THEORIES OF POLITICAL AND SOCIAL VIOLENCE
Johan M.G. van der Dennen
Hartogs & Artzt (1970) first identify organized violence, which is "patterned and deliberate". All organized violence is instrumental in the sense that it is one means of social combat among many, and it functions in a context of group interests and goals. The second type of violence is spontaneous violence, which is an "unplanned explosion set off by the unique chemistry of internal and external conditions". While organized violence is both instrumental and impersonal, spontaneous violence is reactive, compensatory, or gratuitous. As a reaction, it is a way of striking out directly against frustration. As a compensation, it is a way of making up for frustrations suffered in the past. As gratuitous violence, it is a way of displacing aggression from an object which cannot be attacked (because it is too powerful or because it generates ambivalent feelings) to an object which is too weak to resist and which arouses clear feelings. Spontaneous violence may be collective or individual. The third type of violence identified by Hartogs & Artzt is pathological violence, which is committed by individuals and has a basis in either physical or mental illness.
Grundy & Weinstein (1974) propose that organized violence may be divided into criminal and political types. Criminal violence, while impersonal or instrumental, is not directed at the defense, disruption or restoration of a normative order, although it may unwittingly contribute to such outcomes. Political violence is directed at the maintenance or change of a normative order. Nieburg (1968) observes that political violence "addresses itself to changing the very system of social norms which the police power is designed to protect". Grundy & Weinstein would add that political violence can also concern itself with maintaining the normative order under attack. However, the distinctions between criminal and political violence, and organized and spontaneous violence are not always clear because all acts of violence, whether or not they are deliberate, can be used for some political purpose. A rise in criminal violence can become an excuse for officials to change the normative system. Criminal bands can become social bandits and finally guerilla movements. Yung-Teh-Chow (1966) has shown how this process worked in ancient China: "When a bandit group felt strong enough to challenge the central government for control of a region, it would take the designation of i-chun [yi qun], 'righteous troops'".
Similarly, it is often difficult to make a distinction between organized and spontaneous violence. Officials can interpret a riot as a political conspiracy, while oppositionists can interpret a riot as part of a movement for change (Grundy & Weinstein, 1974).
Violence defined in terms of violation
Those dissatisfied with the social structural 'status quo' often tend to define violence in terms of the human experience of 'violation'. The argument of Garver (1958) can serve as an example of this approach, which will be referred to as 'expansive' (Grundy & Weinstein, 1974).
Garver begins his discussion by noting that violence is often connected with the use of physical force: "Violence often involves physical force, and the association of force with violence is very close: in many contexts the words become synonymous". However, Garver argues that the connection of violence with force is superficial. One does not say that a doctor is being violent when he is attempting to repair a dislocated shoulder. The core meaning of violence is the act of violating a basic right of the human being. For Garver, the two basic human rights are the right to one's body and the right to autonomy. Violation of these rights implies that violence has been done.
From this basis, Garver defines four types of violence. Personal overt violence is the "overt physical assault of one person on the body of another". Overt institutional violence occurs when people obeying orders within an organization physically assault others. Garver notes that it is difficult to assign responsibility for overt institutional violence because while the individual is not acting on his own initiative, "the group does not have a soul and cannot act except through the agency of individual men". Quiet personal violence occurs when a human being deprives another person of autonomy through the manipulation of symbols: "If we fail to recognize that a real psychological violence can be perpetrated on people, a violation of their autonomy, their dignity, their right to determine things for themselves, to be men rather than dogs, then we fail to realize the full dimension of what it is to do violence". Finally, quiet institutional violence occurs when some people are systematically denied access to social options open to others. "The institutional form of quiet violence operates when people are deprived of choices in a systematic way by the very manner in which transactions normally take place". Garver notes that denial of options is an attack on autonomy and, therefore, constitutes violence. The definition of violence as violating persuasion is not the most widely accepted usage in the West. It extends or expands the definition of violence to its widest reaches, subsuming a number of acts and conditions deemed immoral and heretofore not regarded as violence. As such it is referred to as an expansive and ethical definition (Grundy & Weinstein, 1974).
The same viewpoint can be found well expressed in the works of Galtung (1959), Wertham (1966), and Bettelheim (1966).
This expansive and ethical definition tends to find the source of violence in the act of violation because those who apply it often want to explain in sympathetic terms the personal overt violence of the oppressed. One way of excusing and justifying personal overt violence is to classify it along with other actions that are called violent because they have some of the same effects on human beings. The oppressed cannot deprive the oppressors of many choices in a systematic way, but they can often resort to personal overt violence. The expansive definition enables one to interpret this personal overt violence as a reaction to the quiet but no less effective and inhumane institutional violence of the oppressor. At the very least one can claim that personal overt violence is no more immoral than quiet institutional violence. When one wishes to make a stronger case, he may claim that personal overt violence is a morally justifiable reaction to quiet institutional violence. At its most ambitious extension, one can claim that it is morally obligatory to smash quiet institutional violence through personal overt violence. In all cases the key to this argument is the equation of physical attacks on persons with institutional denial of opportunity.
Those who use the expansive definition often argue that violence 'is' violation and 'must' be interpreted that way. They seem to be saying that people should use violence to mean violation when they make moral judgments. This is not the same as saying that violence 'is' violation despite claims to the contrary (Grundy & Weinstein, 1974).
Violence defined in terms of physical force
This interpretation is a leading one in contemporary American political thought and many writers have developed it. We might regard this as an observational definition of violence because it is rooted in the observational act of applying physical force and does not distinguish between the source, purpose, or effect of such acts (Grundy & Weinstein, 1974).
Ogle (1950), for example, argues that the term force can be used to refer to a number of different types of action: "It may be purely physical - an overt individual act, such as the striking of a blow or aiming of a gun, or the activity of a crowd as in the case of a mob lynching. Wherever it entails the use of material, measurable force, we may refer to 'violence'". Thus, Ogle does not count psychological denials of autonomy and social denials of access to opportunity as types of violence. He is clear, however, that all acts of personal overt force and overt institutional force count as acts of violence and is careful to declare that he does not want to differentiate acts of force committed by people who are not officials: "A distinction is clearly made between the use of force or coercion on the part of an organized, recognized governmental authority and its use by persons or groups not officially connected with the state. In both cases it would seem clear that the same kind of results, physical or psychological, might be expected from the use of force or violence".
A similar interpretation is given by Nieburg (1969). Nieburg states that violence can be "unambiguously defined as the most direct and severe form of physical power... It is force in action. Its use is a continuation of bargaining begun by other means, whether it is used by the state, by private groups, or by persons". For Nieburg, violence is distinguished from force. Force is a "reserve capability and means of exercising physical power" and "amounts to a threat of violence or counterviolence".
Like Ogle, Nieburg does not recognize psychological or social denials of autonomy as acts of violence, and claims that the distinction between legitimate acts of force by officials and illegal acts of violence by others is superficial.
The observational definition of violence as "destructive physical action against another person" is no more correct than the expansive definition of violence as violation. Essentially, the observational definition has narrowed the expansive definition to include only acts of violation involving the use of physical force. This latter definition is frequently used to excuse and sometimes to justify certain acts of violence by people who are not officials and to judge as immoral certain acts of violence committed by officials.
Thus, those sympathetic with the revolutionary tradition will excuse acts of violence by rebels against totalitarian regimes in which official policies are overtly backed by violence. Counterviolence against tyranny is condoned because the ruling class has not allowed others to engage in peaceful forms of bargaining. At the same time, acts of violence committed by totalitarian regimes are interpreted in the observational definition as illegitimate violations of access to the bargaining process and as violations of the right to one's body.
The ground shifts, however, when violence appears in a regime that pays lip-service to and displays the trapping of representative democracy and yet is bedevilled by pockets of discontent or violent dissatisfaction. The ambiguities of this position become manifest in this context. Here, where social control operates primarily through psychological attacks on autonomy and institutional denials of opportunity, those who employ an observational perspective on violence can see violence committed by disadvantaged groups as either a symbolic demand for a greater share of social values or as an illegitimate attack on institutions for settling disputes which are open to all citizens. Similarly, they may view violence committed by the authorities as either overreaction to legitimate demands or as the efficient use of force to minimize the total amount of violence committed in the long run. The revolutionary stance towards totalitarian regimes and the reformist position towards democratic regimes is, in part, made possible by the observational definition of violence as force or physical harm.
Like the expansive definition, the observational definition may lead to a confusion of 'ought' and 'is'. The key to the discussion of violence is equating force exercised by the state and force exercised by people who are not agents of the state. Like Garver, Ogle justifies this equation on the grounds that "the same kinds of results, physical or psychological, might be expected from the use of force or violence". From the political scientist's perspective, there is no doubt that force exerted by officials and force exerted by others have many consequences in common. However, this does not mean that these two types of exercise of force are the same in all respects, or even in all relevant respects. Attempts to prove that the definition of violence 'is' or 'must be' destructive physical action against another person fail because both definitions meaningfully lead to different interpretations of the term violence. Those using the observational definition imply that their meaning should be the standard definition of violence, but they cannot demonstrate that it is the correct meaning, or even, indeed, that there is one 'correct' meaning (Grundy & Weinstein, 1974).
Violence defined in terms of illegitimate force
A third usage narrows the definition of violence even further to include only acts of violation in which physical force is applied and which are illegal. This is called the orthodox or 'narrow' definition by Grundy & Weinstein (1974). Hook (1945) has argued in favor of defining violence in this way. He states that violence is "the illegal employment of methods of physical coercion for personal or group ends".
From this perspective, physical coercion employed by 'duly constituted authorities' is legitimate and should be called 'force', while the term violence should be applied only to acts of illegal physical coercion.
Allied with this interpretation is the idea that violence is an aberration - an unexpected interruption in the normal course of events. Wolin (1970) has advanced this point of view. He states that "violence denotes an intensification of what we 'normally' expect a particular power to be". He continues that we designate "acts as violent because the amount or degree of force does not seem commensurate with the circumstances or with what we have come to accept as the characteristic style". Wolin concludes that "violence implies that an unusual amount of destruction will accompany a designated act". A similar position is taken by Walter (1964). Walter remarks that violence is "generally understood as immeasurable or exaggerated harm to individuals, either not socially prescribed at all or else beyond the limits established for its use". He states that when violence is "socially prescribed and defined as a legitimate means of control or punishment, according to the practices familiar to us, the destructive harm is measured and the limits made clear". The basis of the narrow approach to defining violence is, thus, neither the experience of violation nor the experience of harmful force, but the experience of having one's expectations in interpersonal relations disrupted by acts of force (Grundy & Weinstein, 1974).
