Trouble in paradise. Legacy review of: War before civilization. By Lawrence H. Keeley, New York, NY: Oxford University Press. 1996. xiv + 245 pp. ISBN: 0-19-509112-4 (hardback)

Only occasionally, does a book catalyze a field of study—Larry Keeley's (1996) War Before Civilization is one of that rare breed. For several decades leading up to the 1990s, archaeologists in the English-speaking world, especially North America, paid scant attention to conflict involving the prestate, kin-based societies that characterized the great majority of human existence. That is not true today. Warfare has emerged as a major focus of research, with much of the recent work consisting of detailed descriptions of skeletal trauma, defensive works, weapons, and conflict-related artwork. They far outnumber insightful excursions into conflict's role in past community life or the evolutionary processes that led to the world as we know it today.

The 1990s transformation—it centered on Keeley's (1996) book, as well as the work of a handful of other researchers (Haas & Creamer, 1993; Lambert, 1997; LeBlanc, 1999; Milner et al., 1991; Wilcox & Haas, 1994)—boiled down to how archaeological data were perceived.¹ Blinkered by naively idyllic visions of life in distant times, it was once possible to ignore ethnographic and historical descriptions of conflict in places such as highland New Guinea (Heider, 1970) and tropical lowland South America (Chagnon, 1968, 1988). Archaeological evidence of intergroup conflict, never in short supply, could be fobbed off as something else entirely, notably some form of otherwise unexplained ritual behavior. Interpretations owed more to researcher preconceptions than to what was actually observed.

Three decades ago, resistance to the thought that prestate societies were tainted by warfare was given intellectual heft by Ferguson and Whitehead's (1992, p. 3) “tribal zone.” Within that zone, people who had previously lived in harmony were reeling from direct and indirect contact with much larger, more powerful, and overtly exploitative nation states. This critique was part of a larger concern about the uncritical use of ethnographic and historical accounts as proxies for past societies known through artifacts, architectural features, and the like. Archaeologists were accordingly challenged to come up with their own data on interactions among societies classified as bands, tribes, and chiefdoms.² Baldly stated, 30 years ago it was not known whether these societies in precontact times regularly fought one another or not. Tarted up with modern language and sensitivities, ruminations about what took place were scarcely more than a reprise of a centuries-old debate over whether human existence was perpetual Hobbesian warre in the absence of “that great Leviathan called a Commonwealth, or State” or a Rousseauian Eden (Hobbes, 1962[1651], p. 19). Keeley (1996, p. 18) believed archaeologists had “pacified the human past” through “studied silence or fashionable reinterpretation.” He was not wide of the mark.

Conflicts encompass a wide range of behavior from verbal abuse to physical, and at times lethal, violence aimed at members of one's own social group up to people belonging to entirely separate societies. Various forms of violence, when they took place, and what constitutes evidence for them have now become a significant part of archaeological, including skeletal, research (Klaus, 2012; Knüsel & Smith, 2014; Lambert, 2012; Martin & Harrod, 2015; Scherer, 2021; Webster, 1993). For his part, Keeley (1996) focused on warfare.

Sticking solely to warfare is not as easy as it might sound. There is no consensus among anthropologists, or other scholars, over what constitutes war. As the humorist Will Rogers (1929) remarked about a decade after the Great War, “[y]ou can't say civilization don't advance … for in every war they kill you in a new way.” Therein lies a definitional problem because prestate societies lack standing armies, with all it might imply in specialized weaponry, training, disciplined formations, logistical support, and fixed systems of command and control. Keeley (1996, p. 41), echoing his book's title, asserted that “[p]erhaps no aspect of prestate societies has been treated with more condescension by civilized observers than the way such groups have conducted their wars.”

That gets us to what exactly constitutes war. For the wide array of structurally dissimilar societies known to archaeologists, it is best to avoid a quagmire of overly fine distinctions about how conflicts played out, who participated in them, the numbers of people injured or killed, and the motives that led to fighting in the first place. Warfare for Keeley (1996, p. x) was “armed conflict between societies.” But he meant rather more than that.

