, 1975).
Whether one applies the older notion of a trend from less to more specialized or the newer concept of a trend from an r-to a K-selected adaptive strategy (ecologists seem to have a love-hate relationship towards r and K selection (Dawkins, 1981)), the phyletically younger organisms would have become progressively more vulnerable to environmental disturbance.
The stratigraphic record has been compared to the traditional life of a soldier  long periods of boredom interrupted by moments of terror (Ager, 1973).
The fossil record suggests that the largest members of a phyletic series usually had the most reason to be apprehensive of the future. [C]
SELECTION IN RELATION TO SEX T.H. CLUTTON-BROCK
In many animal species, differences between the sexes are pronounced.
In fact, there are relatively few morphological, physiological or behavioural characteristics that do not differ to some extent between males and females (see Glucksman, 1974) and the extent of these differences varies widely between species.
For example, sexual differences in body size range from species, like worms of the genus Bonnellia, where females can be over 25 times the length of males (Barnes, 1974) to species like the southern elephant seal, Mirounga leonina, where mature males average eight times the weight of females (Bryden, 1969).
Other sex differences are less apparent  such as those in fat deposition, in haemoglobin levels and in metabolic rate among mammals (see Glucksman, 1974) or those in auditory apparatus among frogs (Narins &amp; Capranica, 1976, 1978).
And many have only recently been explored: for example, recent research shows that in some species there are pronounced sex differences in feeding ecology (Gautier-Hion, 1980) as well as in the effects of starvation on survival (Widdowson, 1976).
It was to provide an explanation for the evolution of sex differences that Charles Darwin formulated the theory of sexual selection, described first in The Origin of Species (1859) and later, in greater depth, in The Descent of Man (1871).
In this chapter, I briefly trace the development of the theory and describe recent attempts to measure variation in reproductive success in males and females, as well as some of the practical problems involved.
However, as I argue in the final section, the extent of sexual dimorphism will depend not on the extent to which reproductive success varies in the two sexes but on the comparative effects of particular phenotypic traits on the breeding success of males and females.
A brief history of sexual selection
The origin of the theory of sexual selection can be traced to a peacock  or, rather, a peahen owned by Lady Tynte.
This bird, born around 1762, lived in such comfortable circumstances that it had already reared eight broods when, to the consternation of its noble owner, it suddenly developed the plumage and spurs of a male and thereafter refused to lay another egg.
No possible confusion of identities could have occurred since Lady Tynte was able to recognize her favourite by the nobs on its toes, which were unaffected by its change in appearance (Hunter, 1837).
It is not known whether Lady Tynte investigated just what had happened to the bird's genitalia but subsequent studies of female pheasants showing similar transvestite tendencies revealed that only sexual differences developing at or after puberty were affected and that the reproductive organs themselves remained unaltered (Yarrell, 1827).
This finally led John Hunter, the eminent surgeon, anatomist and classifier of monsters to produce a seminal paper (An account of an extraordinary pheasant, Hunter, 1837) in which he proposed that differences between the sexes were of two kinds: those involving the sexual organs themselves, which were evident from birth and did not change during an individual's lifetime; and those that did not develop until the animal approached breeding age, such as differences in body size, plumage and in the tendency to be fat which he termed ' secondary ' marks or characters of sex (Hunter, 1837, 1861).
Some of these, like the plumage of gallinaceous birds, could change during an individual's lifetime but these differences were not evident at birth.
Hunter realized both that ' secondary ' sexual characters were functionally related to fighting or display and that their extent varied with ecology.
Hunter's distinction between primary and secondary sexual characters was adopted by Charles Darwin (1871) with a subtle distinction.
Darwin was aware that many ' secondary ' sex differences were evident at birth (or hatching) and distinguished between the two categories on functional rather than on ontogenetic grounds.
Darwin's primary sexual characters were those connected with the act of reproduction itself while secondary sexual characters were used in acquiring mating partners.
To these two categories, Darwin added a third: sex differences' related to different habits of life, and not at all, or only indirectly, to the reproductive functions', among which he included structures associated with sex differences in feeding behaviour (Darwin, 1871).
The theory of sexual selection was intended to provide an explanation only of secondary sexual characteristics.
Darwin realized that many secondary sexual differences were a consequence of the greater intensity of competition between males for access to mates and that many traits were more highly developed in males either because they conferred an advantage in fights or because they rendered their possessor more attractive to females.
He distinguished sexual selection from natural selection on two grounds:.
first, that it was a consequence of competition between members of the same sex rather than between members of different sexes or species; and, second, that it depended on variation in reproductive success rather than survival.
'... This form of selection depends, not on a struggle for existence in relation to other organic beings or to external conditions, but on a struggle between the individuals of one sex, generally the males, for the possession of the other sex.
The result is not death to the unsuccessful competitor but few or no offspring. '
Darwin's theory of sexual selection was less readily accepted by scientists than the theory of natural selection.
Wallace (1889) agreed that combat between males was an important source of selection pressures leading to sexual dimorphism but regarded this as a form of natural selection on the grounds that it increased ' the vigour and fighting power of the male animal, since, in every case, the weaker are either killed, wounded or driven away '.
He regarded Darwin's second mode of sexual selection  female choice of particular males  as unimportant on the grounds that any consequences which female choice might have would be annulled by natural selection  unless females selected the fittest males, in which case the results of sexual and natural selection would be inseparable.
He also pointed to the lack of evidence of consistent female choice for mates carrying particular characteristics.
Some fifty years later, the same points were reaffirmed in two influential papers by Huxley (1938a, b).
Wallace's objection that sexual selection is a form of natural selection is semantically correct  after all, Darwin originally coined the term ' natural selection ' in order to mark its relation to man's power of selection, and the opposite of natural selection is not sexual but artificial selection (see Brown, 1975; Halliday, 1978).
However, his insistence that the process of sexual selection described by Darwin could only increase the average reproductive success or survival of males is clearly wrong (Lande, 1980).
Especially in polygynous species, the costs of combat are frequently high (Geist, 1971; Clutton-Brock, Albon, Gibson &amp; Guinness, 1979) and so, too are the costs of many sexually dimorphic characters associated with combat, such as increased male body size and weapon development (Clutton-Brock, Guinness &amp; Albon, 1982): in species where males are substantially larger than females, both growing and adult males are often more likely to die than females (Robinette, Gashwiler, Low &amp; Jones, 1957; Grubb, 1974; Howe, 1977) and in one reindeer population which crashed from 6000 to 42, only one of the remaining adults was a male (Klein, 1968).
Sexual selection on males may also reduce the average fitness of females (Lande, 1980): in species where adult males are substantially larger than females, producing sons appears to depress the mother's subsequent reproductive success more than producing daughters (Clutton-Brock, Albon &amp; Guinness, 1981).
Wallace's theoretical objections to the importance of female choice as a source of sexual selection on males can also be discounted.
R. A. Fisher (1930) demonstrated that female choice for particular male characteristics (such as tail size) can cause them to develop to a point at which they reduce the average fitness of males.
Subsequent treatments have confirmed Fisher's conclusions (O'Donald, 1980; Lande, 1981) and shown that the process need not depend on the initial female preference favouring more viable males (Kirkpatrick, 1982).
However, while there is extensive evidence of assortative mating (O'Donald, 1980), of the importance of plumage characteristics in courtship (Williams, 1982) and of female preference for males who can defend superior breeding territories (Pleszczynska, 1978) only very recently has it been demonstrated that consistent female choice for any continuous morphological character in males is an important source of variation in male reproductive success.
By experimental manipulation of tail length in widdow birds of the African genus Euplectes, Andersson has been able to alter both the extent to which males are favoured by females and their mating success (Andersson, 1982).
This scarcity of evidence does not mean that the evolution of secondary sexual characters through female choice is uncommon, for mating preferences are usually difficult to demonstrate, particularly where inter-male competition is also involved.
Nevertheless, the possibility remains that, as Wallace argued, many of the sex differences in plumage and coloration ascribed by Darwin to the action of female choice may have evolved because they help the sexes to recognize or locate each other or because they improve male success in competitive interactions.
Measures of sexual selection
Darwin was not specific as to why males should typically compete more strongly for access to breeding partners than females and it was left to biologists of this century to provide the answer (Fisher, 1930; Bateman, 1948; Trivers, 1972).
The reason why males usually compete more intensely is most easily understood by considering the energetic costs of reproduction to each sex.
In most animals, the energetic costs of fertilization to the male are minimal whereas the costs of reproduction to the female are substantial.
Consequently, males are capable of fathering more progeny than females can bear and rear: in current terminology (Trivers, 1972) they invest less heavily in their offspring than females.
In species where successful males can monopolize breeding access to large numbers of females but similar numbers of males and females are recruited, direct competition between males is likely to be intense, aggressive interactions may be frequent and the selective advantages of possessing traits that affect success in combat may be higher among males than among females.
However, neither Trivers' explanation of the prevalence of increased competition among males nor Darwin's description of the theory of sexual selection provide an operational definition of the intensity of sexual selection (Wade, 1979).
Many different definitions of sexual selection have been proposed but two kinds are in common use.
First, some workers argue that the intensity of sexual selection will depend on the relative variability of reproductive success among males and females.
For example, Ralls (1977) argues that ' the intensity of intrasexual selection in a species should be proportional to the ratio of the lifetime number of offspring sired by a highly successful male compared to the number born by a highly successful female in her lifetime ' while Payne (1979) suggests that the extent to which variance in breeding success differs between the sexes is important.
This position is also sometimes mistakenly attributed to Bateman (see Wade &amp; Arnold, 1980).
In contrast, other workers use measures of the extent to which male breeding success deviates from mean male success as estimates of the intensity of sexual selection on males (see Wade, 1979; Howard, 1979; Wade &amp; Arnold, 1980).
Similarly, the intensity of sexual selection in females can be estimated by measuring the extent to which female breeding success deviates from mean female success.
Several statistical measures are in use, including the coefficient of variation, Pielou's (1969) index of evenness and variance in breeding success divided by the square of mean breeding success  a measure derived from Crow's (1958) index of the intensity of selection, where V is variance in fitness and W is the mean fitness of the population.
This can either be calculated using total variance in breeding success or using only variance attributable to differences in the number of mates per individual (see Wade &amp; Arnold, 1980).
Ratios of variation in male and female success can be used to estimate the relative rate at which male and female characteristics can change.
However, since the extent of variation among females differs widely between species (see below), they are of limited value as measures of the comparative intensity of sexual selection (see Wade &amp; Arnold, 1980).
The three measures of the extent to which reproductive success varies within each sex are fundamentally similar though the last is the most convenient since it offers a measure of the potential change in fitness between generations, relative to the average (see Crow, 1958).
It is still necessary to decide whether to use total variation in breeding success in any calculation or just variation due to differences in mate number.
Where the aim of the calculations is to estimate the potential rate of genetic change, total variance in breeding success is clearly the relevant measure to use.
In contrast, where calculations are carried out to assess the comparative importance of differences in reproductive success versus survival for each sex, only variation in mate number may be included (see Wade &amp; Arnold, 1980)  though measures of the intensity of sexual selection as conceived by Darwin should also include variation due to differences in mate quality, which may be an important cause of variation in lifetime breeding success both among males and females in monogamous species.
Lastly, where the aim is to investigate the functional significance of particular phenotypic sex differences, it may be necessary to calculate selection intensities for specific episodes of selection, such as mating success at particular ages (see Arnold &amp; Wade, 1983) for the effect of particular traits on lifetime reproductive success may often be obscured by the influence of other variables or by random variation.
One general point concerning all indices of sexual selection must be emphasized.
Since many of the traits affecting breeding success are strongly influenced by rearing conditions (see Clutton-Brock et al.,
1982), these measures reflect the potential rate of genetic change and may provide little indication of actual rates of change in established populations.
Practical problems
Although it is relatively easy to decide how the intensity of sexual selection should be estimated, collecting the relevant data poses a variety of logistic problems (Ralls, 1977; Davies, 1982).
In particular, in many polygynous species a proportion of males avoid competing directly with larger or older animals and adopt a policy of surreptitious fertilization or kleptogamy (e.g. Clutton-Brock et al.,
1977; Wirtz, 1982).
As a result, it is often difficult to be sure that breeding males fertilize all the females in the groups that they guard: for example, even if territorial male red-winged blackbirds (Agelaius phoeniceus) are vasectomized, their females sometimes lay fertile eggs (Bray, Kennelly &amp; Guarino, 1975).