The narrow definition is based on the grounds that the official use of force functions to support a system of stable expectations, while the use of violence by those who are not agents of the state functions to disrupt such expectations. Underlying this position is the assertion that the use of force is inevitable in human affairs. Mayer (1926) has presented the narrow interpretation with clarity. Mayer, a sociologist, states: "Organization, off hand, is of many forms - cultural, scientific, religious. But underneath them all there is just one type, namely, the organization of force. This is basic. It is the tap root from which all other forms of organization spring and upon which they rest for their ultimate sanction". The socialization of force allows the function of protection to take precedence over the function of defense in human affairs. Adopting a Hobbesian position, Mayer argues: "Protection is the string upon which the social beads are threaded. Take away community protection, and nothing but personal fighting and general anarchy can result". Thus, Mayer traces some important implications of the narrow definition. The narrow constructionists distinguish legitimate from illegitimate force because they observe consequences of force different from those noted by those who employ either the expansive or observational usage of violence.
The first group notes that one consequence of force is violation of human rights to body and autonomy, while the latter group notes that a consequence of illegitimate force is violation of the right to stable expectations of orderly interpersonal relations (Grundy & Weinstein, 1974).
This definition of violence is no more correct than either of the preceding two definitions. The definition of violence as illegal force and the idea, advanced by Hook (1945), that power "which has legal sanction and which expresses itself in the imposition of physical constraint as well as in the use of less conspicuous but more effective social pressures, such as discriminatory economic, cultural, and administrative measures, should not be considered violence"; are important supports for a basically conservative interpretation of political morality. They allow the defender of established institutions to claim that the use of force by officials of the state is justified, while the use of force by others is illegitimate, and hence not justifiable. However, this moral argument cannot be made valid simply by narrowing the definition of the term violence (Grundy & Weinstein, 1974).
Violence in Systems Theory
The standard approach to the interpretation of violence in systems theory is developed in Talcott Parsons' (1954) essay 'Some Reflections on the Place of Force in the Social Process'. Parsons' discussion is based on the functions of legitimate force in supporting systems of stable expectations and of illegitimate force in disrupting such expectations.
Parsons bases his discussion on classifying force as a means of social control. Force is control of human situations in which " 'alter' - the unit that is the object of 'ego's' action - is subjected to physical means to prevent him from doing something ego does not wish him to do, to 'punish' him for doing something that from ego's point of view, he should not have done (which may in turn be intended to deter him from doing similar things in the future), or to demonstrate 'symbolically' the capacity of ego to control the situation, even apart from ego's specific expectations that alter may desire to do things that are undesirable from ego's point of view".
Thus, force is the use of physical means by a social unit to deter, to punish, or to demonstrate superiority. Parsons states that the central meaning of force is deterrence. It is an essentially negative means of social control, which can compel by actual physical intervention and can coerce by threat of physical intervention. It is much more effective in preventing behavior than in inducing action.
While force is only one among many means of controlling human behavior it plays a distinctive role in the social process. Parsons notes that each mode of social control can be 'foreshortened'. For example, Parsons defines power as the ability of a social unit to carry out its plans "regardless of ego's wishes - not necessarily against them, but independently of them". Power is "a generalized medium for controlling action - one among others - the effectiveness of which is dependent on a variety of factors of which control of force is only one, although a strategic one in certain contexts". Force is power foreshortened. Like gold is the base metal for monetary systems, force is the basis of power: "In the context of deterrence, we conceive force to be a residual means that, in a showdown, is more effective than any alternative".
In Parsons' theory the social function of force is to serve as an "ultimate symbolic basis of security", a reserve on which the regime can, from time to time, draw. In the most effective social system negative sanctions are not visible and positive sanctions are present everywhere. This is the case because a stable normative order is defined as a situation in which people have internalized patterns of expectation into their motivational systems. Compliance is the rule rather than the exception, and negative sanctions are only implicitly threatened. People view the force of the state as a means to be used in the last resort, when the other means of social control have failed to elicit obedience. Parsons holds that force is minimized in a constitutional regime because of a tacit agreement among citizens that restraint in demands for immediate fulfilment of promises will be repaid by restraint in coercion of the opposition by those in power. In totalitarian regimes officials are not restrained in coercing the opposition and cannot expect demands to be moderate when there is an opportunity to express them. In Parsons' thought, force (as distinguished from violence) functions to deter deviations from compliance with the requirements of a normative order. Violence is an attack upon that normative order or system of expectations. Force is one of the many means by which human behavior is controlled and is distinguished from the others by its status as the ultimate symbolic basis of security and the residual means of deterrence that are more effective than any other in a showdown. Parsons' theory of the place of force in the social process shows a theoretical background for justifications of violence in legitimist ideologies, although it is not itself an ideology. That it has ideological ramifications, albeit unintended ones, is undeniable. In legitimist ideologies, the force of the state is justified when it protects and defends an ongoing normative order (Grundy & Weinstein, 1974).
In the works of Lewis Coser (1966; 1968) there is an alternative functional account of violence rooted in systems theory. While Parsons relates the use of force to the maintenance of a normative order, Coser relates the exercise of violence to the long-run minimization of physical conflict. Coser attacks theorists like Parsons with the argument that "the curiously tender-minded view of the social structure which has generally predominated in American social theory is seriously deficient and needs to be complemented by a more tough-minded approach". He sets himself the task of showing that violence has important social functions. In his essay 'Some Social Functions of Violence' (1965) Coser argues that violence sometimes functions as a badge of achievement, a danger signal and a catalyst for social action. In each case violence serves a collective purpose. Violence frequently functions as a badge of achievement in disadvantaged groups. In highly differentiated social structures in which achievement is rewarded, those who are judged according to ascribed status and are placed at a permanent disadvantage turn to violence as a way of demonstrating achievement: "In the area of violence, then, ascriptive status considerations become irrelevant. Here, the vaunted equal opportunity, which had been experienced as a sham and a lure everywhere else, turns out to be effective".
Following this line of reasoning, Coser adds that participation in revolutionary violence frequently offers the actor his first chance to participate in political processes. Further, violence symbolizes the commitment of the individual to the revolutionary cause. By striking out against the state overtly, the revolutionary demonstrates that he is willing to risk his body and autonomy for the cause, and to become a member of a new polity that is in the making.
Viewing violence as a badge of achievement does not take Coser out of the tradition of systems theory. His argument is based on instances of role strain in contemporary society. In the Parsonian view, deviance from normative patterns is likely when expectations are frustrated. The person who is told that there is equal opportunity for advancement and that people are judged according to their merit will experience frustration when he is denied access to values because of his skin color or other ascriptive criterion. One reaction to this frustration is violence. Violence is a great equalizer in the sense that it is the residual of power. Where violence functions as a badge of achievement it is likely that there had been inflation in the power system followed by a deflationary spiral. Using violence as a badge of achievement is functional for a few individuals in disadvantaged groups and for the solidarity of some revolutionary groups, but it is not functional for the maintenance of a society-wide normative pattern. On the contrary, it may be specifically employed in this context to destroy the 'status quo'. In the long run it may function to minimize violence if members of the disadvantaged groups are able to gain a stronger foothold in the social structure (Grundy & Weinstein, 1974).
A second function of violence noted by Coser is its appearance as a danger signal, a barometer of social disaffection. He argues that riots in the United States have "constituted quite effective signaling devices, perhaps desperate cries for help after other appeals had been unavailing". In this case the violence of riots is a sign that more violence will come if measures are not taken to meet some of the demands of the protesting group. Thus, the violence of riots can function to minimize violence in the long run by alerting officials and elites of the need to head off more serious revolutionary activity.
Viewing violence as a danger signal is squarely in the tradition of systems theory. In the terms of Parsons, violent activity is evidence that the officials have undertaken either too few or too many commitments. If they are able to rectify their mistakes before the onset of a severe deflationary spiral, the normative structure will be saved and violence will be minimized. Like Parsons, Coser discusses violent activity within a single normative order. For the most part, in the case of violence as a badge of achievement the members of disadvantaged groups accept the norm of achievement present in the wider society. They are members of 'sub-cultures of violence' who breach certain norms so that they can fulfil others. In the case of violence as a danger signal the primacy of a single normative order is even more evident. The rioters are crying for help. They want access to the values of the more advantaged groups and would obey the general norms if granted such access. Thus, violence is a temporary negation of the normative order, not a move to institute a new order (Grundy & Weinstein, 1974).
A third function of violence noted by Coser is its effect as a catalyst. As a catalyst, the violence of some calls forth a sense of "solidarity against their behavior". The commission of illegal acts of violence by agents of the state may "lead to the arousal of the community and to a revulsion from societal arrangements that rest upon such enforcement methods". Thus, the violence of both police and criminals, if publicized, can function to minimize violence in the long run by causing negative reactions in the citizenry.
Viewing violence as a catalyst is also within the tradition of systems theory. There is a difficulty in explaining this function because there would be no need for revulsion against the violence if it had not been committed in the first place. To save Coser's argument it is probably best to assume that the function of violence as catalyst is derived from the functions of violence as mark of achievement and danger signal. If this is the case, to say that violence has a catalytic effect is to claim that there has been a response to the danger signal. From the Parsonian viewpoint acts of violence committed by the police threaten the normative order just as much as acts of violence committed by others. If these acts are made public and there is a revulsion against them future violence will be minimized and the normative order will not be seriously impaired (Grundy & Weinstein, 1974).
At the center of Coser's argument is the claim that violence is a danger signal. In his essay 'Internal Violence as a Mechanism for Conflict Resolution' Coser (1968) remarks: "Internal violence within a social system may be seen as a response to the failure of established authority to accomodate demands of new groups for a hearing. It is a danger signal as well as a means by which such groups make the demands heard". Further, such violence is an indicator of how serious the group is in pressing its claim. Aside from some ill-developed points about how violence serves to strengthen the resolve of group members, Coser rests his case on the point that intelligent reaction to the first appearances of violence can minimize the occurrence of future violence.
While violence may sometimes function as a danger signal, it is not clear that it always does so or that Coser has substantially revised the 'tender-minded' view of social structure. Coser remarks that those in power and authority usually react to political violence by branding it a prelude to revolution and attempting to head it off by resorting to counter-violence. He states that it "behooves the social scientist to point out to them that this is a profoundly mistaken view of the matter". If it takes social scientists to see to it that violence performs its social function, that function seems to be more a possibility than actuality. Perhaps Coser means that violence should function as a danger signal. Further, it is in the name of a 'politics of nonviolent compromise' that Coser makes his suggestions. It is just such nonviolent compromise that Parsons found to be the virtue of a constitutional regime. Yet Coser implies that Parsons is a tender-minded theorist. When subjected to analysis, Coser's theory appears to become another example of the same genre, systems analysis, with its built-in emphasis on systems maintenance (Grundy & Weinstein, 1974).