Keeley (1996), among others such as Vayda (1974) and Webster (1993, 1998), regarded warfare as culturally sanctioned and potentially lethal violence undertaken by members of a group with shared political, social, and economic interests who direct their aggression toward other spatially separated communities.³ The group membership part is important. Victims of attacks are representatives of a community defined as an enemy; they are not targeted as specific individuals. That is, violent attacks are collectively recognized as justified. They are not simply murder.

Potentially lethal confrontations can involve raids organized with the express purpose of carrying out an attack or opportunistic assaults on situationally vulnerable people. It is not unusual for a series of ambushes of targets of opportunity to be periodically punctuated by large-scale and carefully planned attacks on communities, which could result in massacres of many people. Whatever the proximate reasons behind someone taking part in specific episodes of violence—perhaps seeking revenge for recent or long-remembered wrongs, acquiring prestige-enhancing trophies such as scalps, or making off with livestock or captives—the fighting furthered the objectives of antagonistic groups.

While territorial aspirations need not have been an acknowledged reason for an attack, intergroup conflict contributed to a geographical shuffling of populations. The movement of people, of course, has long been a concern of archaeologists. But only over the past few decades has there been enough spatial and temporal resolution to delineate varied demographic histories across large regions (Chaput et al., 2015; Hill et al., 2004; Milner et al., 2001; Shennan & Edinborough, 2007). Increases or decreases in occupation density could have resulted from intrinsic population growth or decline, or migration into or away from particular places. Warfare with its attendant death and destruction was not the sole reason behind such varied population histories. Nevertheless, intergroup conflict surely played a part in the expansion and contraction of the areas that specific groups of people routinely used.

Keeley (1996, also see Keeley, 2016) highlighted the tangible economic advantages and costs for those involved in fighting. People benefited by expanding their territories, hence the resources critical for survival. Those forced off their land—thereby denied access to familiar and much-needed sources of food, such as rich stands of wild plants, standing crops, and stored surpluses—soon found themselves in dire straits. For largely self-sufficient communities precariously perched on a knife-edge between success and failure, jockeying for advantageous positions was an ever-present reality.

The information on intergroup conflict available to Keeley in the early 1990s was largely drawn from ethnographic and historical sources. His overall argument, therefore, was subject to the tribal zone criticism (Haas & Piscitelli, 2013). When obtaining estimates of the frequency of conflict-related deaths, Keeley could rely on only a limited number of archaeological studies. They were the ones that provided quantitative information on conflict-related deaths estimated from reasonably large and well-documented skeletal samples (Keeley, 1996: fig. 6.2, tab. 6.2). When his book was written, the most widely known skeletal sample was from Jebel Sahaba, a terminal Pleistocene site in Sudan (Wendorf, 1968; also see Crevecoeur et al., 2021). He characterized these hunter-gatherers as being as “ruthlessly violent as any of their more recent counterparts” (Keeley, 1996:38).

Keeley (1996: fig. 6.2, tab. 6.2) also cited a community of subsistence agriculturalists from the American Midwest that dated to around ad 1300 (Milner et al., 1991; also see Milner & Ferrell, 2011). These people, classified archaeologically as Oneota, suffered greatly from multiple ambushes, each of which resulted in the deaths of only a few people. Yet cumulatively these casualties amounted to as much as one-third of all the men and women buried in the village's completely excavated cemetery. The ages of the oldest victims were consistent with the upper range of people who were said to bear arms in the late 18th century Midwest. The people ambushed were often disabled, which rendered them especially easy targets.

Although these skeletons provided a good picture of how fighting occurred, the Oneota village was not typical of all such contemporaneous midwestern communities. The site highlights a common archaeological problem. Because there are rarely tight controls over the kinds of sites that are sampled, generalizing from single places is full of uncertainty. The Oneota community cited by Keeley (1996:fig. 6.2, tab. 6.2) consisted of new arrivals in the central Illinois River valley, underscoring the link between intergroup conflict, devastating losses of life, and population movement. The particular circumstances these people found themselves in meant they suffered cruelly at the hands of their enemies.