Moreover, all indices of the intensity of sexual selection beg a problem of fundamental importance: over what period should reproductive success be measured?
It is widely agreed that the lifetime reproductive success of individuals is the most satisfactory measure of fitness that it is usually possible to collect (see Falconer, 1960; Cavalli-Sforza &amp; Bodmer, 1961; Maynard Smith, 1969; Grafen, 1982).
This may not always be the case  for example, in rapidly expanding populations, selection may favour reproductive rate at the expense of lifetime reproductive success (see Lewontin, 1965; Elliot, 1975)  but exceptions are probably rare, especially in long-lived species where variation in longevity greatly exceeds variation in age at first breeding.
While many studies of sexual selection pay lip service to the importance of basing calculations on measures of lifetime success, most use estimates of success calculated over a part of the animal's lifespan.
The time dimension used varies widely: some studies use measures of instantaneous reproductive success (IRS) calculated across different classes of males at a particular point in time (e.g. Mason, 1964; Scheiring, 1977; McCauley g&amp;, 1980), while others measure dally reproductive success (DRS) or seasonal reproductive success (SRS) (see Howard, 1979; Payne, 1979).
Very recently, estimates of individual variation in lifetime reproductive success (LRS) have become available from field studies of a small number of species, including a territorial and promiscuous invertebrate, the dragonfly Erythemis simplicicolis (McVey, 1981), two monogamous birds, the great tit, Parus major (McGregor, Krebs &amp; Perrins, 1981) and the kittiwake Rissa tridactyla (J. Coulson &amp; C. Thomas, personal communication), and one polygynous mammal, the red deer, Cervus elaphus (Clutton-Brock et al.,
1982) while in several other species, including sage grouse (Centrocercus urophasianus: Wiley, 1973) estimates of individual success are available which span several breeding seasons.
These make it possible, for the first time, to test the extent to which variation in instantaneous, dally and seasonal reproductive success reflect variation in lifetime success.
There appears to be no consistent relationship between variation in IRS, DRS, SRS and LRS (see Table 23. 1).
There are several reasons why variation in daily reproductive success may not reflect variation in either seasonal or lifetime success.
In many polygynous species reproductive success varies from hour to hour and day to day and this may cause variation in IRS or DRS to overestimate variation in SRS and LRS.
For example, red deer stags which hold a large harem in a sheltered site on one day may have few or no hinds the next day if the wind changes (see Clutton-Brock et al.,
1982).
As a result, measures of variation in DRS overestimate SRS and LRS in red deer (see Table 23.2).
Estimates of variation in DRS and SRS in sage grouse (Centrocercus urophasianus) show the same trend though estimates of variation in DRS and LRS were similar in McVey's (1981) study of dragonflies.
In addition, variation in SRS will not reflect variation in LRS (see Gadgil, 1972) if individuals that breed successfully show reduced success in future or are likely to cease breeding or die earlier than less successful breeders.
There is evidence that successful reproduction reduces the future reproductive capabilities or survival of females in several species (e.g. Wooller &amp; Coulson, 1977; Altmann, Altmann &amp; Hausfater, 1978; Guinness, Albon &amp; Clutton-Brock, 1978; Bryant, 1979) and in some species successful males are likely to die before unsuccessful ones (Geist, 1971).
However, several other studies have shown that successful breeders live longer (Drosophila melanogaster: Partridge &amp; Farquhar, 198 1).
For example, in red deer stags, where harem size is one of the principal determinants of reproductive success (Clutton-Brock et al.,
1979, 1982), not only do stags that hold large harems hold them for longer within particular breeding seasons than those which hold smaller harems (Fig. 23. 1a), but individuals that are consistently successful in securing large harems throughout their lives tend to live longer than their less successful competitors (Fig. 23.1b).
Trends of this kind will produce an opposite bias and may cause variation in SRS to underestimate variation in LRS.
For example, the fighting ability and reproductive success of red deer stags shows a pronounced peak between the ages of seven and ten years (see Fig. 23,2).
If variation in breeding success is calculated for stags of above a year old (the age of sexual maturity), it greatly exceeds more realistic measures of variation in breeding success such as variation in seasonal success within cohorts or lifetime success (see Table 23.3).
This effect will also tend to overestimate variation in male success relative to variation in female success (see Fig. 23.2).
p 465: TABLE
A final problem is that unless breeding success is measured across the lifetime, a biased set of males may be sampled.
Especially in polygynous species, males that fail to win a breeding territory spend their time on the fringes of the breeding population and often show high mortality.
The field observer approaching a breeding colony is likely to see mostly (if not only) breeding males.
If he calculates the extent to which reproductive success varies across adults that have managed to gain a breeding territory, he will underestimate the real variance since he will exclude non-breeders.
Dividing by the mean success of males in this sample will accentuate this error since the mean success of the males sampled will be higher than that of the male population as a whole.
Table 23.4 shows how the restriction of the sample to harem holding males in red deer reduces estimates of daily variation in the number of females held.
Estimates that are probably biased in this way are already widespread in the literature (see Trivers, 1976;, Arnold &amp; Wade, 1983).
pp. 467C8: TABLES
None of these four biases are likely to be consistent either across the two sexes or across different species: all four are more likely to affect estimates of variation in male success than estimates of variation in female success in polygynous species, though in contrary directions.
Ignoring short term variation and age effects will tend to overestimate variation in male success relative to variation in female success, which is usually less strongly age-dependent in polygynous species (see Fig. 23.2) and less likely to vary widely from day to day.
Conversely, positive correlations between seasonal success and longevity or sampling only successful males will lead to underestimates of variation in male success relative to variation in female success.
That the two sets of biases oppose each other is no guarantee that short term measures of reproductive success will provide reliable estimates of variation in lifetime success for the comparative strengths of the different biases probably vary widely too.
Moreover, sex differences in the length of effective reproductive lifespans and in the influence of age on breeding success are likely to be more pronounced in polygynous species than in monogamous ones.
As a result, short-term estimates of variation in male breeding success are likely to overestimate the extent to which male success varies in polygynous species relative to similar measures for monogamous ones.
Biased estimates of variation in reproductive success may also cause the effects of particular phenotypic traits on reproductive success to be overestimated.
This applies particularly to traits, such as body size, which are themselves related to age.
For example, where the body size and reproductive success of males both increase with age, the effects of size on breeding success may be grossly exaggerated if age differences in size are ignored.
This effect is again likely to be stronger in males than females.
The most obvious conclusion to be drawn is the need for studies of the extent to which lifetime success varies and of the factors which affect it.
Where this is impossible, an alternative approach is to concentrate on studies of variation in breeding success within cohorts but, even here, problems will arise if breeding success is consistently related to longevity.
Polygyny, reproductive success and sexual dimorphism
The theory of sexual selection has led to four common predictions about the relationship between breeding success and sexual dimorphism.
Variation in reproductive success should be greater in males than females in polygynous species but similar in the two sexes in monogamous ones.
Both in red deer and in Erythemis, variation in lifetime breeding success is substantially greater among males than females whereas, in kittiwakes, variation is similar in both sexes (Figs. 23.3, 23.4).
When the red deer samples are restricted to animals that reach breeding age, sex differences in the extent to which breeding success varies are accentuated (Table 23.5).
Considering that male red deer can hold harems of over thirty hinds, it is, perhaps, surprising that male success does not vary more widely.
This is partly because only a proportion of hinds conceive in a given year and individual stags rarely hold harems throughout the whole breeding season, and partly because few stags breed successfully for more than four years.
In contrast, the range of breeding success among hinds is greater than might be expected because their potential breeding lifespans are long (over 12 years) and individuals tend to be either consistently successful or consistently unsuccessful breeders.
Variation in reproductive success should be greater among males of polygynous species than among males of monogamous ones.
One surprising result of the comparison between red deer and kittiwakes is that variation in lifetime breeding success is little greater in red deer stags than in male kittiwakes (see Table 23.5): in fact if all individuals born/hatched are included, it is slightly (though not significantly) greater in male kittiwakes.
The comparison is an unsatisfactory one since there are important differences in the life histories of the two species: male kittiwakes can breed for many more seasons, adult mortality is not so strongly age-dependent and females can fledge up to three young per year (see Coulson, 1966, 1968; Coulson &amp; Wooller, 1976; Wooller &amp; Coulson, 1977).
However, this example serves to emphasize how misleading it can be to assume that the breeding sex ratio necessarily reflects the extent to which male reproductive success varies for, even among closely related species, it is likely to be the case that males have substantially longer breeding lifespans in monogamous species than in polygynous ones (see Wiley, 1974; Clutton-Brock et al.,
1982).
Direct competition for mates will be more intense among males of polygynous species than among males of monogamous ones.
Fights between males can be common and dangerous in polygynous species (see Geist, 1971).
However, competition between males can also be intense in monogamous species (Lack, 1954; Kleiman, 1977) and data are not yet available which would permit a meaningful comparison between the two groups of species.
In fact, it is unsafe to assume that the intensity of direct competition between males should necessarily be reduced in monogamous species.
Though variation in male success may be caused principally by differences in mate or territory quality (whereas, in polygynous species, differences in mate number are the main cause of differences in success: Bateman, 1948; Wade, 1979; Clutton-Brock et al.,
1982), monogamous males might be expected to compete as intensely for the best mates or territories as do polygynous males for the biggest harems.
Where this is not the case, it may be because males can not identify the breeding potential of young females or because female choice pre-empts male competition rather than because variation in success is slight among males.
Sexual dimorphism will be most developed among strongly polygynous species and least developed among monogamous ones.
In many different groups of animals there is an association between polygyny and sexual dimorphism and Darwin himself was well aware of the relationship.
More recently, a variety of studies have demonstrated statistical relationships between the degree of polygyny and the development of sexual dimorphism, though the relationship is not always a close one (Ralls, 1977).
Compared to monogamous species, polygynous ones usually show greater sexual dimorphism in body size (Clutton-Brock, Harvey &amp; Rudder, 1977; Shine, 1979; Alexander et al.,
1979) while weapons used in intraspecific combat, such as the canines of primates and the antlers of deer, are also more developed in the males of polygynous species (Harvey, Kavanagh &amp; Clutton-Brock, 1978; Clutton-Brock, Albon &amp; Harvey, 1980).
However, there are many exceptions (Ralls, 1977).
In some groups of animals, the relationship between the extent of polygyny and the degree of sexual dimorphism is not a close one (Clutton-Brock et al.,
1977).
In addition, some polygynous species, like Burchell's zebra (Equus burchelli), show little or no size dimorphism while others, like the spotted hyena (Crocuta crocuta) and the Weddell seal (Leptonychotes weddelli), even show reversed dimorphism (Klingel, 1972; Kruuk, 1972; Stirling, 1969).
While it is possible that some of these exceptions result from the differing energetic requirements of males and females and attendant selection pressures affecting the relative size of the two sexes (Selander, 1972; Downhower, 1976), the common association between sex differences in size and the development of male weaponry suggests that selection pressures associated with breeding competition are frequently involved.
To interpret exceptions to the general rule that sexual dimorphism increases with the degree of polygyny, we need to remember that it is the comparative effects of phenotypic traits on reproductive success in males and females that will determine the degree of dimorphism and not the amount of variation in reproductive success per se (see Price, 1970; Lande, 1980).
For example, while sexual dimorphism in size is likely to evolve where variation in male success is greater than female success and a given increment in body size has the same effect on breeding success in both sexes, it will also evolve if variation in reproductive success is similar in both sexes but size has a greater influence on success in males or even if variation in success is greater in females but the effects of size are greater among males.
Conversely, sexual dimorphism in size is unlikely to evolve in circumstances where variation in reproductive success is greater among males but the effects of size on reproductive success are slight in both sexes.
In the simplest of all possible worlds sexual dimorphism should, perhaps, be predicted by the relative slope of lifetime reproductive success on body size in males and females.
However, there is no reason to suppose that relationships between size and reproductive success will be linear or that they will follow a similar pattern in both sexes.
Indeed, where size differences are heritable and stabilizing selection is operating, there is every reason to suppose that relationships between size and reproductive success will not be linear.
In addition, even a knowledge of the relationship between body size and lifetime reproductive success in the two sexes will not answer whether the association occurs for reasons connected with breeding competition or because the two sexes differ in their energy requirements (see Downhower, 1976).
To sort out these questions, it will be necessary to identify the particular episodes of selection during which size influences breeding success in males and females (see Arnold &amp; Wade, 1983).