Violence in Theories of Group Conflict
An alternative approach to systems theory places conflict among groups, rather than the maintenance of normative order, at the center of analysis. Nieburg (1969) holds that it is incorrect to interpret political violence as either a deviation from a widely accepted normative order or as only a danger signal to elites so that they will be able to make concessions. Violence is an indication that there is a collective quest underway to discover new normative patterns: "Social fits and seizures can be viewed as a form of search behavior, a pragmatic trial-and-error method leading toward new political and social norms that better satisfy the strain toward security, predictability, and low-risk methods of conflict resolution which organized groups require and crave". This view of the social function of violence is based on the premise that "society is inherently composed of competitive individuals and groups, all struggling to maintain or advance their advantages by a wide variety of means". Those tactics which, through a process of trial and error prove successful in advancing the attainment of group goals tend to be repeated. Thus, when violence proves to be a successful means of gaining advantages it will be widely used.
For Nieburg violence is neither a residual base of power nor a last resort after the onset of a long train of abuses. No social order benefits all members equally, and normative structures tend to represent "the values, interests, and behavior of those who dominate the hierarchical structure of bargaining relationships".
The appearance of violence in political processes is a function of the fit between the formal and informal polities. The informal polity, which determines the direction of social existence, is "the concrete and objective bargaining situation that links the complex webs of interrelatedness among all individuals and groups, and maintains or modifies behavior and personality". The formal polity is made up of the rules and procedures which have grown up to recognize past bargaining situations or states of the informal polity. When there is not a close fit between the informal and formal polities the government loses legitimacy and there will be a more or less violent search for new norms.
At the heart of Nieburg's theory is the notion of bargaining. Nieburg holds that bargaining is a process of adjusting conflict through threatened or actual escalation and counter-escalation of sanctions. The limit of bargaining is the pure test of physical strength aiming at the annihilation or complete submission of the other. Most political violence stops short of such warfare and represents a phase in the bargaining process: "The threat of escalation as a deterrent, made credible by actual escalation from time to time, constitutes a claim to dominance or a challenge of existing patterns of dominance". Thus, violence is an integral part of the bargaining process. Violence and "the threat of violence - pure pain and damage - can be used to coerce and to deter, to intimidate and to blackmail, to demoralize and to paralyze, purposefully and meaningfully in a process of social bargaining". This perspective views violence as a natural form of political behavior, which cannot be dismissed as erratic or exceptional or meaningless, and disorder as intrinsically related to the social process. Violence, then, is "not only the last resort in the bargaining spectrum but also a potentiality or a threat which does in fact change the bargaining equation itself. In a sense, it is the ultimate test of viability of values and customary behavior". Nieburg holds that conflict and violence can be creative as well as destructive: "Conflict, in functional terms, may be viewed as the means of discovering consensus of creating agreed terms of collaboration". Violence is the 'cutting edge' of social integration. It is the ultimate way of measuring where one sovereignty begins and another ends.
Thus, the use of violence is profoundly ambivalent. It defines boundaries which sometimes lead to separation and stagnation and sometimes lead to higher levels of integration. Social existence is defined by competing normative orders and the frequently violent struggle among groups for influence is "society's instrument of creative growth, of continuous adaptation and choice among various options for internal integration". Thus, in Nieburg's view, violence can be creative as well as disruptive and symptomatic.
Nieburg's interpretation differs from those of Parsons and Coser because it recognizes the importance of competition among normative orders as well as adjustment and conflict within a single normative order. Parsons does not recognize a creative function for violence. He holds that the direction of social development is away from force and towards power. Coser recognizes positive, but not creative, functions for violence. Sometimes violence builds commitment to a normative order and sometimes it is a signal that a normative order is threatened. However, Coser stops short of saying that violence may be instrumental in attaining normative reconstruction (Grundy & Weinstein, 1974).
An interpretation of violence similar to Nieburg's has been presented by Roucek (1956; 1957). However, instead of making the bargaining process the core of his analysis, Roucek offers reflections on the proposition that the "relationship of force and justice is, apparently, a problem facing all ages and all cultures". Roucek notes that some theorists have avoided the "dilemma facing them when analyzing the moral approbation of the use of force by distinguishing between the legal and allegal aspects". Roucek does not accept this line of reasoning and points out, like Nieburg, that "society has a series of accomodations which are arranged like the front lines in such a way as to prevent a complete break-through to the home front". Drawing a similar distinction to the one Nieburg makes between formal and informal polities, Roucek argues that the state arises from conflict processes and survives by maintaining processes of accomodation or bargaining. Dominant groups create states of accomodation, often by force, and constitutions result from from new 'balances of power'.
Like Nieburg, Roucek recognizes a social function for violence. He notes that political violence can be committed in the interest of an outside government or a projected form of government, moral system, or society. Thus, a source of violence is conflict between normative orders. For Roucek, violence can help create new forms of political integration: "...violence is a useful social instrument, especially if we acknowledge that law in the modern state is based upon force". However, he warns that violence is least effective when it cannot create moral sanctions for its practices. Brute force accomplishes little permanent change, but violence backed by moral appeals can be quite effective. The conflict approach places the state on the level of all other groups and thereby makes violence a possible means of substantive change (Grundy & Weinstein, 1974).
Ideologies of Violence
Grundy & Weinstein (1974) identified four major justifications of violence:
Legitimist ideologies justify political violence when it is aimed at protecting or restoring a single normative order which the ideologist deems legitimate, or at disrupting an order which he deems illegitimate. For example, many people claim that the state is justified in using protective violence to quell disruptive violence, and others claim that the violence of revolutionary groups is justified when it aims at restoring a traditional normative order. Legitmist ideologies are related to narrow definitions of violence, because they are apt to define violence as an attack on the normative order which they are defending. However, this relation is not logically necessary and legitimist ideologists will often use any definition of violence which helps them further their aims.
A second major justification of violence is expansionist. In expansionist ideologies political violence is justified as a means of imposing a normative order deemed superior on alien groups. For example, an ideologist defending racial inequality may defend the imposition of a political order by force on a racial group he deems inferior. It is important to note here that the term expansionist has an ideological, not a spatial, meaning. The imposition of the supposedly superior normative order may be carried out by a dominant power group occupying the same territory as the subordinate group.
Dialectically opposed to expansionist ideologies are pluralist ideologies of political violence. In pluralist ideologies political violence is justified as a means of winning the right of a group to have its own normative order. For example, an ideologist defending the rights of an ethnic group to political and national self-determination may defend rebellion against established authorities as a means to attain these rights. The aim of the ideologist here is not to show why one group should dominate another, but why a group should escape from domination. This ideology of violence is called pluralist because it usually explicitly or implicitly recognizes the rights of all national or ethnic groups to self-determination. The use of the term pluralist should be distinguished from other popular meanings such as philosophical pluralism (the rejection of a single principle for explaining the universe), pluralist democracy (the competition of diverse interest groups within the institutions of representative government), and social pluralism (the presence of two or more relatively autonomous ethnic or national groups within a single society). Both expansionist and pluralist ideologies of violence are related to operational and behavioral definitions of violence because they defend acts of violence beyond those merely involved with protecting a given normative order. However, this relation is not logically necessary and pluralists, especially, frequently use broad and ethical definitions of violence.
The fourth major justification of political violence is intrinsic. In intrinsic ideologies political violence is justified by its direct contributions to the development of personal character, commitment to cause, and quality of social structure. For example, a revolutionary ideologist may argue that if oppressed individuals undertake violent acts against the ruling classes, they will be purged of their feelings of inferiority and become new persons, opposed to the dominant social structure and firmly committed to the society of the future. Similarly, other ideologists of social change may argue that if oppressed individuals undertake non-violent resistance, important and direct psychological benefits will ensue. The term intrinsic is used to describe such ideologies because they tend to justify violent acts in terms of the experiences which they evoke rather than in terms of ulterior aims. Intrinsic ideologies of political violence are related to expansive and ethical definitions of violence because they focus on attaining personal liberation from all forms of oppression. As was the case with the other ideologies, however, this relation is not logically necessary, and intrinsic ideologies will employ various definitions of violence depending upon the particular context.
'Social Violence': Aggressological Aspects of Abortion and Feticide
(1) 'Spiteful' abortion and feticide.
Among the many motivations for abortion in primitive societies, Devereux (1960) lists 'spiteful' abortion. Marital discord often leads to abortion. A Copper Eskimo beats his wife, thereby causing her to abort (Jenness, 1922). The beating of pregnant wives to induce miscarriage is a frequently reported phenomenon. Tupinamba (Thevet, 1878) and Australian women of Victoria (Ploss, Bartels & Bartels, 1927) abort if they are angry with their husbands. Crow and Assiniboine women abort if they are deserted (Denig, 1930). In Fiji a woman may abort to annoy a husband whom she suspects of infidelity (P1oss, Bartels & Bartels, 1927), or who has just taken a second wife (Williams & Calvert, 1859).
The Harney Valley Paiute woman throws away her live baby if she is hostile to its father (Whiting, 1950). There is indirect evidence that such hostile actions may have also a magical implication. An Araucanian unwed mother sometimes kills her little son and roasts his testicles in a hot pot, to make her faithless lover impotent. This rite is called 'koftun' (Cooper, 1946).
Anthropological literature is replete with examples of two major types of abortion motivated by revenge against the father, or by the desire to harm him in some way.
1. Voluntary abortion and infanticide may occur as a result of a domestic quarrel, or in order to take revenge on a faithless lover, or husband.
2. Compulsory abortion may be required of the woman, so that she would not bear a halfbreed, the offspring of an enemy of the tribe, or the child of an outsider. In each of these cases, the intention is clearly that of damaging the child's father. As regards voluntary spiteful abortions, the husbands are so well aware of the aggression against them which this act implies that they will severely punish their wives for aborting. The hostility against the father implicit in the aborting of a halfbreed offspring, of the child of a tribal alien, and of a potential outside heir, is likewise clearly manifest in the Tupinamba practice of deliberately causing a Tupinamba girl to be impregnated by a prisoner of war, whose offspring is then cannibalized as an act of vengeance against the enemy prisoner. (But since the girl knows what end is in store for her baby, she will often abort it) (Magalhães de Gandavo, 1922).
We do not even have to refer to psychoanalytic experience to show that the spiteful abortion of a child - be it voluntary or compulsory - respresents an attempt to castrate the father, though this element is clear enough when we are dealing with tribes where paternity is either a means of social advancement, or a proud public token of one's potency. In other words, the spiteful abortion of a child is tantamount to a depreciation of the potency of the impregnator, and to a frustrating of his sexual and social aspirations (Devereux, 1960).
(2) Penalties for abortion (Moralistic aggression).
Private penalties: 1. Scolding and anger (Truk: Gladwin, n.d.); 2. Severe beatings (Dobu: Fortune, 1932); 3. Killing (Jakun: Sumner, 1906; Chimariko: Driver, 1939).
Private action against third persons: 1. Damages and compensation; 2. Substitute killing: if an Azande husband whose wife aborted does not succeed in getting either a fine or a substitute wife from his wife's father, he kills either his wife or one of her siblings in revenge (Hutereau, 1909).