Numerous skeletal examples of interpersonal violence, including both healed and lethal injuries, have been published since Keeley's (1996) book appeared. It is often difficult to tell if they were a result of intragroup violence, such as homicides or the abuse of captives or other low-status people, as opposed to attacks by members of different communities who were regarded as enemies upon first sight. Here the nature and patterning of injuries, signs of body mutilation, the number of victims, and their archaeological context help to identify examples of intergroup conflict. Defensive structures, usually walls around settlements, are easier to interpret as evidence of uneasy relations among communities. Walls, however, can be built for purposes other than to protect people and their property. Looking closely at construction details, such as whether bastions studded curtain walls, can help to determine whether they were primarily intended for defense or for some other purpose, as was done by Keeley et al. (2007).

Yet despite many new studies, it is still true that archaeological information about the intensity of warfare is scarce. For the most part, it can be best estimated through community-level victim frequencies, meaning many well-preserved skeletons accompanied by good contextual controls are needed.

It is difficult to identify evidence of conflict of whatever form for mobile Upper Paleolithic hunter-gatherer bands, and even further back in time (Haas & Piscitelli, 2013; Kim & Kissel, 2018). There are relatively few skeletons, and no settlements with readily identifiable defensive structures. With the available skeletons, it is hard to separate the outcomes of intra- and intergroup conflict, although at least some deaths likely resulted from the latter (Haas & Piscitelli, 2013; Kim & Kissel, 2018). It has long been recognized that the composition of small mobile bands of foragers periodically changes as people move from one group to another. Bands fission when people “vote with their feet” to avoid conflicts escalating into lethal fights and the resulting enmities that can last for years, if not generations, afterwards (Lee, 1979:367).

Intergroup violence seems to have become a regular feature of life once people began to live in permanent settlements for much, or all, of the year (Haas & Piscitelli, 2013; Kim & Kissel, 2018). Glossing over many differences in ways of life, these groups include quasi-sedentary hunter-gatherers and subsistence agriculturalists often referred to as Mesolithic and Neolithic, respectively.⁴ An apparent deterioration in relations later in time is, at least in part, related to archaeological visibility. Worldwide there are numerous skeletons from societies broadly classified as Mesolithic and Neolithic, and for the most part readily recognizable defensive works are known from Neolithic and later societies.

Coming to grips with the differential visibility of archaeological evidence and the part it plays in apparent temporal trends in intra- and intergroup conflict is hard. The most compelling way to estimate the intensity of warfare is through frequencies of skeletal trauma. That, in turn, means obtaining multiple contextually well-characterized and large samples from specific culture areas. Yet despite numerous flaws in existing data, it appears that intergroup relations often worsened with shifts to more settled ways of life (Estabrook, 2014; Nakao et al., 2016).

Skeletons, settlements, and artifacts yield information about general tendencies in how warfare involving prestate societies was conducted and perceived by those who took part in it, although details naturally vary from one part of the world to another (Bridges et al., 2000; Brown & Dye, 2007; Crevecoeur et al., 2021; Haak et al., 2008; Lambert, 1997, 2002, 2014; Meyer et al., 2015; Milner et al., 1991; Milner & Ferrell, 2011; Schroeder et al., 2019; Schulting, 2013; Smith, 2014; Steadman, 2008; Wahl & Trautmann, 2012; Willey & Emerson, 1993; Worne et al., 2012). Situationally vulnerable people were often ambushed, presumably when outnumbered by their enemies. These attacks usually resulted in just a few individuals being killed at a time, although cumulatively their deaths could threaten community viability. Ambushes were periodically punctuated by major attacks that devastated entire communities, resulting in tens, if not hundreds, of deaths. Anyone might be killed in the fighting, but victims were usually adults and predominately men. More equal sex ratios could occur in massacres where people of all ages and both sexes were indiscriminately killed, and the same was true when ambushes were frequently repeated, also resulting in many deaths. Prior injuries or diseases rendered people especially vulnerable. Regardless of how attacks were carried out, bodies might be mutilated and trophies, including scalps, heads, and limbs, taken. Specialized weaponry and defensive gear were sometimes used, although often the weapons were everyday tools, such as arrows and axes. Success at war was folded into the very ethos of societies. That is archaeologically visible, for example, as artwork celebrating prowess at war and ceremonial weapons associated with individuals who held highly esteemed social positions. In short, intergroup conflict's archaeological signature is what would be expected from ethnographic and historical descriptions of such fighting.