This argument raises the question of why it is that sexual dimorphism and polygyny are related at all.
The most likely explanation is that the factors determining breeding success in males and females tend to be most similar in monogamous species and most different in highly polygynous ones.
The factors affecting breeding success in males and females certainly differ widely in polygynous species.
For example, in red deer, longevity, offspring survival and home range quality have a greater effect on the reproductive success of hinds than on that of stags.
In contrast, fighting ability, body size and (because of its effect on adult size) early growth have a more important effect on the reproductive success of stags than hinds (Clutton-Brock et al.,
1982).
The factors affecting lifetime success in males and females of monogamous species have yet to be described.
While it is clear that they will not be identical (see McGregor et al.,
1981), it is reasonable to suppose that, especially among species that pair for life, they are likely to be more similar than in polygynous species.
Selection pressures on males and females
Emphasis on the importance of understanding the factors affecting breeding success in males and females has the advantage that it forces us to ask specific comparative questions concerning the functional significance of particular sex differences.
For example, does body size have a less important effect on the reproductive success of male zebras than male bovids because zebras fight with their teeth and hooves rather than by pushing?
Similarly, does size have a greater effect on reproductive success in female Weddell seals compared to land or pack-ice breeding species because they breed on fast ice and defend access to water holes?
Conversely, is body size less important in male Weddells compared to other species because they defend underwater territories where success depends on manoeuvrability and because females are widely dispersed?
Thinking in these terms may help us to understand the distribution of many other sex differences.
For example, among hermaphroditic reef fish, some species begin life as females and a proportion of individuals later become males (protogyny) (Warner, Robertson &amp; Leigh, 1975; Robertson &amp; Hoffman, 1977).
But in a few species, such as the clown fishes (Amphiprion), individuals start life as males and a proportion later become females (protandry) (Fricke &amp; Fricke, 1977).
The females of most protogynous species spawn on the edge of the reef, releasing their eggs into the plankton, and seldom compete for spawning sites.
In contrast, female clownfish lay their eggs around sea anemones which they subsequently help to defend.
Does resource defence by females again increase the benefits of large body size to female clownfish while reducing the benefits of size to males as a consequence of female dispersion?
Understanding the comparative effects of size may even help to explain variation in birth sex ratios.
It has been suggested that in species where reproductive success varies more widely among males than females and is influenced by parental investment, parents who can afford to invest heavily in their offspring should produce sons while those that can not do so should tend to produce daughters (Trivers &amp; Willard, 1973).
In apparent contradiction to the theory, dominant female baboons and macaques produce more daughters than sons while subordinates produce more sons than daughters (Altmann, 1980; Simpson &amp; Simpson, 1982).
One possible explanation is that even though reproductive success probably varies more widely among males in these species, maternal rank has a stronger effect on the success of daughters since females remain in their mother's troop and inherit her rank while sons disperse to other groups and may be unable to benefit substantially from their mother's rank (Altmann, 1980).
Perhaps most importantly, emphasis on examining the effects of particular traits on the breeding success of males and females should encourage us to investigate the adaptive significance of sex differences whose functions are not immediately obvious.
For example, in many mammals growing males lay down less body fat than females (Glucksman, 1974) and suffer heavier mortality during periods of food shortage as a consequence (Clutton-Brock et al.,
1982).
One possible explanation is that because early growth exerts a greater effect on reproductive success in males than females (see above), selection favours increased investment by young males in growth at the expense of laying down body fat to assure survival during periods of food shortage.
This prompts the question as to whether young females lay down less fat in species showing reversed size dimorphism.
Recent research shows that this is the case in at least one species belonging to this category (the European sparrow hawk: I. Newton, personal communication).
These explanations are cautiously worded  and necessarily so, for our current knowledge of the factors affecting reproductive success in males and females is rudimentary.
Nevertheless, the way ahead is clear.
If we wish to fulfil Charles Darwin's ambition of understanding the reasons for the distribution of differences between the sexes, we shall need to examine the causes of variation in lifetime breeding success among males and females in natural populations.
RULES FOR CHANGING THE RULES PATRICK BATESON
Few biologists have not at one time or another marvelled at the exquisite fit that can be found between the characteristics of an organism and the characteristics of its environment.
Darwin was not the first to marvel but he made the notion of such adaptedness scientifically respectable by providing an explanation of how it might have come about.
Nowadays, most discussion about adaptation assumes that when it is found, a fit between organism and environment is the result of evolutionary selection pressures and nothing else (e.g. Lewontin, 1978).
Furthermore, each adaptation is supposedly transmitted from one generation to the next by genetic means alone.
However, these assumptions are manifestly false when applied to behaviour  particularly the behaviour of complex animals (e.g. Lorenz, 1965; Hinde, 1968).
Quite obviously a great many animals are able to tune their behaviour to their environments by learning and by other developmental mechanisms which rely on external triggering.
When faced with an example of adaptedness in behaviour, at least three explanations can be offered for the process of adaptation.
Consider an actual case, the way in which a long-tailed tit makes a strong, cryptic and well-insulated place in which to lay eggs.
Once a site in a bush or tree has been selected, the pair of long-ta fled tits search for moss and bring it back to the site.
When some moss has stuck, each tit collects spiders' webs and stretches them across the moss.
More moss is collected and then more spiders' webs until a platform has been formed.
The bird that is building can now place moss and webs around itself  building up the sides of the nest.
When the nest-cup is well formed the bird fetches lichen and weaves this on to the outside of the nest.
Building up the sides of the nest is resumed but is periodically interrupted so that more lichen can be added to the outside.
Eventually, the bird builds the walls up and over itself to form a dome, but leaves a neat entrance hole at the side.
Finally, the nest is lined with a large number of feathers (Tinbergen, 1953, cited in Thorpe, 1956).
In principle, the behaviour of the tits could be adapted to the job of building a safe, warm nest for offspring in three separate ways.
First, birds performing the appropriate actions could have had more surviving offspring than those making less good nests; consequently, in the course of time, genes necessary for the expression of the appropriate actions spread through the long-tailed tit population.
Secondly, the bird could copy what another more experienced bird had done; the process of selecting the actions best adapted to the environment had gone on in previous generations and been transmitted socially.
Thirdly, by experimenting on its own with different materials and different actions, each bird could assemble the appropriate repertoire for building nests.
The three processes of adaptation and the three sources of adaptedness for an individual are shown in Table 24.1.
All three processes could contribute to the adaptedness of the nest-building.
We should not expect three classes of behaviour corresponding to the three processes of adaptation.
Furthermore, we should not be surprised by the extent of learning and imitation that can be found, particularly in complex animals.
Cultural transmission of adapted behaviour is by no means confined to humans (see Galef, 1976; Bonner, 1980).
One of the first examples to be discovered in animals was the opening of milk bottles in Britain by great tits, blue tits and coal tits.
The spread of the habit from a few scattered locations before 1930 to a great many in 1947 was well documented (Fisher &amp; Hinde, 1949; Hinde &amp; Fisher, 1951).
The animal which is learning does not operate like an idiot photographer attempting to take a snap of everything.
What an animal learns is highly selective and highly ordered.
The instruments for changing behaviour show an adapted regularity which suggests that they themselves have been subject to natural selection during the course of evolution.
With precisely this point in mind, Konrad Lorenz (1965) referred to the ' innate school marm ' who, he imagined, was busy directing the course of learning.
It is worth noting that what Lorenz meant by the memorable phrase was not that there were unlearned instructions for learning, but the instructions were adapted for their present use by natural selection.
The importance of this distinction will become apparent later.
It is more usual nowadays to use computer metaphors and refer to the programming of learning (e.g. Pulliam &amp; Dunford, 1980).
I mildly distrust such metaphors.
It can be easily assumed that because we know how computers work we therefore know how learning is programmed.
My distrust turns into hostility when phrases like ' genetically programmed ' are used instead.
Such phrases confuse the way in which coded information is transmitted from one generation to another with the regularities of a nervous system, which itself is the outcome of an ordered developmental process.
For these reasons I prefer to use the phrase ' rules for changing the rules' (Bateson, 1976).
A necessary preliminary is to clarify the meaning of the term ' rule '.
Behaviour is very far from being disorderly.
It may be complex but it is certainly not chaotic.
Something provides direction and keeps it in order.
Something is responsible for the regularity.
' Rules' refer to these consistencies of behaviour.
In more complex animals, consistencies are not easily found on the surface and many of us feel the need to postulate structural regularity beneath the surface if we are to make sense of what we see.
A famous example of this theoretical approach is Chomsky's (1965) analysis of language in terms of underlying grammatical rules.
Not everybody likes this style and, among psychologists, Skinner (1959) in particular has derided the use of concepts based on inferred structures and processes.
It must be admitted that concepts referring to unseen processes tend to acquire additional meanings that are not suggested by the evidence they are intended to explain (see MacCorquodale &amp; Meehl, 1948).
Muddle ensues when the rules are treated as though they are tangible and can be observed directly.
But the confusion can be avoided if we treat unseen high-order rules for what they are, namely, as explanatory devices.
When that is done the thinking can be creative and rewarding.
The usage of ' rule ' by biologists is clearly different from that employed by social scientists when they talk about verbally transmitted instructions for what humans may and may not do.
It would obviously avoid punning and confusion if different terms were used.
However, it is unlikely that biologists will be misunderstood when they apply the term to animals.
In this chapter I first consider the underlying rules for associative learning and suggest that some useful general principles have already been uncovered.
I go on to argue that, despite the underlying regularities, the behaviour of an individual animal is only predictable when a lot is known about the conditions in which the animal has grown up.
In order to emphasize the importance of this point, I devote the rest of the chapter to a discussion of the developmental rules that may influence mate choice in humans.
A growing understanding of such rules has tempted biologists into making exaggerated claims about the invariance of human mating preferences and also about the origins of incest taboos.
I argue that even if only one process were involved, the outcome would depend on conditions and, since conditions vary, so must the behavioural outcome.
For all that, I conclude that the postulated underlying rules for development may usefully account for some of the variation in human sexual behaviour and possibly even the variation in marriage laws.
Rules for associative learning
Any animal with even a rudimentary nervous system will be better placed if it can compute the arrival of impending danger or the location of valuable resources such as food or mates.
Its nervous system does not have to be modifiable in order to work with reasonable efficiency in this way.
Even so, the power to predict and control the environment is enormously enhanced by a capacity to associate neutral events with those that already have some importance for the animal.
With such capacity, initially meaningless cues and initially haphazard or exploratory acts can acquire causal significance.
What could be the rules for the necessary associative learning processes?
One very obvious possibility would be a time-window preceding the important event.
If a neutral event occurs within this time-window then it loses its neutrality.
To give a text-book example, college students were trained in a situation in which a buzzer was sounded before, after or together with the delivery of a mild electric shock to the finger (Spooner &amp; Kellogg, 1947).
Periodically, the students' responses to the buzz alone were tested.
The students who, in the training trials, heard the buzz half a second before the shock, jerked their finger back more consistently than those who had had a longer gap between the buzz and the shock during training, and they did so much more markedly than those who heard the buzz at the time of the shock or after it (see Fig. 24.1).
The brief interval allowed for the establishment of a link between neutral and significant events was at one time elevated into a general law of associative learning.
Equipped with this rule and with knowledge of what are important events, the animal seems to be well set up to acquire the ability to use initially meaningless environmental cues as predictors of what will happen, and initially haphazard acts as instruments for controlling the environment.
While it might seem to make good intuitive sense that a time-window should be small, substantial delay in detectable effect can sometimes follow the performance of an activity.
If you eat some contaminated food, you will not necessarily feel the ill-effects immediately.
Indeed, it is well known now that many mammals and birds can develop aversions to novel foods that were followed by ill-effects hours after ingestion (reviewed by Domjan, 1980).
Experiments involve a spurious association between the novel food and the illness which is usually induced chemically or by X-rays.
Nevertheless, the animals subsequently avoid the novel food.
They do not avoid familiar food which has similarly been followed by illness, and only certain cues such as smells and tastes associated with the novel food are attended to.
Others, such as noise, are treated as being irrelevant (see Revusky, 1971; LoLordo, 1979).
The implication is that the animal is able to classify neutral events prior to learning and has a rule for what classes are relevant to particular outcomes.
The phenomenon of modifiable taste aversion is often taken as one of the prime pieces of evidence for doubting general principles of associative learning.
It has led some people to argue that the only sensible way to study learning is by examining it in the ecological conditions to which it is adapted (Johnston, 1981).