Semiprivate penalties: 1. Divorce; 2. Ridicule; 3. Contempt; 4. Indirect criticism. Public penalties: 1. Fines; 2. Beating (Chinese: Ploss, Bartels & Bartels, 1927; Creek: Foreman, 1934; Sanpoil: Ray, 1932); 3. Exile: to the primitive, exile is nearly as severe a punishment as death (Chinese: Ploss, Bartels & Bartels, 1927; Cheyenne: Hoebel, 1936); 4. The death penalty was imposed among the Aztecs (Murdock, 1934), except when the woman was aborted to save her life. The Assyrians impaled the woman who aborted (Sigerist, n.d.). The Mojo drowned such women at once, lest the village should suffer an epidemic of dysentery. The same 'punishment' was also inflicted on the Mojo woman who was unfortunate enough to have a miscarriage (Devereux, 1960).
Hoebel (1981) relates infanticide, invalidicide, senilicide and suicide and considers them to be privileged acts: approved homicide. A basic postulate, 'life is hard', and its corrolary, 'the insupportability of unproductive members of society' govern these acts. 'Life is hard' is a very general statement lacking specific content. Weyer's (1932) approach is even broader. Female infanticide, which he holds to be an artificial check against population increase, he considers to be a reaction to a population problem and part of the 'population policy' of the Eskimos. He admits that scarcity of food sets a low maximum on the density of population in the Eskimo region. Unchecked, population will increase beyond the optimum. Infanticide and senilicide are essential in establishing an optimal balance between resources and society, a balance that can be expressed in quantitative terms. Weyer perceives the importance .for the subject of infanticide, "for it involves a fundamental question in racial survival which is governed by laws closely analogous to those of organic evolution".
The reasons for infanticide among the Netsilik Eskimo as described by Rasmussen's (1931), Van de Velde's (1954) and Balikci's (1957) informants show a remarkable consistency. Balikci groups them in three categories:
1. Cases of attempted (or accidental?) child murder that seem to have purely social causes and cannot be related to environmental factors.
2. Second, conditions of extreme ecological pressure may produce infanticide.
3. To the third category belong the numerous cases of female infanticide that take place shortly after birth, under apparently normal conditions of life. Informants agree on one of reasons behind infanticide; namely that women do not hunt, they are not self-sufficient, and they are less independent than men. A hunter has to feed his daughter for many years, and when grown up she gets married and leaves the family at the very time when she is becoming useful. A comparison is made with dog teams, where usually only one bitch is to be found in each team: "bitches simply don't pull as hard as the male dogs do". A positive reason is also added. Boys, who will one day become reliable hunters, are greatly desired. If a woman is to suckle a female infant for two or three years she has not the chance of having a son during that period. The daughter is thus killed in order to make room, hopefully, for a son (Balikci, 1967).
Numbers were kept low among the Netsilik by causes of death other than those related to disease or old age. There are numerous cases of starvation, cases of infanticide under extreme pressure or for social reasons, several cases of murder related to wife-stealing and many cases of suicide motivated by purely social factors (Balikci, 1961). It may seem obvious, as Rasmussen (1931) noted, that a consistently high rate of female infanticide would disrupt the sex ratio in the society, leading the tribe towards extinction. To a considerable degree, however, the high death rate for adult males was a counterbalancing factor. While Rasmussen's census of the Arviligjuarmiut indicated a gross imbalance between the numbers of boys and girls, the same was not true for adults; he counted 19 men and 17 women. This tendency towards equilibrium in the sex ratio for adults can be explained by several factors. First, by the "higher death rate among men due to the natural hazards in hunting" (Weyer, 1932). This statement is valid for the whole of the Eskimo area although we do not have quantitative data for the Netsilik in the traditional period. Among the Netsilik, drowning, either during kayak-hunting or when crossing lakes and rivers seems to have been the most frequent type of accident. Hunters also were devoured by polar bears. Starvation, too, seems to have found more victims among men. Of the 25 deaths from hunger counted by Rasmussen (1931), 16 were men and only 9 women. Furthermore, men had a greater tendency to commit suicide than females (Balikci, 1961) Finally, in all of the murder cases described by Steenhoven (1959) the victims were men.
The higher death rate affected adult men of all ages. As a consequence of female infanticide, many young men reaching adulthood suffered from the lack of marriageable women. "Men fight among themselves for a wife, for a simple consequence of the shortage of women is that young strong men must take women by force if their parents have not been so prudent as to betroth them to an infant girl" (Rasmussen, 1931). The Netsilik knew of a number of ways of securing a wife (Balikci. 1963). Parents usually arranged for their children's marriages before birth. Young men did not rely exclusively on their parents to find them wives. They looked for themselves, too. When all these wife-securing techniques failed, men resorted to violence. Women were brutally grabbed. There are cases of wives who were stolen from visiting strangers or weak husbands. In addition, Steenhoven (1959) has described cases in which a husband was murdered for the purpose of stealing his wife.
Faced with the shortage of women, the Netsilik could adapt in various ways, including polyandry and the importation of women from neighboring tribes. This scarcity undoubtedly involved hardships: that of waiting for many years before a woman became available, that of the threat of murder involved in polyandrous arrangements, or, if everything else failed, that of emigration to a different land where matrilocal residence had to be accepted (Balikci, 1967).
A further means of socioculturally induced (and violent) population limitation - for which I suggest the term 'social spoliation' or 'social spoiling' - may have been the mutilating practices of clitoridectomy and infibulation ('pharaonic circumcision') in large parts of Asia, Europe, and, especially, the African continent. Apart from mutilation, pain, shock, and psychic trauma, bleedings, anorgasmia, and sterility due to infections of the urogenital system and uterus, the lethality of these practices was and is undoubtedly extremely high. Until very recently this ritual form of gynecocide has received little scientific and humanitarian attention.
The primary checks on population growth recognized by Malthus and by several of his predecessors in the field of population theory were those age-old scourges war, famine and disease. In the days when armies lived off the countryside, little remained for the inhabitants to eat. Furthermore, the filthy army encampments were breeding grounds for disease, with the result that epidemics usually followed in an army's wake. Of course, war was not the only cause of famine and disease; crop failures were frequent. Such failures were usually local, but in the days when transportation was difficult, famine relief was all but impossible. The local population was reduced to eating roots, bark and even grass, and many starved to death.
Disease tended to follow along trade routes, settling primarily in the unsanitary towns. Visitations of the plague came at intervals of 10 years, the worst epidemics often carrying off from a fourth to a third of the population. As late as 1720 a severe outbreak of the plague in southeastern France left 40,000 dead out of a total of 90,000 in Marseilles, the corpses being piled in the streets for lack of men to remove them.
Along with the plague went other lethal diseases. For example, only about four in every 100 persons failed to contract smallpox, a hideous illness that chiefly afflicted children in the first year or two of life. Smallpox was fatal in about one case in seven; in the latter part of the l8th century it was the cause of 10 percent of all deaths. In an expanded version of 'An Essay on the Principle of Population', published in 1806, Malthus noted: "The small-pox is certainly one of the channels, and a broad one, which nature has opened for the last thousand years, to keep down the population to the level of the means of subsistence". Plague and smallpox were undoubtedly the greatest killers, although tuberculosis and typhus were not far behind. Viewing the pre-Malthusian period as a whole, one can understand that humanity lived in constant dread of war, famine and disease, and that this terrible trio ensured a steady and substantial check on population growth. As Malthus noted, pestilence can sweep off thousands and tens of thousands. Then "gigantic famine stalks in the rear, and at one mighty blow levels the population with the food of the world".
Malthus recognized that in addition to the three major levelers there were other "active and able ministers of depopulation", among them such sexual irregularities as promiscuous intercourse, violation of the marriage vow and improper arts to conceal the consequences of illicit connections. He did not, however, attempt to describe or analyze the role played by these lesser influences.
Apart from enforced or voluntary, temporary or permanent, celibacy which was prevalent (e.g. among women of childbearing age in 18th century France 30 to 40 percent remained unmarried, and only some 40 percent of men founded families), and late marriages, thus considerably reducing the women's period of fertility in precontraceptive times, still another factor making for reduced fertility was the trade in mercenary soldiers that was so popular during those times. For centuries the cantons of Switzerland had done a thriving business in hiring out their young men to serve long enlistments in the armies of France, Spain, the Netherlands and various Italian states. Only a few of these men ever returned home and married; the mortality in war was heavy. It has been estimated that over the years the mercenary system did away with 35 to 40 percent of the natural increase in the Swiss population. There were still 20,000 Swiss mercenaries in the French army in 1748 and another 20,000 in the Dutch forces. In the course of the 18th century the Swiss cantons, with a total population of only 1.5 million or so, are thought to have lost 500,000 men on foreign battlefields (Langer, 1972).
Equally significant, if not more so, were the methods employed to dispose of unwanted children. Plato and Aristotle, and other ancient authors as well, accepted abortion and advocated infanticide as necessary means of ridding society of infants who were deformed or diseased, and as a means of regulating the size of the population. It is also well known that the exposure of newborn children was practiced by many societies, even in recent times. What is not so generally realized is the extent to which infanticide continued to prevail in Western civilization. In England, as late as 1878 about 6 percent of all violent deaths could be classed as infanticides.
The easy methods of committing infanticide included dosing the baby with gin or opiates; starvation and strangulation were often resorted to, and cases of a skull broken by a hairbrush were by no means unknown. A simple and popular method was 'overlaying', or smothering an infant who shared a bed with his parents. No doubt this occurred unintentionally when the parents were drunk or careless, but it happened so often that in. some countries parents were forbidden to have small children sleep in the same bed with them. Of all procedures the most unobjectionable was simply to leave the infant at some church entrance or in a secluded doorway, in the hope that perhaps someone would rescue him. Most of these abandoned children died quickly of exposure or starvation.
In the 18th century it was not an uncommon spectacle to see the corpses of infants lying in the streets or on the dunghills of London and other large cities. Infanticide in one form or another steadily increased in England during the 19th century.
Scores of accounts tell of dead infants found in rivers, canals and drains or lying under the bushes in parks and fields. Coroners' reports show that between 1855 and 1860 inquests were held on 3,900 dead children in the London area, most of them newborn. In 1,120 instances the finding was murder; in 904 others it was 'accidental' suffocation. In the 18 months ending in June, 1862, inquests were held on 5,113 dead children throughout England and Wales. The finding was murder in 902 cases, 297 of them in London. In 1862 one of the coroners for Middlesex stated infanticide had become so commonplace "that the police seemed to think no more of finding a dead child than they did of finding a dead cat or a dead dog". The 'Morning Star' (June 23, 1663) declared that infanticide "is positively becoming a national institution"; the 'Morning Post' (September 2, 1863) termed it "this commonest of crimes" (Langer, 1972).