Bioarchaeological studies, in particular, have enhanced our understanding of warfare among prestate societies since the publication of Keeley's (1996) book. Wound distributions can indicate how fighting took place; for example, victims might have been facing their attackers or fleeing when cut down. Particularly important are studies based on large samples. They allow one to identify what took place and the intensity of fighting, as estimated from the number of casualties (Crevecoeur et al., 2021; Milner et al., 1991; Steadman, 2008) as well as compilations of conflict-related skeletal evidence, sometimes accompanied by settlement information, from large geographical areas (Arkush & Tung, 2013; Fibiger et al., 2023; Milner, 1999; Milner et al., 2013; Smith et al., 2020; Worne et al., 2012).

Nevertheless, much more remains to be learned about the nature and impact of warfare on local communities. That includes the individuals who were injured or killed, and their relationships to one another. Such topics have been explored at one late Neolithic site in southern Poland through an analysis of ancient DNA (aDNA) from 15 adults and children who were killed all at once and buried in a common grave (Schroeder et al., 2019). These people were closely related, and they included members of several nuclear families. This is an outstanding example of where an aDNA analysis of just a few individuals is sufficient to produce results of great interpretive value. Of critical importance is the tightly controlled archaeological context where people who were alive at the same time were interred together. The usual situation, in contrast, is skeletons that had accumulated in a cemetery over some lengthy period. Not only were the people who fill most cemeteries alive at different times, the composition of the groups that contributed to the skeletal sample could have changed over the duration of cemetery use.

Beyond the immediate deaths, conflict's cost to communities is poorly understood. Merciless violence undoubtedly caused great physical and emotional suffering for community members. The sudden and unpredictable loss of people in the prime of life must have affected the acquisition of food, among other resources, essential for household and community survival. Warfare's disrupting and demoralizing effects certainly led to hardship and excess mortality, although their magnitude is difficult to estimate.

One way to address this issue is to look at the disease load of hard-pressed groups. The Oneota villagers mentioned above, for example, were haunted by ill health (Milner et al., 1991). Their subsistence practices must have been adversely affected by numerous conflict-related deaths and, quite possibly, a curtailment of foraging tasks that put people at risk when venturing far from the relative safety of their village. Unfortunately, interpreting lesion frequencies in mortality samples, which is precisely what archaeological skeletons happen to be, is difficult. A more effective means of tackling this problem might involve looking at general trends in age-independent mortality to see if deaths distributed broadly across the age distribution were associated with high-conflict situations (Milner & Boldsen, 2023). That would be consistent with more pressure placed on communities. It would be noticeable skeletally as a greater representation of the adolescent to young adult fraction of the sample than is found in other cemeteries. No matter how it is done, detecting and measuring the downstream effects of warfare-related deaths on community life is far harder than identifying victims of attacks and estimating how many people died in such a fashion.

Even with deficiencies in existing evidence, it is difficult to avoid Keeley's (1996:175) conclusion that conflicts among the kin-based and usually small societies of the precontact era were “total war conducted with very limited means.” Warfare-related mortality—the immediate deaths and those that followed from suffering, malnutrition, and disease—was likely to have been one of the forces that collectively held global population growth in check for most of our species' existence. That should be justification enough to systematically study warfare in band- to chiefdom-level societies.

While there is considerable evidence for warfare involving archaeologically known quasi-sedentary hunter-gatherers and subsistence agriculturalists, it is not found everywhere. An oft-repeated caution about interpreting an absence of evidence as evidence of absence certainly applies here. Yet it seems that truly peaceful relations occurred among prestate societies, and that rather pacific periods could span many lifetimes and encompass large geographical regions. The existence of generally peaceful relations might be anticipated from historical and ethnographic accounts (Fry, 2012). But the admonition about relying on information from near-recent times to inform us about conditions in the distant past is as true for peace as it is for war.