However, common features can be found, whether a rat is learning to avoid poison or shock.
If two neutral events of the same class are used, the second one interferes with learning about the first.
For instance, rats were given novel saccharin solution and 1 5 min later were given novel vinegar solution, and finally they were made ill with lithium chloride.
Subsequently, they were much less likely to avoid the saccharin solution than rats which had been given water instead of vinegar (Revusky, 1971).
The vinegar had over-shadowed the saccharin.
The time-window idea probably has to be retained in a watered-down form because if an animal is given novel food followed by weeks of familiar food and finally made sick, it is unlikely that it will avoid novel food.
Nonetheless, the notion of a time-window is not sufficient to account for what is found.
The rule would seem to need a triple condition attached to it.
If a neutral event has occurred within a certain time of an already important event, if it was of a certain category and if another of the same category had not been interposed between it and the important event, then that event itself acquires significance for the animal.
At this point it would be fair to ask: what has all the work on the avoidance of shock and poisons got to do with social behaviour?
The answer is that associative aspects of learning enable the individual to cope not only with its physical environment but also with its social environment.
Humphrey (1976) has argued convincingly that animals are in many ways over-equipped for the inanimate environment, but the environment provided by other animals (particularly clever ones) is especially complex, difficult to predict and difficult to control.
Predators and prey have to be coped with one way or the other, but strong pressures also come from social companions.
Social groups are clearly not just bands of competitors and individuals often need each other for survival.
Nevertheless, group members also have to compete with each other for many necessary resources.
In such competitions, it is not simply the one who is strongest who wins.
It can frequently be the one with the best abilities to make complex calculations about what the others are up to.
In terms of predicting and controlling the social environment, high technology can quite clearly be every bit as important as brute force.
It follows that in complex animals in particular, the rules for learning the rules can be as much to do with social interaction as anything else.
Development of rules
If we are right in our inferences about the rules for associative learning, they clearly do seem to have adapted qualities.
They fit the animal's information-gathering equipment to particular problems and, presumably, they have been subject to natural selection during evolution.
Therefore, they must be transmitted in some way, usually genetically, from one generation to the next.
At this point, though, we should not forget the hard-learned lesson that an evolutionary argument is not the same as a developmental one.
The rules for modifying behaviour do not spring fully armed out of the genome.
They themselves have to develop and, clearly, they represent the workings of an already functional nervous system and body.
The extent to which their development involves various kinds of experience raises an entirely separate issue.
As a matter of fact we know that, at least in complicated animals, many features of the rules are profoundly modified by experience.
I discovered this painfully myself when I went as a visitor to a beautifully equipped laboratory to work on learning in rhesus monkeys.
The laboratory had some elegant computer-controlled apparatus for teaching the monkeys to discriminate between visual forms such as letters.
If the monkey pressed the correct letter it was rewarded with a peanut by a mechanical dispenser which was specially designed for this kind of food.
Everything was perfect except that when I came to train experimentally naive monkeys, I discovered that they did not like peanuts.
The monkeys had to be deprived of their regular food and accustomed to the peanuts for weeks before they would take them with any readiness, let alone treat the nuts as rewards for appropriate behaviour.
In this case, which is not exceptional (see Weiskrantz &amp; Cowey, 1963), experience expanded what the monkeys regarded as acceptable food, and at an earlier stage in development experience had narrowed the range.
It could be argued that in such instances an unlearnt program could still be detected at work behind the scenes since the general category of food, and its effectiveness as a reward, was in some sense built in.
In other cases though, it becomes more difficult to pinpoint what might or might not act as a reward without very extensive knowledge of the animal's previous experience.
For instance, the conditions in which it becomes possible for an animal to perform an act that would bring it food become rewarding themselves.
So the animal will work in order to provide itself with those conditions.
In this way lengthy chains of behaviour can be developed with any one event providing the terminating condition for one action and the enabling condition for the next (see Kelleher, 1966).
This is the basis for many complex circus acts performed by animals.
Beyond this, the knowledge of the ways in which initially neutral cues are treated as potentially relevant or ignored is growing, and suggests once again that the rules for learning can be influenced by the nature of prior experience (e.g. Dickinson, 1980).
Presumably, if the rules for learning are to have any universality in natural conditions, the experience which affects them must be a common feature of all the animals having the rules  in other words, the variance due to the environment is normally small in the environment to which the animal is adapted.
Also, when considering development, it must be stressed that we do not have to depend on an infinite regress.
Quite clearly learning does not have to be involved in the development of the rules for learning.
For instance, even very young rats selectively associate taste with poison and texture with electric shock (Gemberling, Domjan &amp; Amsel, 1980; Domjan, 1980).
So, in this case, it looks as though the rules for forming some associations, but not others, are not dependent on learning for their development.
But that should not make anybody complacent about the developmental processes.
The classic mistake has been to confuse experiences involved in learning with all other kinds of experience which the animal can have during its development (see Lehrman, 1970).
All sorts of environmental conditions can have non-specific but profound influences on behavioural development without involving learning (see Bateson, 1981).
Change the social or physical conditions in which the animal is growing up and you may find it ends up with a different set of rules for learning.
It will seem obvious, I hope, that a rule for learning or for any other kind of developmental process, is not a gene written large.
We have no reason to suppose that there is any simple correspondence between gene and rule for changing behaviour any more than there is an isomorphic relationship between gene and behaviour.
So why make an issue out of it?
The reason is that some of the influential popularizers of modern evolutionary theory still manage to confuse the developmental issue with the evolutionary one.
Because gene frequencies are generally presumed to change in phylogeny, then it is suggested that genes must be doing the real work in ontogeny.
As a result of this category mistake, total non-communication has occurred between the sociobiologists and their critics (see Bateson, 1982b).
The distinct issues are relatively easy to sort out in the case of associative learning but, as will become apparent in the remainder of this chapter, they are all too easily muddled when functional approaches are brought to bear on development problems.
A functional approach to morality
I am going to consider now a famous case in which the supposed regularities of human morality are attributed to the workings of adaptive rules, so providing an evolutionary explanation for part of human culture.
The evolutionary costs in this case are those due to inbreeding, and the cultural outcome is the incest taboo.
Two quite distinct arguments are mounted.
Since these are sometimes confusingly conflated, it is helpful to keep them separate even though they are not mutually exclusive (see Fig. 24.2).
The first argument is the classical one and runs as follows.
Human beings, being observant and intelligent, spot the consequences of matings between close relatives and make safety laws about them.
The incest taboo is equivalent to a legal requirement to wear seat belts or crash helmets.
It is sometimes claimed that people in many cultures are aware of the ill-effects of inbreeding (Lindzey, 1967), but nobody, as far as I know, has claimed that such knowledge is universal.
Certainly in modern statistical studies, differences between children of inbred and outbred marriages can be scarcely detectable, particularly when the inbred marriages occur in communities where spouses are traditionally at least first-cousins (e.g. Rao &amp; Inbaraj, 1977).
Even in outbred communities it would be very difficult to detect inbreeding costs when infant mortality was high and its causes and those of deformities in offspring were numerous and varied.
The second argument about the origins of the incest taboo is the one that relates to the major theme of this chapter and I shall consider it at some length.
I should emphasize first, that the two possibilities shown in Fig. 24.2 are not the only explanations for the origins of the incest taboo.
Most anthropologists would prefer to look elsewhere, partly I suspect because of the way the biological arguments have been overstated.
The biological part of the second argument shown in Fig. 24.2 is well-trodden ground and has been reviewed in numerous places recently (e.g. Alexander, 1980; Fox, 1980; Thiessen &amp; Gregg, 1980; Bixler, 1981).
Westermarck (1891) believed that satisfying sexual relationships are not formed between people who have spent their childhood together.
This view is supported by the behaviour of members of Israeli kibbutzim who very rarely marry the people they have grown up with (Spiro, 1958; Talmon, 1964; Shepher, 1971).
Other evidence in support of the Westermarck hypothesis comes from the work of Wolf and his colleagues, summarized in a recent book (Wolf &amp; Huang, 1980).
They have analysed a form of arranged marriage which was practised in Taiwan.
The wife-to-be was adopted into the family of the husband-to-be when she was a young girl.
The marriage was formalized and consummated when the partners were adolescent.
This form of arranged marriage, the ' minor marriage ', could be compared with a more common form of arranged marriage, the ' major marriage ', in which the partners met each other for the first time when they were adolescents.
In a great many respects the minor marriages were less successful than the major marriages.
They generated fewer children, the rates of infidelity were higher, and so forth.
For instance, 15% of the 1117 minor marriages ended in divorce whereas only 6% of the 1651 major marriages did so.
On a note of caution, it should be pointed out that the people involved in minor marriages were considerably younger at the time of the formalization of the marriage than those involved in the major marriages.
Also a minor marriage was a much cheaper option for the parents than a major marriage and was, therefore, considered to be socially disgraceful.
These factors may have contributed markedly to the relative lack of success of the minor marriages.
Another independent piece of evidence corroborating the Westermarck hypothesis is provided by stable incestuous relationships (Weinberg, 1956).
Weinberg found that whereas most incestuous relationships were unstable and short-lived, six that he examined involved strong and lasting attachments between the partners.
In each case the siblings concerned had been separated from each other when they were babies.
Three criticisms are commonly directed at the evidence for the Westermarck hypothesis.
First, it is pointed out that a most preferred sexual partner is not necessarily a spouse (e.g. Solomon, 1978); however, it is difficult to see how this perfectly valid point is relevant to the evidence given above.
Secondly, overt sexuality is found between siblings (Finkelhor, 1980) and among kibbutz members of the same age (Spiro, 1958; Kaffmann, 1977).
Finally, despite supposed indifference to familiar members of the opposite sex, incest does occur quite frequently (Livingstone, 1980).
I shall not ignore these criticisms, but at this point I think it is helpful to consider the data from animals.
Apart from strengthening the view that reduced sexual responsiveness to familiar members of the opposite sex is quite widespread in birds and mammals, the animal studies also suggest ways of dealing with criticisms of the human evidence.
They also point to other sources of variation in mate choice.
Imprinting and discrepancy hypothesis
The effect of early experience on the mating preferences of birds and mammals has been known for a long time.
The effect of imprinting on the sexual preferences of birds was made famous many years ago by Konrad Lorenz (1935).
Numerous quantitative studies have been done on both birds and mammals in the last twenty years (reviews in Immelmann, 1972; Bateson, 1978a) and have shown that early experience can have profound and lasting effects on sexual preferences.
Admittedly, imprinting was usually thought of as the process by which animals normally learn about the characteristics of their species, even though a reference was occasionally made to ' asexual imprinting ' (Aberle et al.,
1963) and a number of studies were done on the reduced sexual responsiveness to familiar members of the opposite sex in rodents (reviews in Dewsbury, 1982; D'Udine &amp; Alleva, 1983).
In general, though, thinking was retarded by the dichotomous classifications that were in use at the time.
Assortative mating was either positive or negative.
Animals, like humans, were either endogamous or exogamous, they preferred the familiar or they preferred the novel.
It is possible to break out of the straight-jacket by applying an idea that had been used for many years in thinking about the psychology of classification and aesthetics (McLlelland &amp; Clark, 1933; Berlyne, 1960).
This is known as the ' discrepancy hypothesis'.
In general, what people find most stimulating and most attractive is a bit different but not too different from what they know already.
As is shown in Fig. 24.3, it was a simple step to translate this into a hypothesis about the effects of early experience on mating preferences (Bischof, 1972; Bateson, 1978a).
Experimental studies on both birds and mammals followed quickly (Bateson, 1978b; Gilder &amp; Slater, 1978; McGregor &amp; Krebs, 1982).
However, it was not all plain sailing.
When they were given a choice between a familiar and a novel member of the opposite sex, birds might actually choose the familiar even when the novel really did not look so very different to our eyes (Miller, 1979; Bateson, 1980; Slater &amp; Clements, 1981).
The results suggested that the birds might have sharply tuned preferences only slightly displaced away from siblings when normally reared.
The difficulty is that, if the most preferred mate is slightly different from a familiar member of the opposite sex and if we do not know how to measure the difference, we can unwittingly present the animal with a novel object which is less attractive than the familiar.
It might seem as though the hypothesis is so slippery that it can not be falsified.
Indeed, this point has been used as a general criticism of the discrepancy hypothesis (Thomas, 1971).
However, the issue is settled by positive evidence not by ingenious explanation of the failure to confirm the idea.