The abundance of information on this overwhelming social problem leaves no doubt that late in the 18th century and early in the 19th fully 80 percent of the foundlings died in their first year, most of them in the first days of life. The Rouen records show that mortality in the hospice there regularly reached and at times exceeded 90 percent. Conditions were somewhat better in Paris. There in 1818 the Hôpital des Enfants Trouvés took in 4,779 babies. Of these, 2,370 died in the first three months and another 956 within the first year, so that the birth-through-12th-month mortality was less than 70 percent. The three winter months, says one commentator, regularly produced 'a veritable hecatomb'. There was no mystery about these matters. Mothers who left their babies in the box knew that they were consigning them to death almost as surely as if they dropped them in the river.
It seems reasonable in the light of this painful record to conclude that infanticide, covert or flagrant, was at least as important as celibacy in checking population growth from the middle of the 18th century to the middle of the 19th. It is also evident that a marked change of attitude with respect to children, particularly the newborn, took p1ace thereafter. What was then commonplace now seems intolerable cruelty. Yet the earlier attitude is not beyond understanding. The newborn bear little resemblance to more mature human beings or even to year-old children. Strong maternal feelings evidently are aroused only on the second or third day after giving birth, when nursing begins. Thus it is not surprising that among so many people at so many different times in history the disposal of unwanted babies has been accepted as a matter of course. In Europe from 1750 to 1850, at any rate, infanticide persisted in spite of all the denunciations of the church and all the penalties of the law (Langer, 1972).
The Destruction of Offspring: Dickeman (1975)
The destruction of offspring, far from being a uniquely human phenomenon, is widespread in living organisms. In plants, biochemical inhibitors may prevent germination and normal growth of conspecific seedlings except under propitious conditions; in a variety of invertebrates and vertebrates, egg cannibalism, aggressive neglect of young, and cannibalism of injured and dead have been reported as density-dependent behaviors.
There are reports of infanticide and cannibalism in nonhuman primates under apparently normal natural conditions, for example among hanuman langurs and chimpanzees (Sugiyama, 1967; 1973; Yoshiba, 1968), where infants were killed by dominant males. This is not, apparently, a regular practice in all troops of either species.
Cannibalism of infants in time of famine is of course known in some human societies, as among the Eskimos and Australians. Consumption of newborns, fetuses, dead infants, and even, in some districts, regular consumption of all firstborn has a scattered distribution in Australia (Krzywicki, 1934; Weyer, 1932). Infant cannibalism is known to have existed in Pleistocene man. But these instances, like the Aztec consumption of human sacrifices as a regular part of the upper-class diet (Cook, 1946; Padden, 1967), are exceptional. The mass of human infanticides is never consumed (Dickeman, 1975).
What is distinctive in the human species is selective removal on the basis of of conscious intent. The rationale varies from case to case, but the usual act of human infanticide is a product of individual or group decisions in a healthily functioning social system, and not at all the mere product of accidental proximity in conditions of social disorganization. It is a normative, culturally sanctioned behavior. This truism needs emphasis in the context of the current, often muddy, dialogue regarding human population increase. It reveals, bluntly, the irrelevance of the facile analogies of Calhoun (1962) and others to an understanding of human responses to increased density. The human act requires the capacity to discriminate between individual offspring, both one's own and others, in terms of a variety of cognitive categories. This capacity for selective removal in response to qualities both of offspring and of ecological and social environments may well be a significant part of the biobehavioral definition of Homo sapiens (Dickeman, 1975).
As Neel (1969) has suggested: "...man is to my knowledge the only (mammal) who regularly and without 'external' provocation purposely and knowingly commits infanticide... There have been many attempts to define that point at which our primate ancestors crossed the threshold to 'true' man. Most definitions involve tool-making or speech. I suggest that an equally sound definition is the point at which parental care evolved to the level of permitting rapid population increase, with the concomitant recognition of the necessity to limit natural fecundity".
Note that this capacity depends not only on cognitive discrimination, but on the relaxation of innate maternal attachment or imprinting mechanisms that allowed abandonment only in situations of extreme stress. This relaxation may also be a significant criterion of humanity (Dickeman, 1975).
If infanticide is an alternative to natural mortality, or to death through aggressive neglect or child abuse, then frequencies of directly postnatal events must be compared with those occurring long afterward in the life of the child or even adult. Abandonment likewise presents problems. Exposure is clearly a form of intentional killing, but may have other functions, as in the Netsilik Eskimo (Balikci, 1967) where the exposed infant announces by its cries that it is available for adoption by less fertile couples. In Western Europe until the mid-nineteenth century, abandonment, country wet-nursing, and foundling-home donation constituted systematic, if covert, means of infanticide (Langer, 1974; Trexler, 1973). Early students of population, as educated eighteenth and nineteenth century Europeans, were familiar withe the practice of infanticide from classical literature. Malthus (1807) reported its existence in Australia, the Pacific, North and South America, central Asia, India, and China, in addition to ancient Greece and Rome. He emphasized its occurrence in nonurban societies and Asian civilizations, however, and neglected the more covert European forms. Yet even in the Greek and Roman cultures, the importance of infanticide was lost sight of in Malthus's more general treatments, such as his 1830 summary, in part because he classified it with epidemics, famine, and warfare as a 'positive check'. Moral bias caused him to confound control of population growth with control of natality, preventing recognition of infanticide as a preventive check, along with delayed marriage and reduced coitus. In eddition,his uncertainty about its effectiveness led him to conclude that in the "ancient nations and less civilized parts of the world... war and violent disease were the predominant checks to... population" (Malthus, 1830). Thus began a long neglect of cultural controls on the part of demographers, for whom 'savage' society was typified by maximal natality and maximal mortality, resulting in moderate or low levels of increase (Dickeman, 1975).
Carr-Saunders (1922) was an early notable exception to this general view. In his first work he devoted much space to nonurban and non-Western societies, concluding that infanticide, abortion, and abstention from intercourse were widespread in human culture: "There is no indication of the correlation of any one practice with any one economic stage". His notion of 'optimum balance' was a theory of internal regulation culturally achieved; thus "the evidence shows that there is even among the most primitive races at times at least some deliberation as to whether a child shall be allowed to live... some semi-conscious adjustment of numbers...". Human societies maintained themselves at or below carrying capacity of their environments, whether by conscious or by unconscious means, although Carr-Saunders recognized that all were not equally effective and proposed the operation of some sort of cultural selection. But all men were distinguished from other animals by the restriction of fecundity (potential fertility) through artificial means: "reproduction among men is never 'mechanical'". However, Carr-Saunders's later work moved away from a concern with non-Western societies, end the impact of his first book was regrettably slight (Dickeman, 1975).
Until recently, such matters as cannibalism, warfare, human sacrifice, wife capture, torture, slavery, sexual mutilation, and infanticide were either largely neglected in anthropology or became the subjects of embarrassed dispute. Anthropology became, and to some degree remains, the twentieth century purveyor of the myth of the noble savage (Dickeman, 1975).
In recent years, several biologists interested in human ecology have come to positions similar to those of Carr-Saunders and most anthropologists. Cook (1947) long ago pointed to the minimal role of disease in regulating precontact North American populations, whereas Bates (1955) emphasized the likely antiquity of infanticide and abortion. Most influential at the moment is of course Wynne-Edwards (1962; 1964; 1965), who, in his search for homeostatic mechanisms, has declared himself the intellectual heir of Carr-Saunders. For him, primitive man's cultural customs, "conscious or not, kept the population density nicely balanced against the feeding capacity of the hunting range". The agricultural revolution, for some reason, resulted in the breakdown of human homeostasis. (Note that this contradicts Carr-Saunders). A paradoxical shift has occurred. 'Primitive' man is once again seen in terms of a biological model, but one no longer red in tooth and claw, derived rather from modern population biology, and permitting a role to formerly embarrassing customs. In Eden, Adam and Eve infanticidal lay down with the territorial lion and the socially hierarchical lamb. Expulsion from the hunting-and-gathering paradise, with the invention of agriculture, marked an end to homeostatic innocence of which Cain and Abel were the first fruits (Dickeman, 1975).
In the context of his general analysis of human population control, Carr-Saunders (1922) was the first to review evidence for infanticide on a worldwide basis. Available data on the small family size of 'primitive races' led him to explore, as possible explanations, lower fecundity as compered with civilized groups, mortality through disease and warfare, customary abstention from intercourse, abortion and infanticide. Reviewing a mass of governmental, travellers', and ethnographic accounts, he concluded that the primary means of limitation were the latter three practices. "...Everywhere among primitive races either abortion, infanticide or prolonged abstention from intercourse are practiced in such a degree and in such a manner as to have as their primary result the restriction of increase".
Unable to distinguish between hunter-gatherers and agriculturists in this regard, he was puzzled by the overpopulation of traditional civilizations such as India and China. Carr-Saunders proposed that cultural checks had ceased to be employed systematically in these societies, being used only in response to famine and other crises, and that this cultural breakdown was in some way a correlate of political oppression, aggravated by the influence of Christianity, resulting in a state of 'apathy and listlessness'. He noted the need to restrict numbers for nursing and mobility among nomads, and the widespread removal of defectives as a 'quality' control; he recognized the frequency of female infanticide, but offered no explanation.
Although no detailed analysis of forms and motives is given, Carr-Saunders's list of societies in which infanticide occurs is impressive. He believed the practice dated from the Upper Paleolithic. Only for two societies are quantitative data given, drawn from other sources. A study of 160 Chinese women over 50, who had borne 631 sons and 538 daughters, revealed that 158 of the females but none of the sons had been destroyed. Another calculation was that 20-30 % of all female livebirths in two Chinese provinces were removed. For India, Carr-Saunders repeated estimates of the annual destruction of 33,000 livebirths. Neither estimate is discordant with more recent ones. An attempt has been made to dismiss the evidence marshalled by Carr-Saunders as 'out of date' and the argument as 'very naive' (Douglas, 1966). Yet his sources are often detailed and his estimates of frequency and impact are confirmed by our few modern studies. Although a thorough review of early sources is long overdue, Carr-Saunders was probably correct in assuming that the positive bias of observers and the influence of Western contact had resulted in an underrepresentation of the practice in existing literature (Dickeman, 1975).
Aptekar's (1931) review of infanticide, abortion, and contraception is a rejoinder to Pearl's (1922; 1924) emphasis on the logistic curve in human population growth, and hence a defense both of the impact of cultural practices and of primitive rationality. His treatment of infanticide does not go beyond Carr-Saunders in use of sources, nor is he able to deal with the then-popular problems of depopulation and of the relation between polyandry and infanticide. Aptekar concluded that although societies at all stages of economic development engage in abortion, infanticide, and abstention from sex, infanticide is the 'primitive population check', due to a lack of precise knowledge of the fertilization process and to the lack of effective alternatives. Carr-Saunders's contention that primitive societies experienced lower fecundity than civilized ones was challenged in detail by Wolfe (1933). Wolfe regarded the low fertility of 'the primitive hunter' as a consequence of the 'low state of culture and ...high mortality', rather than culturally induced. Cultural checks, he believed, appeared only with the advent of agriculture. Carr-Saunders's evidence for infanticide was ignored. Adherence either to the anthropological or to the biological-demographic model seems to have more to do with these interpretations than do available data (Dickeman, 1975).