Keeley (1996) naturally focused his attention on intergroup conflict—it was why he wrote his book. Furthermore, it is relatively easy to identify projectile points embedded in bones, ax wounds, body mutilation, defensive walls, and the like. This emphasis on what is archaeologically visible should not be interpreted as implying that peace was nothing more than an absence of conflict. Harmonious relations among separate communities with, at times, competing interests require effort to establish and maintain. Identifying archaeological evidence of such relations and how they came about is not easy. But it has met with some success, such as estimating when the League of the Iroquois originated (Snow, 1994).

To be sure, uneven archaeological coverages result in murky pictures of the past. But when looking at sufficiently large geographical regions and long periods of time, it is increasingly apparent that the distributions of traumatic injuries and defensive works vary greatly over space and time (Arkush & Tung, 2013; Bartelink et al., 2014; Fibiger et al., 2023; Lambert, 1997, 2002; Milner, 1999; Milner et al., 2013; Parkinson & Duffy, 2007; Smith et al., 2020). Keeley (1997; Golitko & Keeley, 2007) was an early contributor to this slowly accumulating body of literature, which is based on deep dives into the details of specific regions. A key aspect of this work involves plotting the spatial and temporal distribution of evidence for conflict.

Simply identifying that variability, difficult as it is, does not get us very far. The bigger purpose of such work has to do with the sociocultural, demographic, technological, and environmental conditions that lay behind such variability. Of particular interest are what provided opportunities and imposed constraints on societies that either led to, or ameliorated, competition over resources, and all else that followed in train. That included outright warfare that could take on a life of its own as cycles of retaliatory killings extended far beyond whatever initial grievances or circumstances provoked the fighting. Unsurprisingly, periods of greater violence accompanied deteriorations in the material conditions of life, specifically resource scarcity and unpredictability (Allen et al., 2016; Jones et al., 1999; Kennett et al., 2013; Kohler et al., 2014; Kuckelman, 2016; Lambert, 1997, 2002, 2012, 2014; LeBlanc, 1999). Resource availability, of course, was not the only important aspect of life associated with either more or less violence because it is embedded in larger sociopolitical, technological, and demographic contexts (Kennett et al., 2013; Kohler et al., 2014). Humans, for example, are notable for their collective memory, which can prolong intergroup antagonisms for generations after the original real or imagined wrongs took place.

Just what led many, but not all, prestate societies in the distant past to engage in warfare remains a vexing issue. That is partly related to the peculiar strengths and weaknesses of archaeological studies. Skeletons, artifacts, and sites are best suited for coarse-grained views of the general sociopolitical, technological, demographic, and environmental conditions under which intergroup conflicts were likely to break out. But archaeological coverages of time and space are incomplete, and what comes to light is susceptible to alternative interpretations, some demonstrably better than others. Analyses of archaeological remains generally cannot come to grips with a multitude of proximate causes that, in one way or another, led to conflicts among separate communities. Yet shedding such rich context-specific detail helps when trying to identify the general conditions that favored war or peace.

After several decades of research, it is apparent that there were protracted periods when warfare was rather common across large geographical regions. The same was also true for generally peaceful intergroup relations. Keeley (1996:183) is quite rightly associated with putting warfare squarely within archaeologists' sights, yet he found it “disappointing” that insufficient attention was directed toward investigating why certain times and places were dominated by peaceful interactions among different groups of people. The trick for us today will be to assemble the contextually rich information needed to discover the reasons behind such varied pictures of intergroup conflict.

At this point, only one thing about intergroup conflict involving prestate societies is crystal clear—Hobbesian and Rousseauian characterizations of life are best cast into history's dustbin. That alone is a notable accomplishment, considering the tenacious and centuries-long hold these ideas have had on scholarly and public thought. People in the distant past certainly engaged in brutal and destructive warfare resulting in many casualties and having long-lasting consequences. But they were also fully capable of taking effective steps to ensure peace. And that is a lesson for us in our own troubled world.

George R. Milner: Writing – original draft (lead).