If members of the opposite sex were graded along a continuum from familiar to very novel and the animals were allowed to choose between all the possibilities, then progress can be made.
While we do not yet know what cues the animals might use, we can exploit the likelihood that, when other qualities such as physical well-being are equal, an optimal choice of mate is likely to be one that minimizes the costs of both inbreeding and outbreeding (Bateson, 1980, 1983).
In other words, the genetic relatedness of the partner is likely to be important.
It is a relatively easy matter, when we know the pedigrees of animals, to arrange choices between members of the opposite sex of different degrees of relatedness.
I have done this with Japanese quail.
Birds that had been reared with siblings were tested in apparatus that allowed them to be given up to six alternatives (Bateson, 1982a).
In one experiment birds were given choices between members of the opposite sex that were either familiar siblings, novel siblings, novel first-cousins, novel third-cousins, or novel unrelated individuals.
Fig. 24.4 shows the mean percentage durations spent in front of each category of stimulus bird by both adult males and females.
The time spent near novel first-cousins was significantly greater than the time spent near both the familiar and novel siblings and novel unrelated individuals.
Despite the clear overall preference for first-cousins, the data were highly variable.
The variability was expected as the degree of relationship between two individuals only indicates the probability that the two share heritable characters.
Knowledge by a bird of a sibling's appearance must necessarily be an imperfect guide to what a first-cousin will look like.
Of course, remaining near a member of the opposite sex is not the same as mating with it.
However, other experiments have shown that, in adult male Japanese quail, the time spent near a female in a choice test is strongly linked to the copulation preference (Bateson, 1978b).
Furthermore, the males are observed to court the females in the choice tests.
Finally, I had found in other experiments that birds show no consistent preferences for members of the same sex.
The results with quail provide direct support for the discrepancy hypothesis as applied to sexual preferences and indirect support for the notion of optimal outbreeding.
I have written elsewhere about the evolutionary pressures which might generate a balance between inbreeding and outbreeding (Bateson, 1983).
However, it is worth emphasizing here that at least four costs have been proposed for inbreeding, and at least seven for outbreeding.
The costs of outbreeding may include the risks of infections from pathogens carried by the partner and the breaking up in the offspring of co-adapted complexes of genes found in the parents.
Not all costs can apply to all species and some naturally outbreeding species may use other mechanisms for avoiding the costs of inbreeding, such as dispersal away from the natal area by one sex (reviewed by Greenwood, 1980).
Even in those species that choose a mate that is a bit different but not too different from close kin, other factors are also important.
Qualities such as the physical condition of the member of the opposite sex, the resources it holds and the extent to which it bears characters that have been subject to sexual selection can all affect whether or not it is chosen (see Halliday, 1983).
It would be quite wrong to suggest that the only influence on mate choice is relative familiarity.
Despite the variety and the complexity, the animal work indicates first and foremost that mate choice can be profoundly influenced by early experience.
Secondly, in some species the choice is remarkably finely tuned so that under certain circumstances familiarity may be preferred over novelty.
The fine tuning might be achieved by employing two well-known mechanisms as shown in Fig. 24.5.
Filial imprinting is known to restrict preferences to the familiar (see Bateson, 1979), and sexual imprinting could operate in exactly the same way.
Habituation, by contrast, reduces responsiveness to the familiar.
The net effect of superimposing habituation on imprinting would be to displace the preference away from the familiar.
The combination of the two learning processes could produce a sharply peaked preference for something a bit different from the familiar when other things are equal.
Mating preferences in humans
The animal evidence enriches the discussion of human mating preferences in several important ways.
First, by emphasizing that preferences are displaced away somewhat from the familiar, it is possible to explain two facets of the data from humans that would otherwise have seemed incompatible.
Apart from the evidence of reduced sexual interest in familiar members of the opposite sex, which I have already mentioned, the great mass of data shows that freely chosen human spouses are more like each other than would be expected on a chance basis.
Similarities are not only social and psychological, but also found in measures of body dimensions such as length of earlobe (e.g. Eckland, 1968; Lewis, 1975; Thiessen &amp; Gregg, 1980).
A second point is that the method of testing choices draws attention to the relative nature of a measured preference.
It will rarely be the case that either an animal or a human will be provided with the opportunity to mate with an absolutely ideal member of the opposite sex.
Furthermore, the best available mating may be with a sibling or an offspring on certain occasions.
Sexual responsiveness to a familiar member of the opposite sex may not be zero  particularly when the time allotted to searching for an alternative has run out.
The conclusion is, therefore, that when an individual has no choice or an impoverished set of choices, he or she may inbreed.
Finally, the precocious sexual behaviour, which is often observed between siblings and was the basis for Freud's (1950) thinking about the development of sexual preferences, may play a role.
If habituation is involved in displacing preferences away from the familiar to individuals that are slightly different, then the learning process may be facilitated by the performance of precocious sexual behaviour which is common enough in humans (Finkelhor, 1980) as it is in other animals.
I am not convinced that overt sexual behaviour is essential for the development of indifference, even though it may help.
But I think it is highly misleading to suggest, as Shepher (1971) has done, that development of sexual preferences is complete by the age of six in humans.
He based this conclusion on a very few individuals who married within their peer group in the kibbutz and were found to have entered the kibbutz during their childhood and usually after the age of six.
To demonstrate a sensitive period of the type he was proposing, it would be necessary to show that adults who had left a kibbutz at the age of six were not sexually attracted by members of the opposite sex whom they had been reared with while still in the kibbutz.
As things stand, the existing evidence has been wildly over-interpreted both by Shepher and by others who have uncritically accepted his conclusions (e.g. Lumsden &amp; Wilson, 1981; van den Berghe, 1982).
If we reject the naive application of the sensitive period concept and accept that familiarity of a certain kind does reduce sexual attractiveness, then it may be possible to reconcile the thinking of Freud with that of Westermarck.
The sexual attraction of their siblings and parents, which people under psychoanalysis reported they felt, may have created the conditions for developing subsequent indifference.
This line of thought might also be applied rewardingly to explain one striking feature of divorce statistics.
For instance, in British women who married before the age of 20, the proportion of marriages that ended in divorce has been approximately double that of the marriages of women who married between 20 and 24 (Office of Population Censuses and Surveys, 1978).
This has been true at any time between four and 25 years after marriage.
Many factors, such as differences between social classes in attitudes to marriage, could explain or contribute to explaining the difference.
Clever research design could sort out some of the confounded variables, so I shall add to the possibilities a speculation arising from my point about habituation.
Early marriages may involve a great deal of intimacy but relatively little sexual satisfaction.
Indeed, people often report that their early sex lives were relatively unrewarding.
If the effects of habituation are not powerfully offset by rewarding sexual experience, the partner may lose his or her attractiveness and become the equivalent of a sibling.
While a great deal is stili unknown about the development of sexual preferences in both animals and humans, the similarities are quite striking.
In my view a good case has been made for the view that the learning processes involved in the formation of mating preference of humans have been subject to natural selection during the course of evolution.
However, acceptance of this point has to be tempered by an awareness that mate choice is influenced by many qualities.
Early experience with particular individuals is not the only source of variation in adults' mating preferences.
So even if Westermarck was right, as I believe he was, it would be extremely surprising if his hypothesis explained all of what humans do.
Incest taboos and marriage laws
In a society in which spouses are freely chosen, it is easy to confuse the influences on sexual preference with those on marriage.
The anthropologists have to point again and again to the great many societies in which spouses are arranged and not freely chosen.
A biologist who was so minded could counter by arguing that, if by and large the marriages generate the children, then marriage laws could still have been influenced by evolutionary pressures.
While that argument would beg an important question, on the face of it the marriage laws do seem to promote the function of optimal outbreeding.
Marriage with close kin is generally forbidden in most societies and so, commonly, is marriage with people of dissimilar culture.
A general bias in favour of spouses who come from nearby  both spatially and socially  has often been noted by anthropologists (Fortes, 1962).
Charles Darwin married a cousin and, indeed, such marriages were quite common in nineteenth-century Europe.
It must be said that the apparently convincing evidence has been processed into non-significance by elegant mathematical analysis (Hajnal, 1963).
This seems to have been done in the interests of retaining the theoreticians' assumption of panmixia (Charlesworth, 1980).
But maybe Darwin knew better, since the assumption of random mating could hardly apply in those numerous societies that actually favour first-cousin marriages (Murdock, 1967).
Is there some functional similarity with mate choice in quail?
Two important and unresolved problems are raised by this line of thought.
How does an inhibition get translated into a prohibition?
And how much of the variation in the prohibition is explained by the character of the inhibitions?
Incest taboos take many different shapes and forms.
They sometimes include certain types of cousin and people related by marriage only.
Prohibitions on sexual relations with parents, siblings or children, are nearly always universal  but not quite.
Hopkins (1980) has analysed the marriage records of Roman Egypt in which the census data were especially complete.
He confirmed the view that among perfectly ordinary people, who were neither Pharaohs nor priests, full brother-sister marriages occurred in a minimum of nine out of the 113 marriages he analysed, If the less certain cases are also included along with marriages between half-siblings, the proportion of incestuous marriages was of the order of 20%.
Whatever view one takes of the origins of the incest taboos, it would be intellectually shoddy simply to ignore these data.
Can we find a common underlying principle that explains the variation?
Levi-Strauss (1969) proposed that, women being the most important resource that men have, a system for exchanging women always underlies the social control of marriage.
His arguments have something of the character of the Ptolemaic theory of the universe  brilliant, logical, grand, but despite all these things, extremely complex.
His theory is in stark contrast to the biologists' attempts to find a relationship between the prohibitions on certain types of marriage partner, and the inhibitions about having sexual contact with such classes of people.
These ideas are simple.
So simple, indeed, that usually the argument is not stated at all.
The correlation between prohibition and inhibition is offered as though it explained everything, or at best, the prohibition is held to have arisen by ' myth making ' or ' ritualisation ' (Bischof, 1972).
As Williams (1978) pointed out, such statements are nothing more than promissory notes.
They do not provide an explanation of how one turns into the other.
In a recent book, Lumsden &amp; Wilson (1981) attempted to deal with the issue mathematically.
However, their efforts have not provided the explanation that we need.
They merely assumed that inhibition generates the incest taboo without pointing to any behavioural mechanism that could translate one into the other.
In lieu of anything better I shall make a suggestion.
Prohibitions may have arisen from the social pressure directed against unorthodox behaviour.
What is normal behaviour is itself influenced by the pattern of early experiences which are common to that society.
The implication is, therefore, that there will be some correlation between child-rearing practices and taboos (see Fig. 24.6).
People often strongly disapprove of others who behave in unusual ways.
The most obvious example is the moral repugnance that many people show for homosexuality between consenting adults.
Why should they mind?
They are not harmed by the homosexuality.
But the conventional response is nonetheless a violent one  in some societies homosexuality may be punished by death.
If fear of nonconformity and the unusual has driven the cultural evolution of incest taboos, then a comparable argument should apply to taboos on marriages with strangers or members of other castes and races.
Levi-Strauss (1969) noted that such taboos certainly exist and a notorious modern example of it is found in the immorality laws of South Africa which forbid sexual relations between blacks and whites.
If this approach is anywhere near correct, we should expect a correlation between the class of people who are prohibited as sexual partners and the likelihood that they will be familiar or extremely novel (see Fig. 24.6).
Obviously many things might muddy the correlation between social structure and prohibited partners, but one piece of existing evidence points in the right direction.
While first-cousins are often favoured as marriage partners, a distinction is very often made between parallel and cross-cousins.
The sibling parents of parallel cousins are the same sex and the sibling parents of cross-cousins are of the opposite sex.
Sometimes parallel cousins are forbidden as spouses and cross-cousins are favoured.
Alexander (1980) has gone through Murdock's (1967) ethnographic atlas and found that this asymmetrical treatment of cousins is strongly associated with the type of marriage common in that culture.
The results are shown in Table 24.2.
Alexander argues that because brothers may share wives in polygynous societies, parallel cousins may in fact be half-siblings.
He seems, therefore, to be using the evidence as an ingenious updating of the classical safety law argument.
However, parallel cousins who may in reality be half-siblings are also likely to have lived in the same household.
Once again, genetic relatedness is confounded with familiarity.
So, another possible explanation is that in a polygynous society, the parallel cousins will be much more likely to grow up together than the cross-cousins.
In effect, the parallel cousins are as familiar as siblings.