The first significant contribution to anthropological demography after Carr-Saunders is Krzywicki's (1934) Primitive Society and its Vital Statistics. Although the theoretical framework is strongly unilinear, that perspective has little impact on his painstaking demographic analyses. Although he deals with group size, warfare, depopulation, dependency ratios, and marriage rates, his most valuable contribution is an extended analysis of female fertility, childhood mortality, and family size in hunter-gatherers and agriculturalists. The data on family size go far beyond Carr-Saunders, but confirm the latter's original contention: on the average, from two to three children are reared by hunter-gatherers, from three to four by North American agriculturalists, and from two to three by African agriculturalists (Of course, effective natality, completed family size, and survivorship are often conflated in early sources, but their consistency is striking).
Discussing the means of limitation, Krzywicki offered the first functional analysis of infanticide, based on the imperatives of long lactation end mobility in hunter-gatherers, and of long lactation even in sedentary groups. In so doing he discriminated between systematic infanticide for child spacing and those more occasional and demographically less significant forms such as removal of defectives, twins, and illegitimates, and idiosyncratic motives. Analysis of Australian sources is especially detailed. Regional differences, for which we still lack explanation, emerge clearly in preferences for one form of limitation over another: abortion in North and South America (except Eskimo) and infanticide in Australia; the rarity of both customs in agricultural Africa, except in the case of twins and defectives, is apparent.
Krzywicki also gave the first serious attention to distorted sex ratios in these societies, but was unable to offer either causal or functional explanation. In some ways this work is unique: use of sources is cautious and demographically informed, and the appendix is a valuable entré to early materials. Krzywicki's appreciation of the role of conscious economic motives in population control is important, as is his ability to discriminate between society's ideological imperatives regarding limitation and the emotions of the individual actors who must implement them (Dickeman, 1975).
A cross-cultural sample of 64 societies, later enlarged to 200, was used by Ford (1945; 1952) in his analysis of human reproduction and birth control. The two publications form a useful introduction to cultural behaviors and attitudes, although they do not provide a functional or demographic analysis, nor are the data handled quantitatively. Ford's discussion of the apparently contradictory nature of attitudes toward infanticide, one of the sources of Aptekar's confusion, is useful. While it may be said that infanticide is both universally practiced and universally condemned, these generalizations refer to several distinct ideological domains. Both abortion and infanticide are practiced occasionally in every society, but may be strongly disapproved deviant acts: of Ford's early sample of 64 societies, 19 expressly forbade infanticide. Yet even where allowed, "...it is also clear that strong social pressures are brought to bear against any excessive indulgence in abortion and infanticide. Most societies permit the riddance of the child under specially defined circumstances only. Generally... both abortion and infanticide come under rather strict social control". Ford offered two explanations, the first concerning the regulation of homicide: "...infanticide is most readily condoned if it occurs before the infant is named and has been accepted as a 'bona fide' member of its society. It seems that the primary and fundamental restriction in most societies is the taboo on murder, i.e. killing a member of the ingroup. The less eligible a child is for membership in the group, the less seriously the act of killing the child is viewed".
Ford's other explanation is of greater import to human biology. He stated: "The widespread insistence upon not killing newborn infants suggests a tendency on the part of mothers to do away with their own babies. Were this not the case it would be difficult to explain the very general development of sanctions directed against infanticide. Nor, indeed, does it seem unlikely that a women should occasionally wish to kill her offspring... this would be especially true after she had fulfilled the obligations imposed by society and had already borne and reared a number of children... The tendency to practice infanticide does not contradict the assumption that women instinctively desire to rear and protect their young. The facts do support the view, however, that the maternal instinct, if indeed there be such an instinct for human beings, is not nearly strong enough to counteract unaided the tendency to destroy unwanted infants" (Ford, 1945).
Nag's (1962) study of human fertility is an extremely valuable synthesis of physiological and ethnographic materials, including a cross-cultural sample of 61 societies. However, he does not discuss infanticide in detail, perhaps from lack of confidence in Carr-Saunders's and Krzywicki's sources, but more likely because his taxonomy of factors influencing fertility is derived from the work of sociological demographers, who naturally do not include it. Nag concluded: "The circumstances under which infanticide is or was practiced generally in these societies suggest that the mortality due to this factor is negligible in comparison with the mortality due to natural causes".
Incidental data on infanticide are found in Whiting's (1969) article: 'The effects of climate on certain cultural practices', an explanatory hypothesis for the tropical distribution of circumcision, based on cross-cultural correlations with such factors as polygyny, patrilocality, protein deficiency, and postpartum sex avoidance. Whiting's assumption is that postpartum abstention is a means of preserving both quantity and quality of milk supply to the nursing infant. "Infanticide is as efficient as a postpartum sex taboo for solving the problem posed by nomadism, but only the latter solves the problem of protein deficiency". Consequently, infanticide is not involved further in the hypothesis; however, a tabulation is provided. Of 88 societies that could be coded for it, 15 lacked infanticide, 8 practiced it (although it was unascertainable whether for child spacing or not), 54 practiced it but not for child spacing, and only 9 used it as a means of spacing. Whiting recognized that his sample is skewed in favor of Africa and North America. Granting sampling weaknesses, this distribution suggests the necessity of a search for ecological and societal correlates of infanticide in terms of the specific functions that it serves, rather than as a unitary phenomenon (Dickeman, 1975).
An initial step in this direction is Grantzberg's (1973) treatment of twin infanticide. Pointing to the inadequacy of explanations based on ideology, Grantzberg tested the hypothesis that twin infanticide is a response to the burden of caring for two infants simultaneously. He used a sample of 70 societies rated in terms both of availability of childrearing assistance to the mother, as inferred from family size and settlement patterns, and of the mother's freedom from work responsibilities, as measured by subsistence duties and childcare tasks such as nursing and carrying.
A significant correlation with absence of twin infanticide emerges; indeed, only 2 of the 33 societies scored as having no difficult conditions actually practice twin infanticide. However, of those that do suffer "conditions that would cause a mother to disrupt social patterns if she attempted to rear two children at once" only 16 out of 37 practice twin infanticide. Clearly Grantzberg's results support the contention that maternal workload is a significant causal factor, if they cannot yet explain how some other societies manage maternal burdens by other means. "Contrary to the common practice of treating infanticide as one monolithic institution,... (it) is... rather a heterogeneous phenomenon consisting of different practices, each aimed at solving different problems...". The analysis confirms Carr-Saunders's view that infanticide is a response to the "difficulty of transporting and suckling", but demonstrates, as did Krzywicki's studies, that these burdens and their resolution have much wider application than merely to nomadic societies. We must therefore demur from Sussman's (1972) contention that "once human groups became sedentary, social spacing mechanisms were no longer a necessity". Dickeman (1975) thoroughly discusses the removal of defectives, control of numbers, and manipulation of the sex ratio.
Removal of defectives
Maintenance of the genetic health of a population through removal of congenital malformations at birth would appear on first thought to be a likely function of infanticide. Many sources report such a motivation; Ford (1945), for example, stated that such removal prevents the development of useless dependents in societies without specialist roles for them and "discourages the transmission of hereditary strains toward certain abnormalities". However, further consideration suggests that the role of infanticide as a means of artificial selection is equivocal at best. It is estimated that 5-6 % of all births have defects partially or wholly of genetic origin, and that 5 % of all conceptions result in major congenital defects. Incidence of major malformations identifiable at birth, based on Caucasian, Japanese, North American Indian, and American Black series, is about 1 %, ranging from 0.63 to 2.44 (Neal, 1958; Salzano & Freire-Maia, 1970; Woolf & Turner, 1969). Such rates would appear to present adequate opportunity for effective artificial selection.
Unfortunately, the relevance of these figures is not great. Series are to some degree incomparable, since they may rely on registration data, examination at birth, or hospital nursery records. The first are of course subject to underreporting while examination at birth, as in Neel's Japanese series, includes both stillbirths and livebirths. Only nursery records, as in Woolf & Turner's Utah study, exclude stillbirths, which are irrelevant to infanticide. Thoroughness of examination and methods of coding also vary from study to study. These figures are all products of examination by personnel trained in Western medicine, whereas any assessment of general potential for selection must refer to malformations detectable at birth by an indigenous midwife or experienced mother without benefit of Western training. An inspection of the classification of malformations devised by Neel (1958) suggests that the majority of medically major defects is not readily identifiable by the layman. Most musculoskeletal and nervous system complications would be superficially recognizable, while only some of those involving digestive and urogenital systems would be; the greater number of abnormalities of the respiratory, cardiovascular, and lymphatic systems, and of the sense organs, would not be recognizable at all. Additional testimony to the assumption that the majority of defects is not available for normal human intervention at birth is the number of additional defectives uncovered both through autopsy of neonatal deaths and nine-month follow-up of livebirths in Neel's study.
Neel identified that "group of malformations readily diagnosable at birth under almost any conditions" and "sufficiently frequent that their incidence may with meaning be compared with the findings in other extensive series" as: anencephaly, spina bifida manifesta, harelip-cleft palate, atresia ani, anophthalmos-microphthalmos, and polydactyly. Rates of incidence for these six major defects range from Neel's 0.49 (Japanese) to McIntosh's 1.23 (New York city, as recalculated by Neel). A Brazilian study reported a frequency of 0.5 % for the 'severest cases', listed as "harelip-cleft palate, club foot, limb bone aplasias, congenital blindness, polydactyly, mental retardation, albinism, etc." (Salzano & Freire-Maia. 1970). (Mental retardation and blindness are, however, rarely identifiable by the layman at birth). With the exception of harelip-cleft palate, polydactyly, clubfoot, and albinism, much of this readily observable group will not live beyond the first year; the likelihood of perinatal death in Neel's six categories is extremely high. And even for such traits as cleft palate, survival without modern medical intervention may be difficult (Dickeman, 1975).
Populations do of course vary, both in the incidence of specific congenital defects and in the resulting totals. We would expect a higher incidence of defectives both presented and removed in populations with high inbreeding coefficients. Consanguinity does seem to increase the occurrence of defectives and hence the frequency of multiple malformations, but this only elevates the level of lethality on the average, thus making artificial selection less necessary. In any case, the effect of consanguinity (other than that involving primary kin) in controlled studies is small (Neel, 1958; Neel & Salzano, 1967; Schull & Neel, 1966).