It does not follow from the argument I have mounted here that the conformism generating prohibitions is an adaptive response that evolved in the service of maintaining optimal outbreeding.
The conformism might have arisen for quite different reasons and among its other consequences happened incidentally to amplify the beneficial effects of the inhibitions.
Certainly, it would be difficult to argue that all the variation in human marriage laws could be explained in terms of their evolutionary benefits.
The brother-sister marriages of Roman Egypt probably had a great deal to do with the preservation of property and nothing to do with the preservation of genes (Hopkins, 1980).
And similar explanations can account for a lot of the variation found in human societies (Goody, 1976).
It would be absurd to adopt a rigidly determinist view of what has gone on in the formation of culturally transmitted marriage laws.
To say the least, it is unfortunate that a multiply influenced process with many stages in it should be thought of by sociobiological proponents and their critics alike (e.g. Solomon, 1978) as having an invariant outcome and a single explanation.
It is a bit like arguing that if the hypothesis that smoking causes lung cancer is to be believed, everybody who smokes must get cancer; what is more nothing else (such as asbestos) can be admitted as having the same effect.
The opposed parties in such disputes, evidently believing that causality has a chain-like character to it, have an impoverished notion of how things actually work.
In most complex systems the sources of variation are likely to be numerous.
It follows that alternative explanations do not have to be mutually exclusive.
The best that can be said about marriage laws is that some of the variation may be explained along the lines proposed here.
Even within this explanatory framework, little useful understanding will be obtained without first studying the problem at many different levels.
What is needed, therefore, is constructive collaboration between biologists and social scientists and a proper respect for the insights that the different disciplines can provide.
Conclusion
The arguments presented in this chapter do not lead to comfortable conclusions that can be instantly assimilated.
However, they are not, I trust, obscurantist and a positive point does emerge.
Regularities can be found in the way that behaviour is tuned to the environment during development.
To be effective, though, the rules have a conditional character to them which means, of course, that they generate variation in behaviour in response to variable environments.
It is not inexplicable variation, but it is variation nonetheless.
The old reductionist's vision was that one day when we knew enough about genes we would be able to predict every detail of every adult's behaviour.
It always was a pipe-dream.
Imagine civil servants working in a capital city and trying to make all the decisions required for running a large country.
They Simply would not have the flexibility or the speed of reaction to cope with the complexities of everyday life.
And so it is with the genes, the natural bureaucrats.
On their own they are too clumsy in their form of regulation to provide the necessary adaptations, especially those required for social living.
The genes had to delegate control.
When we examine animals with nervous systems that were built with conditional rules for dealing with the external environment, the business of predicting how they will respond on the basis of knowing how they were made becomes impossible.
It is like trying to predict the outcome of a game of chess before anyone has made a move.
What we can do is attempt to get hold of the rules of the game so that we can make sense of a game as it is played.
At that stage I concede happily that we may be able to predict what a clever animal will do in a particular set of circumstances.
In the meantime, we should expect to be surprised very often. [C]
SOCIOBIOLOGY AND THE DARWINIAN APPROACH TO MIND AND CULTURE EDWARD O. WILSON
On 3 October 1838, Charles Darwin wrote in his N notebook that ' to study Metaphysics, as they have always been studied appears to me to be like puzzling at astronomy without mechanics...
Experience shows that the problem of the mind can not be solved by attacking the citadel itself... the mind is function of body... we must bring some stable foundation to argue from... ' (in Barrett, 1980).
Although Darwin had turned in the right direction, he could do very little with mind and culture during his lifetime for the same reason that he was helpless before the mysteries of heredity: the basic information and modes of thought were lacking to produce the stable foundation which he correctly viewed as essential.
Today, one hundred years after his death, we may at last be approaching a sufficient understanding to bring to fruition Darwin's proposal.
If so, we can verify another insight, which was entered into the M notebook on 16 August 1838: ' Origin of man now proved...
Metaphysics must flourish...
He who understand baboon would do more toward metaphysics than Locke. '
What will be the outcome of this most problematic, controversial extension of evolutionary theory?
It is a common perception that during the 1950s biology replaced physics as the most exciting domain of science.
I will make a brash prediction: that by the year 2000 the social sciences, in conjunction with brain studies, will commence to replace biology in the central role.
If such an advance is realized, it represents one more step in a progression in which the antidiscipline, that is, the field treating the next level of organization below the one under scrutiny, is partly replaced by the synthetic enterprise to which it gave rigour and impetus (Wilson, 1977).
In other words, biology advances the social sciences.
Just as physics and chemistry helped to modernize biology and moved it to centre stage during the past thirty years, I believe that biology is about to augment the social sciences greatly and move them to centre stage.
The principal remaining obstacle in this enterprise is the unknown relation between genes and culture.
But of course when one uses the phrase ' an unknown relation ', he means that it is a puzzle to be solved.
In this case the problem is surely one of the most important in all of science, not to mention philosophy, as Darwin perceived 144 years ago.
A few writers still speak of a permanent discontinuity between the biological sciences and the social sciences, grounded in epistemology (Eccles, 1980) or at least forced by a fundamental difference in goals (Hampshire, 1978).
But others, principally in cognitive science and evolutionary biology, have come to see the gap as a largely unknown evolutionary process, a complicated and fascinating interaction in which culture is generated by biological process while biological traits are simultaneously altered by genetic evolution in response to cultural innovation.
Charles J. Lumsden and I have recently studied this dual evolutionary process, which we call gene-culture coevolution (Lumsden &amp; Wilson, 1981).
We have attempted to align a previously independent field of inquiry, cognitive and developmental psychology, with evolutionary biology and particularly sociobiology, and in so doing have constructed an ensemble of models that trace, at times clumsily and imperfectly, behavioural development from the genetic blueprint to the assembly of the nervous system to the learning process  and then back down to the alteration of gene frequencies by natural selection operating within the context of particular cultures.
The full sequence covered by our models is referred to as the circuit of gene-culture coevolution.
This research is a logical extension of sociobiology, the systematic study of the biological basis of social behaviour and a growing division of evolutionary biology.
The particular problem addressed by the theory is the following.
We know that human social behaviour is extremely variable.
It is also open-ended, in the sense of being always subject to rapid change due to innovation and importation.
Cultural evolution is often characterized as Lamarckian in quality, in other words, dependent on the transmission of acquired characters, and relatively fast; while genetic evolution is Darwinian, that is, dependent on changes in gene frequencies across generations, and slow.
But exactly how are these two processes coupled?
The solution to the problem can be found by shifting emphasis from the terminal products, the genetic blueprint and the final cultural product, and concentrating on the developmental procedures that connect them.
The reason such analysis has not proceeded more vigorously in the past is that evolutionary biologists have virtually ignored developmental psychology, now a vast field in its own right, while psychologists for their part have not appreciated the great potential of evolutionary theory for their own studies.
p 547: FIGURE
The theory of gene-culture coevolution (see Fig. 26.1) proposes the following process.
First, human genes affect the way that the mind is formed  which stimuli we perceive, how information is processed, the kinds of memories most easily stored and recalled, the emotions they are most likely to evoke, and so on.
These effects, which have been well documented in recent psychological research, are called epigenetic rules.
The rules are rooted in the particularities of human biology, and they affect the way culture is formed.
For example, outbreeding is much more likely to occur than brother-sister incest because of the apparently innate rule that individuals raised closely together during the first six years of life are inhibited from full sexual intercourse at maturity.
Certain colour vocabularies are more likely to be adopted than others because of another rule: the retinal colour cones and certain interneurons within the brain encode light into four basic colours, even when the wavelength of light falling on the eye varies in a continuous manner.
The Dani of New Guinea have one of the poorest colour vocabularies in the world, in fact consisting only two terms, for ' bright ' and ' dark ' respectively.
Eleanor Rosch (1973) took advantage of this fact to conduct an experiment in learning propensity.
She gave one group of male volunteers a new colour vocabulary to learn in which the terms were centred on the four basic colours.
Another group of men received a vocabulary centred on the wavelengths at the margins of the basic colours.
Individuals in the first, ' natural ' group learned the words twice as quickly and retained them longer.
When given a choice between the two terminologies, Dani men preferred the natural vocabulary.
Both of these cases, incest avoidance and the development of colour vocabularies, illustrate nicely how biological constraints in cognition, based on specific genes, can influence the formation of culture.
Epigenetic rules have been demonstrated in virtually every category of cognition and behaviour investigated in such a way as to distinguish choices among stimuli.
Examples include odour and taste discrimination, with important effects on the evolution of language and cuisine; preference from infancy onward for certain basic geometric designs over others; phoneme formation; rules of transformational grammar; the development of particular, species-wide facial expressions to denote the emotions of fear, loathing, anger, surprise, and happiness; various other forms of nonverbal communication; the pattern of mother-infant bonding; the method of infant holding by women; fear of strangers (a usually strong response that persists from about six to eighteen months); phobias; and others (see the review by Lumsden &amp; Wilson, 1981).
It is clear that during the past twenty years developmental psychologists have come to the edge of a vast array of structural processes in the development of the mind, and an exciting era of experimental research has begun.
Most of the ontogenetic patterns occur early enough in life and are sufficiently strongly marked and stereotyped to suggest that they are genetically canalized.
It is further true that some degree of heritability has been indicated by twin and pedigree analysis, of varying degrees of sophistication and reliability, in virtually every measurable category of cognitive ability and personality trait thus far studied.
Many of these properties form components of the epigenetic rules just cited (Ehrman &amp; Parsons, 1981).
By 1980 about 3100 human genes had been distinguished, mostly by biochemical techniques.
Of these, 340 were pinpointed to a particular chromosome, with at least one on each of the 23 pairs of chromosomes (McKusick, 1980).
Some of the genes and chromosome aberrations affect behaviour in selective ways.
A notable example is the major gene recently identified that reduces spatial ability in three standard tests but not in twelve others (Ashton, Polovina &amp; Vandenberg, 1979).
It is equally significant that the analysis of complex, multiple-locus systems (polygenes) is well advanced.
Recent advances include the calculation of the numbers of chromosomal loci and genes involved in such relatively complicated behaviours as dominance and drug aversive behaviour in mice and the epigenetic rules of colour preference in birds (Thompson &amp; Thoday, 1979).
An important principle of gene  culture coevolutionary theory is that a tabula rasa mind, open to all choices equally and hence totally dependent on the accidents of history, must still have a biological foundation  and a very finely adjusted one at that.
The sensory apparatus and brain have to be tuned precisely in order to process all stimuli without bias.
Such an effect, which Lumsden and I have called the ' pure cultural transmission ' of culture, can not be achieved merely by removing genetic constraints on cognition and learning.
Quite the contrary: it requires a formidable array of homeostatic devices, in order to achieve uniform information processing and hence independence from all but the most generalized set of internal reinforcement mechanisms.
And indeed there is no evidence that the human brain works in such an extreme behaviourist manner.
Nor does the grain mediate a ' pure genetic transmission ' of culture, in which (like the song of the white-crowned sparrow) only a single form of behaviour can be taught and learned.
All of the evidence from cognitive studies thus far indicates that human behaviour lies in between these two extremes, that is, in the category we have termed ' gene-culture transmission ' of culture.
Multiple choices are learned and any one of them can be followed  as for example incest versus outbreeding  but there is an innate predisposition to learn certain ones in preference to others, or else to choose them once they have been acquired.
Let us now review the essential steps in the proposed coevolutionary circuit.
Consider for example the case of the avoidance of brother-sister incest, which is based to a substantial degree on an inhibition developed during close domestic association in the first six years of life.
Because this epigenetic rule occurs across cultures and is strong enough to defeat countervailing social pressures, it can reasonably be supposed to have a genetic basis.
Moreover, those who follow the rule benefit in natural selection.
Incest results in higher rates of homozygosity, the more frequent expression of lethal or subvital recessive genes, and hence a greater incidence of hereditary disease and early death among the offspring.
The epigenetic rule thus directs the developing mind to avoid brother-sister incest.
The summed preference of members of the society lead to particular cultural patterns, including reinforcing taboos and laws, that prohibit incest.
However, because the cultural transmission is of the intermediate, ' gene-culture ' form, the preference is not absolute, and scattered individuals in many societies still prefer and may even practise brother-sister incest.
The result is some variation among cultures in the frequency of its members who adopt this preference.
Consider, for example, groups of 25 individuals.
This is the size of many hunter-gatherer bands, the social organization in which mankind has existed throughout most of its history.
At any given moment most such bands can be expected to contain no incestuous members.
A smaller percentage of the bands will contain one such member, a still smaller percentage will contain two incestuous individuals, and so on.
The full array of such fractions, comprising a frequency distribution across all cultures sampled, is called an ethnographic curve.
We have devised methods for predicting such curves from a knowledge of two functions: the magnitude of bias in the preference for one cultural choice (such as incest) versus another (outbreeding), and the degree to which the expressed preference of the remainder of the group affects the magnitude of the individual bias.
Two results of general interest emerge from this preliminary analysis.
First, it is technically possible to predict patterns of cultural diversity, expressed as the ethnographic curves, from a knowledge of individual cognitive development, and also to perform the reverse: to infer at least some of the principal properties of cognitive development from a knowledge of the pattern of cultural diversity.
The second result of broad interest is that cultural diversity is to be expected even if the underlying cognitive development is rigidly programmed.
Cultural anthropologists have commonly argued that the existence of substantial differences among cultures is evidence of the absence of underlying biological influence (Sahlins, 1976; Harris, 1981).
But this conclusion is entirely wrong.
Cultural diversity per se is evidence neither for nor against such control.
Rather, what matters is the pattern of the diversity.
As biological bias is increased toward one choice as opposed to another in the course of genetic evolution, the mode of the ethnographic curve can be expected to shift in that direction.
And as the influence of peer activity is increased -this influence itself may well be biologically determined  there will be a tendency for the ethnographic curve to change from a unimodal to a multimodal form.
By examining the pattern of cultural diversity in an explicit form such as the ethnographic curve, the nature of the underlying epigenetic rules can be partially inferred.
A recurrent working hypothesis of gene-culture coevolutionary theory is that the epigenetic rules are shaped by natural selection over many generations.
Returning to the brother-sister incest case for illustration, we note that individuals who conform to the aversion leave more offspring.
As a result, genes underwriting the avoidance of incest remain at a high level in the population.
Consequently the predisposition is sustained as one of the epigenetic rules.
In general, the rules leading to higher rates of survival and reproduction tend to increase in the population.
Thus the assembly rules of the mind build up during evolution, element by element.
In the genetic models, the tabula rasa brain, in which the mind is created solely by the circumstances of history, proves to be a very improbable outcome in the evolution of any conceivable intelligent species.
Even if a species somehow managed to begin with such a brain it would soon evolve in the direction of structural and biased epigenetic rules.
And as the evolution proceeds, small changes in the degree of bias can be expected to result commonly in much greater changes in the final cultural product.
For example, a barely detectable innate bias toward the use of body adornment, if combined with a moderate sensitivity to peer usage, would result in most or all members using such adornment in all societies.
Finally, a detectable amount of genetic evolution in the brain and mind can occur within only thirty or forty generations, or very roughly a thousand years.
If correct, this still purely theoretical conclusion implies that epigenetic rules and mental traits might have continued to evolve into historical times.
The conventional view, that such biological evolution ceased tens of thousands of years ago and human change has consisted entirely of cultural evolution since then, may be incorrect.
In closing, I want to take note of the familiar lament that science and technology have created not just a cornucopia but terrible dangers as well.
What is meant by science in this case is of course the physical sciences and to a lesser degree the biological sciences.
But the solution is not, as a few modern Luddites have suggested, the curtailment of science itself, including sociobiology and the social sciences.
Quite the opposite: the solution is to make every effort to extend new scientific procedures into the deeper reaches of human nature in order to provide solutions to those residual problems that continue to defy simple economic and technological solution.
The peculiar clockwork of the human mind, not scientific knowledge itself, is the source of the danger.
If evolutionary theory can be successfully extended to the assembly of the mind and the creation of cultural diversity, the result may well rank as the completion of the Darwinian revolution.
Whether the particular scheme summarized here can contribute substantially to that end remains to be seen, but I hope at the very least I have been able to express why I believe that the social sciences will eventually be fused with biology.
No natural boundary appears to exist between the natural and social sciences.
Their blend zone, a mysterious and sometimes prohibited domain, offers a great immediate potential for scientific discovery in the postulational-deductive and experimental tradition of the natural sciences. [C]
EVOLUTION, ETHICS, AND THE REPRESENTATION PROBLEM BERNARD WILLIAMS
This paper is concerned with culture and with evolution, but not with cultural evolution.
It discusses the relations between biological evolution and the areas of human culture which may broadly be called ' ethical '.
The concept of cultural evolution is problematical, and there are notorious difficulties about applying the notions of evolution and natural selection to cultural development; in particular, the ends served by various cultural developments are themselves defined by culture, as are the ' choices' to which Wilson refers in his paper (this volume).
That area, however, is not the concern of the present discussion.
There are two kinds of connection between evolutionary theory and ethics: one normative, and one explanatory.
There is also a connection between these two, to which I shall come later.
The first of these is older than the second, and has acquired a bad name; indeed, it acquired it fairly early, in some part from the monumental and unappealing system of Herbert Spencer.
In fact, as John Burrow has shown (Burrow, 1966) a lot of this material ante-dated The Origin of Species; the concept of ' the survival of the fittest ' (Spencer's own phrase) was already implicit in earlier sociological work which Spencer derived from Malthus.
Darwin himself had little sympathy for these ideas and not much, personally, for Spencer, though he did once say  I quote Burrow (p. 182)  ' in a moment of enthusiasm... that Spencer's Principles of Biology made him feel that he ' is about a dozen times my superior ', and thought that Spencer might one day be regarded as the equal of Descartes and Leibniz, rather spoiling the effect by adding, ' about whom, however, I know very little ' '.
The bad normative applications of evolutionary theory to ethics which were made by Spencer and others also, of course, involved a lot of bad evolutionary theory: if normative lessons could be drawn from Darwinian theory, there is certainly no reason why they should take the form suggested by Social Darwinists.
However, there is in addition a standard objection which holds that no such lessons can be drawn at all, at least in any directly logical way, since any project of deriving ethical content from premisses of evolutionary theory commits the ' naturalistic fallacy ', an error which is today often equated with that of trying to derive ought from is.
Interesting questions about ' naturalism ' in ethics in fact go beyond these purely logical issues.
Naturalism in a broader sense consists in the attempt to lay down certain fundamental aspects of the good life for man on the basis of considerations of human nature.
If this project falls, it is not for purely logical reasons; it will rather be for the more interesting reason that the right sort of truths do not exist about human nature.
I shall come back to this wider question at the end of this paper.
The point about ought and is, so far as it goes, does have some force.
It can be put in the following way.
Suppose that considerations of evolutionary theory show that certain behaviour is in some sense appropriate for human beings.
Either human beings can diverge from this pattern, or they can not.
If they can, then the biological considerations are not going to show that they ought not to; while, if they can not so diverge, then there is no question of ought.
This argument seems to me sound so far as it goes, but it does not go very far.
Implicit in this last argument is another logical relation which is more interesting for this question than that between ought and is: the relation, that is to say, between ought and can.
This relationship underlies some important negative arguments which by citing certain claims to the effect that human beings can not, as they may suppose, live in a certain way, lead to the conclusion that certain ethical goals or ideals are unrealistic and should be revised.
By arguments of this kind, biological or similar arguments could coherently yield constraints on social goals, personal ideals, possible institutions and so forth.
To say that human beings can not do certain things is, of course, an extremely vague form of statement.
At one extreme, it may mean that the world will not contain an example of any single human being doing that thing; at the other end, it may merely mean that if a group of human beings adopt a norm requiring that behaviour, the norm will often be broken, its observance will give rise to a good deal of anxiety, those who comply without anxiety to the norm will be unusual in other respects, and so forth.
This vagueness will not matter so long as one is clear about the level at which the formula of ' ought implies can ' is being applied: thus the latter, and weaker, kind of ' can not ' would be enough to provide a strong argument against the behaviour being made into a norm for a human society, but it would not be enough if the question concerned the adoption of a personal ideal in an individual case.
Here, as so often, it is a centrally important question, who is supposedly being addressed by a given piece of ethical discourse.
Granted that one is clear about that, the fact that relevant statements of what human beings can and can not do come in various strengths is not so important.
What is vitally important is the difficulty of knowing which of them, relevant to difficult ethical issues, are on biological grounds in fact true.
In this area there is an important connection, which I mentioned before, between what I called the normative and the explanatory interest.
If some biological constraint can rule out, or make unrealistic, some normative practice or institution, then knowledge of it may not only encourage us to decline that practice if it is suggested, but may also contribute an explanation of why human communities do not in general display that practice or institution.
Might biological considerations then go further and explain the human adoption of other practices, which are conformable to biological constraints?
This raises a general question which is central to these areas, and which I shall call the representation problem.
It is a problem which comes up at various points in considering the relations between biology and human practices, and may be put in the following way: how is a phenotypic character which would present itself in other species as a behavioural tendency represented in a species which has a culture, language and conceptual thought?
It may be said that in some cases, at least, such a tendency will show up in that species merely as itself  that is to say, as a merely biological character of that species.
But, in fact, virtually no behavioural tendency which constitutes genuine action can just show up in a cultural context ' as itself '.
Where there is culture, it affects everything, and we should reject the crude view that culture is applied to an animal in a way which leaves its other characteristics unmodified.
(Related to that view is the naive assumption of certain sociobiologists that sociobiology should expect to be more closely related to social anthropology than to other social sciences, because the ' primitive ' peoples studied by social anthropology are nearer to nature than human beings who live in large industrialized societies.)
None of this is to deny that there is a biological basis for elements in human behaviour which are culturally affected, moulded and elaborated.
It is not to deny that some culturally elaborated behaviour can usefully be explained from a biological perspective.
It is simply to recall the fact that almost all human behaviour, at least that which deserves the name of ' action ', is in fact culturally moulded and elaborated.
In accepting that there is a representation problem, I reject two views according to which there would be no such problem.
First is a simple reductionist view, which would neglect the way in which culture not only shapes but constitutes the vast mass of human behaviour.
When ancient Greek thought first discovered the opposition of ' nature ' and ' convention ', it also discovered that an essential part of human nature is to live by convention.
The study of human nature is, in good part, the study of human conventions, and that is what it is from the strictest ethological point of view.
That is how this species is.
It is a claim additional to this, but one which I also believe to be true, that human conventions, at least beyond a certain state of elaboration, can be understood only with the help of history, and that the social sciences accordingly have an essential historical base.
To pursue the question of whether that is so, would go beyond the limits of the present discussion, but it is worth bearing in mind, when the relations are discussed of biology to the social sciences, that an essential social science is likely to prove to be history.
The second point of view which is excluded by taking seriously the representation problem is one which I am disposed, perhaps unfairly, to call ' the Wittgensteinian cop-out '.
This is a view implicit in the idea that the central concept for gaining insight into human activities is that of a ' language game '.
Since ' language ' in this formulation is regarded both as the key to human convention, and also as something which human beings possess and animals do not, the phrase itself implies the lack of interesting explanatory or constraining connections between human and animal behaviour.
It suggests an autonomy of the human, under a defining idea of linguistic and conceptual consciousness, which tends to put a stop to any interesting questions of the biological kind before they even start.
It therefore does not give any help even in the areas, such as sex and hunger, where we most obviously need means of describing the relations between culture and the biological.
The feature of human culture and human activities that gives rise to the representation problem is above all that human communities embody norms, and it is this notion that I shall principally discuss.
However, there are other ways, as well, of picking out differences between human activities and those of other animals.
One is the very general feature that humans possess conceptual and reflexive consciousness; this, and the very large philosophical problems introduced by those three terms, I shall happily leave on one side.
Another distinction between human and animal behaviour is that considerations of motive are appropriate to the assessment of human action.
This is a matter that is worth some brief discussion, since it is closely connected with the fuss that has been made about the application of the term ' altruism ' to animal as to human behaviour.
In other animals there is behaviour which benefits another individual, and moreover there is behaviour the end of which is to benefit another individual, in a sense of ' end ' which requires a lot of work to make clear, but which is uncontentiously illustrated by behaviour the end of which is that the animal should take in food.
In the human case, many more layers can be added, and other distinctions drawn.
There are questions of intention, where this concerns what thoughts produce the action, and what features of the action are, relative to that thought, accidental.
There are questions of underlying desire.
Some actions which benefit others come from the desire just to benefit that particular person, while others flow from some more general disposition, while the desire to benefit a particular person or group may be accompanied by a variety of other desires, for instance to extract goodwill from them, or the possibility of a reward.