Available demographic studies do not help us much in regard to this question, but they do suggest that infanticide has limited effects in small populations. Thus, in the highly endogamous Xavante (Neel et al., 1964; Salzano, Neel & Maybury-Lewis, 1967), out of a total of 287 individuals of all ages in two villages, seven congenital defectives were identified (one with polydactyly, one with clubfoot, three mental defectives, and two or three with possible congenital heart disease) (Weinstein, Neal & Salzano, 1967). Although infanticide is likely in this group, if defectives are already being removed by this means, the rate at birth must be unusually high. Similarly, of 274 Yanomamö, one achondroplastic dwarf and six cases of congenital heart disease were observed. Yet these defectives are believed to be a "residue of a relatively high proportion of defective children" removed by infanticide or early death, in a society which admits removal of defectives. Mission birth records for other Yanomamö villages likewise contained five major congenital defects out of 93 births (Neel, 1969; 1971).
Another small society, the Buang of New Guinea, which formerly practiced infanticide against defectives, contained eight living dwarfs in a population of 9000 (Malcolm & Zimmerman, 1973). The authors reported, "although it was difficult to obtain a clear and coherent description of dwarfs during childhood, the people stated that the condition was in most cases apparent at birth, both due to the associated deformities and the small size". It is interesting that such individuals, although fully accepted members of society, are not permitted to marry. This injunction may, however, reflect society's assessment regarding the defective's ability to assume the normal socioeconomic sex role, rather than an attempt to prevent genetic transmission.
Woolf & Dukepoo's (1959) study of Hopi albinism explains the high rate of this condition as a result of high extramarital fertility of affected males who spend their days as craftsmen in the village, in a society which does not recognize a hereditary basis for the condition, and which, while it does not find these individuals acceptable marriage mates, does find them attractive as sexual partners. Among the Peruvian Cashinahua, albinism is reported to be intentionally removed as spiritually dangerous. One albino woman in her forties, was identified in a demographic study of five villages totalling 206 people. Of 15 pregnancies, she had terminated 7 in abortion or infanticide. (This scarcely suggests avoidance of the spiritually dangerous lady). The authors further observed two infants, one with spina bifida and one with syndactyly, which would have been killed at birth in the past but were saved as the result of Western influence (Johnston et al., 1969).
As stated, populations differ in the incidence of specific malformations. But, while the Ashanti may have spared hunchbacks to become court jesters (Ford, 1945) and the Hopi male albino had a role in the village as a weaver, it is difficult to see any relationship between the present incidence values in most populations and previous selective infanticide. High rates of anencephaly and spina bifida in the British Isles, of polydactyly in US Blacks, or of harelip-cleft palate in Japanese make no sense as artifacts of cultural selection, at least at present.
Finally, the relation of this puzzling situation to reproductive compensation and the dependency ratio may be mentioned. If a defective infant, with short life expectancy, lowered economic efficiency, and lowered fertility, is preserved through investment of nursing and general childcare, we may speak in theoretical terms not of fetal but of 'live wastage'. This is especially so in societies that observe postpartum sex avoidance, in which case only two years of life for a defective child would significantly delay the return of the mother to reproductive status. One would anticipate that the burden of numbers of defectives in subsistence societies would be too great to bear, but such does not appear to be the case. Weinstein, Neel & Salzano (1967) have remarked that, given the relative infrequency of chronic and degenerative diseases in small societies, "congenital defects may play a relatively greater role in the medical burden of the community than in contemporary, more civilized groups".
We need much better measures of dependency ratios and energy investments in human economic systems. But a consideration of the data reviewed above leads one to wonder whether what appears dysfunctional on the surface may not in fact be adaptive. The maintenance functions of genetic systems are not reducible to simple input-output models. On the genetic level, Neel (1958) proposed:
"The similarity of malformation frequencies in such diverse populations as Japanese and European thus finds an explanation in the fact that there is a malformation frequency representing the optimum balance between, on the one hand, fetal loss and physical handicap from congenital defect, and, on the other, population gain from those very same genes which in certain combinations may sometimes result in congenital defect. The differences between populations as regards the frequencies of specific defects would seem to indicate that within the framework of this optimum figure, different populations have evolved genetic systems differing significantly in their details".
On the level of cultural systems, a parallel proposal may be made, if somewhat hesitantly. Perhaps some level of 'live wastage', some negative cost-benefit ratios in the production of young, are tolerated as a means of reducing overall natural increase. Insofar as a population may possess excess reproductive potential in relation to available resources, some of those scarce resources might be channeled into the production and support of individuals of low potential, and the devotion of productive capacity to their support might have some effect in reducing rates of natural increase. In regard to congenital malformations, then, most infanticide would appear to be no more than the disposal of some excess population of low productivity, most of which is in any case destined for early mortality. Available demographic and ethnographic data are not likely to assist us much in testing this rather surprising hypothesis. However, on the basis of what we now know, the impact of infanticide on the incidence of congenital malformations appears to be, in most cases, insignificant (Dickeman, 1975).
Regulation of numbers
The human potential for excess reproduction, a property of all successful species, is of interest both in terms of the history of human expansion from an early African homeland and of the more general question of the evolution of self-regulating mechanisms. Estimates of human growth potential have been based on the physiological maxima of fertility, observed natality rates, and maximal known rates of increase. Lorimer's (1954) fecundity model produced a maximum of 8.32 births per completed reproductive span, but this is too modest, as the hypothetical potential of the human female is about 15 children, even allowing for the effect of lactation on ovulation. Recent natality studies give mean rates of the number of children ever born from 7 to 10, the most famous being that of the North American Hutterites (Eaton & Mayer 1953; Tietze, 1957), which is 10.6 in spite of a mean female age at marriage of 20.9 years. However, these modern populations enjoy at least some benefits of modern medicine, and so do not provide us with the best notion of past human rates of increase (Dickeman, 1975).
That human growth can be explosive is well demonstrated by Birdsell's (1957) analysis of the rates of modern subsistence societies in unoccupied areas. His four cases (Tristan da Cunha, Pitcairn, Bass Straits Islands, and the Australian Nanja Maraura) underwent one-generational doubling in their early histories; more recently Birdsell (1968) has revised this to a tripling within a thirty-year generational period. These recent estimates support Malthus's (1798) original proposal, based largely on North American colonial data, of a potential human doubling time of 10-25 years. Such growth must soon initiate either internal or external limitation.
It is natural in the historical context discussed above that living hunter-gatherers should have received the most attention as possible models for early human response to the limits of growth. Krzywicki's (1934) extended analysis of Australian infanticide includes a computation of the contribution of infanticide to total infant mortality from early sources, noting correctly that infanticide affects offspring who would in any case die in infancy of natural causes, and is thus substitutive. Australian societies removed on the average every child above the desired number of three per family. Based on estimates of natality, spacing, and the short reproductive span of Australian women (natality being rare after 35), the infanticide rate is 20-40 % of livebirths. Any revision of Krzywicki's assumption of a short reproductive span of 15-20 years would result in an increase in this proportion. Assuming natural infant mortality in such conditions as 35-40 % without the intervention of infanticide, systematic removal for child spacing would raise total infant mortality to about 45-50 % of livebirths, according to Krzywicki. However, this overlooks the role of child spacing in reducing some mortality by improving conditions in infancy.
Birdsell (1968; 1972) also emphasized the importance of infanticide in unspecialized hunter-gatherers and its role as a response to infant transport problems, based on extensive research on aboriginal Australia. In a recent summary (1968), he reported the adult sex ratio in precontact genealogies as 150:100; this, and the evidence of early sources, led him to estimate a 15-50 % infanticide rate, imposed through three year child spacing. "There is reason to believe that the limiting forces involved in the necessary spacing of children to be saved would apply broadly to all hunting groups, although sex preferences might vary".
The problems of mobility and lactation are induced by a complex of environmental and technological factors preventing both sedentism and early weaning. Infanticide is thus an indirect response to environmental limitations, culturally mediated. If so, then we might well posit moderate levels of increase for all mobile hunter-gatherers. The populations from which estimates of maximal human natality are derived are sedentary, with the exception of Birdsell's Maraura case. This means that explosive colonization based on their high rates may not be the best model for the spread of early man, although it may well be appropriate for early populations in areas of abundance that allowed sedentary collecting, at least until other limiting factors intervened (Dickeman, 1975).
Recent field investigations of Kalahari Bushmen have raised more questions than they have answered. The most thorough demographic study (Howell, 1975) involved census of 840 Dobe !Kung Bushmen and interviews of all women over 15. On the basis of childhood life tables, Howell estimated a life expectancy of 32 years at birth and an annual growth rate at 0.5 %. Her study agrees with others regarding late menarche (average 15.5 years), and even later age at first livebirth (19.5 years), she also reports a long nursing period of 4-5 years (other reports range from 2½-4 years: Lee, 1972; Marshall, 1960; Schapera, 1930; Silberbauer, 1972), and a high female marriage rate, only two mental defectives being unmarried. However, Lee's (1959) !Kung sample had an unusual and unexplained sex ratio in all age groups favoring females. Some sort of birth-spacing mechanism may well be involved in producing resulting low rates of completed fertility (Aptekar, 1931; Balikci, 1967), but what it may be is unclear. Both Howell (1975) and Lee (1969; 1972) maintain that infanticide is uncommon, practiced only on defectives, on the death of the husband, or "sometimes when one birth follows another too closely" (Howell, 1975); in Howell's population 6 out of a total of 500 livebirths had been destroyed. She believed that neither disease nor malnutrition were significantly involved. Consequently, explanations based on unusually long suppression of ovulation through lactation (Lee, 1972) or control of ovulatory cycle through fat intake (Frisch & McArthur, 1974; Kolata, 1974) have been proposed.
However, other Bushmen ethnographies suggest a different interpretation, according to Dickeman (1975). Marshall (1960) reported that Nyae-Nyae !Kung space their children 2-4 years apart through infanticide to ensure lactation during that period; Schapera (1930) likewise reported systematic infanticide for reasons of nursing and mobility, but not on defectives, removing one or two out of every three children 'without exception'. Early sources in Krzywicki concur. This implies an infanticide rate of at least 50 %. The evidence of postpartum abstinence is less consistent, varying from none (Schapera. 1930), to one year's abstinence (Lee, 1972), to abstinence until weaning (Silberbauer, 1972).
The sedentary Central Pomo of Northern California practiced infanticide, in addition to abortion, against all excess births over the family ideal of two or three, as well as against twins and illegitimates. The everpresent fear of food shortage as a motivating factor is clear from an early study; unfortunately it provides no quantitative data (Aginsky, 1939).
Horticultural societies have also received little demographic treatment until recently. Still in many ways most valuable is Firth and co-workers' study of Tikopia, based on fieldwork in 1928-1929 and again in 1952-1953 (Borne, Firth & Spillius, 1957; Firth, 1957; 1959; 1965; Spillius, 1959).
In Tikopia, the ideal completed family contained two or three children. Systematic controls were coitus interruptus, abortion, infanticide, male and female celibacy, suicide (including overseas voyaging), and, on two occasions in remembered history, warfare between kin groups and mass expulsion.