The cultural and psychological elaboration of these various motives of course raises difficulties for any simple relation of them to the biological.
Some of those difficulties arise just from the general problem of applying biological models to a species which engages in intentional thought; to that extent there is no special problem about altruism and morality.
People think that there is a special barrier here to the application of biological models, I believe, because they take ' altruism ', in a ' properly moral ' sense, to refer to some quite peculiarly pure motive, such as the intention to benefit others derived from impartial reflection on their interests and associated with no other desire whatsoever.
But it is extremely unreasonable to suppose that all (perhaps any) human beings act from that motivation, either, and if morality is to be a generally human phenomenon, it is simply a mistake to equate it from the beginning with such exigently Kantian formulations, and it is a mistake even from the point of view of the human sciences.
It is no doubt true that a biological perspective will make one more suspicious of extremely intellectualist or, again, very purist views of morality; but equally, so will a reasonable historical and psychological understanding of morality.
It is the notion of a norm that perhaps gives rise to the central representation problem.
The main point is condensed in the question raised by Pat Bateson in his paper (this volume), about the relation between an inhibition and a prohibition.
The most, it seems, that a genetically acquired character could yield would be an inhibition against behaviours of a certain kind; what relation could that have to a socially sanctioned prohibition?
Indeed, if the inhibition exists, what need could there be for such a prohibition?
If the prohibitory norm is to be part of the ' extended phenotype ' of the species, how could we conceive, starting from an inhibition, that this should come about?
This is a central example of the problem, but it is not the only example even with respect to norms, and it will be helpful to distinguish various things that fall under the general heading of a ' norm '.
Not everything that falls under this heading is a sanctioned prohibition.
We can distinguish various items; I will represent them as stacked, in a way which is typical, but not by any means universal.
(1)
Behaviour which is normal.
This does not just mean ' frequent ': exceptions are perceived as' odd ', but are not necessarily disapproved of, sanctioned, etc.
(2)
(1) together with an institution.
This can be applied to the case of marriage, where there will of course be usually sanctions of varying degrees against behaviours that threaten marriage, and sexual activity outside marriage may be disapproved of, but this does not imply that merely not engaging in marriage is disapproved of, nor that an unmarried condition is sanctioned.
(3)
Behaviour which lies outside (1) and (2), and to which in addition there may be strong personal disinclination: e.g. homosexuality as regarded in enlightened circles.
(4)
(3) together with rejection and disapproval of the deviant behaviour: e.g. homosexuality as regarded in less enlightened circles.
(5)
(4) together with sanctions personal or legal: e.g. homosexuality in the least enlightened circles.
Among the cases in which the options are not stacked like this, is that in which the sanctions and disapproval exist against behaviour which is in fact frequent and not the subject, perhaps, of any deep personal disinclination; this, at the limit, is pure humbug, like the old school-master's attitude to masturbation.
(1) to (3) of course raise some difficulties for a biological approach, particularly with regard to institutions, and an adequate treatment of the representation problem will deal with all these levels.
The question of inhibitions and prohibitions arises most clearly at levels (4) and (5).
There is, moreover, a specially paradoxical version of it which arises from certain cases in which not only does extra conceptual content have to be introduced to characterize the human prohibition, but also the introduction of that content stands in conflict with the proposed biological explanation of it.
A clear example of this arises with the famous example of the incest taboo, which has been discussed by Bateson (this volume).
There are of course many incest taboos, that is to say, prohibitions on sexual relations between persons of various degrees of familial relation, and some of these are hardly even candidates for biological explanation.
Moreover, there may well be some very severe doubts about the application of the biological model even to the favourite cases.
The present discussion, is not, however, concerned with the factual merits of these explanations, but only with the shape that they take.
In other species, there are behavioural drives the function of which is to avoid inbreeding.
Such a drive, however, has to be operationalized in some other way, since the animals do not have any direct knowledge of the matters relevant to inbreeding: the inhibition against mating has to be triggered by the recognition of or reaction to some property adequately correlated with the kin relationship, such as being an individual with which the animal has been brought up.
It is this inhibition that is allegedly displayed, in the well-known case, by those brought up in the kibbutz.
But we have not yet reached any incest taboo.
There are no sanctions against marrying those that one is brought up with (as such); the sanction is against marriages which would constitute close in-breeding.
The conceptual content of the prohibition is thus different from the content that occurs in the description of the inhibition.
It indeed relates to the suggested function of that inhibition, but that fact will not explain how the prohibition which is explicitly against in-breeding will have arisen.
It certainly does not represent a mere ' raising to consciousness' of the inhibition.
It can have come about, in fact, only given human knowledge of relevant facts  presumably, of the ill-effects of in-breeding.
But once that is an essential step in the explanation, we no longer need the biological element in the explanation (of the prohibition, that is to say, rather than of the inhibition).
It turns out that we have to appeal in any case to something like a rational collective agency, directed towards avoiding recognized and agreed evils, and that already provides an adequate explanation  a fairly traditional one  of the incest prohibition.
A similar paradox can arise with other norms supposedly based at a biological level, but there are cases that avoid it.
Consider for instance the ' double standard ' in sexual morality, traced by Symons (1979) to the disparity between ovum and sperm.
This account, though it applies much more widely, is essentially the same as an explanation of these social phenomena which goes back at least to Hume, who accounted for ' the artificial virtues of chastity and modesty in women ' by referring to the naturally greater disposition of males to protect children that they believe to be their own.
Here again, there may be serious doubts about relevant anthropological facts, but the present point concerns the principle of the explanation, which involves an important difference from the incest case.
Here, it is natural to think in terms of the institutionalization of a disposition which could be displayed in a simpler form pre-culturally.
The conceptual content required in this case to describe the institution, though it involves a great deal of cultural elaboration, does not display the same kind of break between the pre-cultural and the cultural as is found in the incest case; and the biological pattern of explanation could recognizably run through such ideas as human beings finding certain institutions' natural ', which does not require any appeal to a rational collective agency to understand the basic biological idea, as is damagingly the case with the incest example.
In fact, an explanation which went back to a biologically grounded disposition could in this case precisely avoid the invocation of rational collective agency, which is rather an intellectualist embarrassment to the story as Hume (1738C40) tells it.
None of this implies that even if such biological elements did play some role in explaining these institutions, the institutions would then be necessary or unchangeable  even if the explanation were true, this could still be a case in which becoming conscious of their rationale was a help in changing them.
In one of the two cases we have considered  incest  the prohibition is paradoxically related to pre-cultural dispositions: it expresses their function, but not their content.
In the case just considered, social institutions could in principle be an expression of a pre-cultural disposition.
In other cases, again, the existence of norms seems to be a substitute for a pre-cultural disposition.
This might well be so with the control of aggression and of self-seeking behaviour; I shall make one or two remarks about this question without pursuing it at length.
In the work of Maynard Smith and others (see, for example, Maynard Smith, this volume) games theory is applied to explaining selection for certain genetically based patterns of behaviour.
Games theory can equally be applied to characterizing human norms which are instituted against aggression and other non-cooperative behaviour.
(Ullmann-Margalit (1977) gives a recent analysis, though the outlines of the idea that sanctioned norms can represent a solution to the Prisoners' Dilemma can be found in Hobbes.)
The principles of the two applications of games theory are in many ways the same but their results point, in a sense, in opposite directions.
If sanctioned norms are necessary in the human case, or socialization into rule-observing behaviour, this must be because constraints on human responses in these areas are not, or not significantly, genetically based.
Granted structures of the Prisoners' Dilemma type, cooperative behaviour can be secured only granted a certain level of assurance, and the need of norms (in particular of sanctioned norms) to produce that assurance shows that the assurance can not be adequately delivered by genetically based signals.
It is very tempting to suppose that the lack of any such reliable signals, and the perilously low level of security often reached in human communities, must be connected with a high level of conceptual and, in particular, predictive thought, and also an associated capacity for deceit.
This perhaps gives a special force to the Voltairean remark about the function of language being to conceal thought.
The previous remarks have raised some questions about the relations between human norms and possible underlying dispositions determined at a biological level.
They represent some aspects of what I have called the representation problem, and it is only through further investigation of that problem, and by becoming clearer about how the various kinds of norm could relate to our biological inheritance, that we can come to see much about what biological constraints there might be, beyond the obvious ones, on social and ethical arrangements.
I do not believe it to be excluded a priori that there could be some, and I do not believe that very much is to be achieved by very general assertions or denials of the possibility.
What is needed is more detailed analysis, not only anthropological but philosophical, of the demands that any explanations of this sort would have to meet.
It will be needed, above all, if we are to be able to read the historical record.
It is only if we can read that record that we can discover some very important biological characteristics of human beings, since (to repeat an earlier point) it is through convention, convention that has a history, that human nature is expressed.
It is not merely that without a hold on the representation problem we can not discover the relevant content in the historical record; without understanding that problem, we can not adequately control the idea that there is any relevant content at all.
If a biologically grounded disposition showed up simply in the form of what human beings could not or would not do, then there would be no real problem of alternative behaviours.
The alternatives will simply be absent from the record, and it is unlikely that anyone, except as the most extreme perversity, would want to undertake them.
This, of course, is the area in which the is/ought argument scores its clear but uninteresting success.
What is much more interesting, I have already suggested, is the idea that there could be patterns of behaviour which human beings are entirely capable of wanting and indeed, on an individual or limited scale, of achieving, but which for biological reasons are bound to be psychologically costly, or confined to a small group of otherwise unusual individuals, or otherwise bound to fail as general social institutions.
To understand how this could be involves some understanding of the representation problem, and to decide that any given pattern of behaviour has this character of being, as one might put it, ' biologically discouraged ' requires one to be able to read the historical record.
It hardly needs emphasizing that on any question that is interesting, such as social roles of the sexes, we would have to be able to read the historical record better than we now can in order to arrive at any strong conclusions about what is biologically discouraged.
I come back finally to what I mentioned at the beginning of this chapter as the area of ' naturalism ' more broadly conceived: that is to say, the question of founding human ethics on considerations of human nature, in some way which goes beyond merely respecting the limits, biological or other, on what human beings are able to do.
This is the project of thinking out, from what human beings are like, how they might best and most appropriately live.
Such a project continues to attract some philosophers.
Its attractions are obvious.
It does not, in any obvious way, require any supernatural warrant, while it is less arbitrary or relativistic than other secular ways of looking at the content of morality.
It seems to offer some promise of being both well-founded and contentful.
It seems to me that a correct understanding of human evolution is very relevant to projects of this kind, but that the effect of that understanding is largely discouraging to them.
This is for two different kinds of reason.
The first is a reason at a more particular and factual level and is correspondingly more sensitive than the other to changes in hypotheses about the emergence of human beings.
It is simply that the most plausible stories now available about that evolution, including its very recent date and also certain considerations about the physical characteristics of the species, suggests that human beings are to some degree a mess, and that the rapid and immense development of symbolic and cultural capacities has left man as a being for which no form of life is likely to prove entirely satisfactory, either individually or socially.
Many of course have come to that conclusion before, and those who have tried to reach a naturalistic morality which transcends it have had to read the historical record, or read beyond the historical record, in ways which seek to reveal a partly hidden human nature which is waiting to be realized or perfected.
The evolutionary story, to the extent that it can now be understood (and to the much more modest extent to which I understand it myself) seems to me to give some support to the view that in this respect the historical story means very much what it seems to mean.
The second and more general reason lies not in the particular ways in which human beings may have evolved, but simply in the fact that they have evolved, and by natural selection.
The idea of a naturalistic ethics was born of a deeply teleological outlook, and its best expression, in many ways, is still to be found in Aristotle's philosophy, a philosophy according to which there is inherent in each natural kind of thing an appropriate way for things of that kind to behave.
On that view it must be man's deepest desire  need?  purpose?  satisfaction?  to live in the way that is in this objective sense appropriate to him (the fact that modern words break up into these alternatives expresses the modern break-up of Aristotle's view).
Other naturalistic views, Marxist and some which indeed call themselves' evolutionary ', have often proclaimed themselves free from any such picture, but it is basically very hard for them to avoid some appeal to an implicit teleology, an order in relation to which there would be an existence that would satisfy all the most basic human needs at once.
The first and hardest lesson of Darwinism, that there is no such teleology at all, and that there is no orchestral score provided from anywhere according to which human beings have a special part to play, still has to find its way fully into ethical thought.
REFERENCES etc. [ pp. 565C566 ] OMITTED; p 567: EPILOGUE; p 568: BLANK