According to informants, infanticide was a major family means: "A male and a female are born; they are allowed to live; two only... If another child is born also it is buried in the earth. And two only are left, corresponding to the scarcity of food" (Firth, 1957). Firth noted only one case in 1929, with about half of the island Christianized; his informants were contradictory on the matter of sex preference. Borne's analysis of sex ratios by cohort showed distortions as great as 153:100 in 1929 (ranging from 110-153:100 for cohorts under 27), and persisting though less extreme in 1952 (106-125:100). As Borne stated, this persistent imbalance implies female infanticide in most years; a rate of 10-30 % of female livebirths is not unreasonable (Borne. Firth & Spillius, 1957). During the famine, Spillius (1959) recorded 62 pregnancies, of which 14 failed to terminate in livebirths. He believed the majority of these were deliberately destroyed by induced abortion or infanticide. Four certain cases of abortion-infanticide were identified.
Mortality data for 1929-1952 include 16 female suicides out of a total female mortality of 187. More significant is the practice of overseas voyaging by males, regarded by the Tikopia as tantamount to suicide, and engaged in for adventure, from personal shame, rejection, or as a legal banishment imposed by the chiefs.
Demographic literature on nonindustrial aggrarian societies is immense, but infanticide, though widespread, has been little studied. Japan has received both ethnographic and demographic treatment, especially in Taeuber's (1958) classic demography. National censuses for early periods of Japanese history indicate slow growth from the thirteenth to sixteenth century, as arable lands were increasingly brought into production, but political instability and recurrent earthquakes, epidemics, and famines resulted in high mortalities. Population increased rapidly in the early Tokugawa period, under a more unified government, then remained stationary for about 150 years, from 1700 to 1852, varying, in national censuses introduced at this time, little more than 10 %. Several constraints may be identified: isolation was imposed not only by distance from a hostile mainland, but by political exclusion of most foreign agents and by the prohibition of emigration. Thus neither external outlet nor energy inflow through trade or conquest were available. By about 1750, limits of arable land had been reached. With heavy rice taxation and loss of lands to the growing merchant class, levels of farm debt rose. Escape to the cities was an alternative only for few, and mortalities there were high. Malnutrition and famine were common (Bowles, 1953; Taeuber, 1958).
In this situation, a variety of techniques of population limitation was used, varying by region and class. Contraception seems never to have been important. Celibacy was not significant in a society where marital rates were high in all classes, but delay in age of marriage was important for some, especially the samurai. Childbirth was generally disapproved of for women over 40 or for those who had daughters-in-law. Abortion, both chemical and mechanical, was widespread. This practice was favored by noble and samurai classes and by urbanites, probably as much to destroy illegitimates as for limitation per se, although the increasingly impoverished samurai were often forced to prevent excess births. Abandonment was also common, but its extent is unknown. For peasants, 90 % of the population, the preferred means was infanticide. This practice is old in Japan, with references occurring as early as the ninth century, but its use apparently increased in late Tokugawa. Peasant preference rested on several considerations. Abortion was expensive or risky, and illegitimates could be socially assimilated, so concern for concealment was less. Infanticide allowed removal of defectives and twins. Most importantly, it permitted manipulation of the sex ratio to guarantee patrilineal perpetuation of the localized peasant family.
According to Dickeman (1975), data indicate a mean number of children retained somewhere between 4 and 6. Assuming a reproductive span of 25 years (ages 15 to 40), this would imply limitation, by whatever means, of almost 50 % of potential livebirths, out of a maximum potential of about 8. (This crude estimate could be refined by analysis of age-specific female mortalities). We must then assume an even lower completed family size for the late Tokugawa period. Sex ratios imply maximum removal of 50 % of all female livebirths, but an overall mean of 10-30 % of females. Remembering that some males were destroyed, and that infanticide was the primary means for the majority of the population, a late Tokugawa infanticide rate of 10-25 % of livebirths is not unreasonable. Historical assertions that 30,000 children were removed annually by 100,000 households in Chiba Prefecture, or that two out of five livebirths were destroyed in Kyushu, become more believable.
Manipulation of sex ratio
The studies reviewed above reveal the role of infanticide not only in control of numbers, but in the production of distortions in sex ratios as well. Preferential male infanticide is extremely rare in the ethnographic record, though it occurred in Carthage (Weyl, 1968) and in some East African groups (Douglas, 1966). Obviously it merits further investigation. Female infanticide, on the other hand, is fairly common, as Carr-Saunders (1922) and Krzywicki (1934) have attested, resulting in significant sex ratio distortions in early age groups. Similar distortions occur in some Pleistocene human populations (Vallois, 1961). Its structural importance is dramatically demonstrated in the following cases. Eskimo hunter-gatherers from Alaska to Greenland practiced preferential female infanticide at rates resulting in childhood sex ratios up to 200:100, averaging around 125:100. These rates were offset by heavy adult male mortality from hunting accidents, homicide, and suicide resulting in most cases in parity by middle age, and a disproportion of females in older age groups. Moreover, optional polygyny intensified the shortage of females, and encouraged conflict between males, wife stealing, and prepubertal betrothal (Balikci, 1967; Weyer, 1932). Among the horticultural Yanomamö and Xavante of South America, sex ratios in the 0-14 year cohort are 128.6 and 123.7 respectively; in the Peruvian Cashinahua they reach 148:100. Infanticide is known for Yanomamö and Cashinahua and suspected for the Xavante. Yanomamö male aggression, both inter- and intragroup, produces 24 % of all male mortality, resulting in a more balanced total sex ratio of 110.5 (Xavante 115.1). Again, competition for mates is aggravated by polygyny and encourages child betrothal and wife capture ( Chagnon, 1968a,b; De Oliveira & Salzano, 1969; Johnston et al., 1969; Neel, 1969; Neel et al., 1964; Salzano & Cardoso de Oliveira. 1970). Similarly, the horticultural Bena Bena of New Guinea suffer a shortage of females (no sex ratios given) resulting from female infanticide and polygyny, encouraging wife capture and a 30 % rate of pre-pubertal betrothal (Langness, 1967).
Preferential female infanticide in agrarian societies is represented not only by the Japanese case. The Jhareja Rajputs in northwestern India removed virtually 100 % of female livebirths prior to British suppression of the practice. This practice occurred in the context of a hypergynous marriage system within the endogamous Rajput caste, which was composed of ranked exogamous subcastes, of which the Jharejas were the highest. Destruction of Jhareja females guaranteed the upward movement of both brides and dowries from lower ranking subcastes. Again, female shortage was aggravated by polygyny in the higher classes.
A similar hypergynous system probably occurred at least in part of preindustrial Europe, judging from the recent analysis of fifteenth century Florentine materials (Trexler, 1973). There, sex ratio distortions were greatest in the upper class (124.5) and clear evidence of selective aggressive neglect and abandonment, if not outright infanticide, is provided. The importance of the dowry in traditional Europe is well known.
These cases could undoubtedly be multiplied by reanalysis of primary historical and ethnographic sources. Their significance of terms of the model of male competitive aggression and reproductive success proposed by Trivers and others is apparent (Alexander, 1974; Crook, 1972; Fox, 1972; Trivers, 1972; Trivers & Willard, 1973). Female infanticide guarantees the operation of that model; in stratified societies it magnifies the natural discrepancy in primary sex ratios between upper and lower classes that is a result of the influence of socioeconomic status and birth order (Teitelbaum, Mantel & Starr, 1971). Granting the noncomparability of these materials in some respects, and the larger numbers of currently unanswerable questions they raise, they do confirm that preferential female infanticide operates in a variety of human socioeconomic systems as a significant contributor to the maintenance of social structure at rates ranging from 10-100 % of female livebirths per social unit (Dickeman, 1975).
She agrees with Nag (1962) that traditional oral contraceptives are rarely effective and abortion is ordinarily of little demographic importance due to the high risk of mortality or sterility to the subject. In most societies, the effective means of abortion are direct physical trauma to the cervix and external trauma resulting in expulsion of the fetus as a stillbirth at or very near term. In the former, risks of sepsis (with mortality rates as high as 25 %), tetanus (mortalities of approximately 50 %), and massive hemorrhage are great. As for the use of external trauma, little is gained over natural parturition except increased maternal risk. In consequence abortion serves mainly for the concealment of illegitimates, except in societies with strong prohibitions against infanticide. Truly effective methods of systematic control other than lactation effect are thus limited to abstinence through postpartum avoidance or chaste celibacy, incomplete coitus, and infanticide, grading into abandonment and early aggressive neglect.
Both infanticide and postpartum avoidance occur in polygynous societies, either in combination or separately. Female infanticide occurs both where women contribute little to primary production and where they contribute much. Likewise, systems of hypergyny may depend on the practice of differential infanticide in upper classes, or merely on socioeconomic differences in sex ratio and perhaps in female celibacy; they may provide productive alternate roles to males denied mates, or they may produce large numbers of unproductive lower status males who must be controlled by repression. Our explanatory difficulties arise not so much from the inadequacy of existing data as from the fact that differing socioeconomic necessities may produce similar adaptive responses, and vice versa.
Population control practices in man exist, moreover, in culturally variant complexes of cognition and value, as Rappaport's (1968; 1971) exploration of the cybernetic functions of ritual systems in relation to population dynamics implies. Douglas's (1966) critique of Carr-Saunders and Wynne-Edwards emphasized the role of prestige consumption in motivating population control, but failed to specify the nature of the link between the two. Extending Rappaport's approach, the exchange and consumption of prestige items probably plays, as does ritual, an informational and regulatory role, relating the level of available resources to competitive and other demographic processes through socially significant symbols. Meggitt (1972) has demonstrated the operation of such a system for the Mae Enga, while Lindenbaum's (1972) analysis of the idea of female pollution in New Guinea defined a network of ideological components, from sorcery to attitudes toward sex, within which the practice of female infanticide is enmeshed, and proposed a relation between intensity of the ideological complex and population density. These analyses indicate a direction for future research: attention to the density-dependent functions of systems of exchange, ritual, and warfare must now be integrated with data on indirect and direct means of population control. If human societies are adaptive, all these must be parts of a single overarching cybernetic system, operating on both conscious and unconscious levels (Dickeman, 1975).
Moreover, Dickeman points out that ideological disapproval and even shame and secrecy are by no means incompatible with significant rates of infanticide. If ideological, that is, learned behavioral, systems have adaptive value as contrasted with innate behavioral systems, they do so because of their flexibility, their rapidity of response to changing conditions. Thus the ambivalence and inconsistency that characterize all human thought are probably powerfully adaptive, providing immediate response to the needs of the moment by permitting appeal to alternate available rationales. Tolerance of infanticide provides a rationale in situations of familial economic stress, whereas its disapproval supports and encourages nata1ity in improved economic conditions. It is no surprise that the more public, authoritative ideological statement is always that which encourages fertility: species success as measured by dispersal and colonization depends upon it. But the inherent flexibility of inconsistent ideologies allows fine tuning in relation to external variables without requiring readjustment in the ideological structure itself (Dickeman, 1975).