AZTECHU1\IANSACRIFICE:CROSS-CULTURAL
HYPOTHESIS1
OF THE ECOLOGICAL
ASSESSMENTS
'm'
A
;
MichaelWinkelman
ArizonaStateUniversity
Ecological, religious, and social predictors of institutionalized human sacrifice are
assessed through cross-cultural analysis. While human sacrifice has no signifcant
correlations with measures of agricultural potential, protein, total food, food storage
adequacy, and famine risk, there are significant positive correlations with population
density, population pressure, and war for land and resources. Population pressure and
war for land and resources have independently signiflcant correlations with human
sacrifice, and together account for 38 per cent of its variance (multiple R = .62, p < .006).
A measure of low hierarchical focus of religion provides signifslcantadditional explanation
of variance (multiple R=.72, R2=.S1, pc.OOO), suggesting human sacrifice may play a
cannibalism, religion, ethnology, ecology)
role in ideological integration. (SacrifWlce,
Concernwith the explanationof humansacrificeandcannibalismgainedheightened
interestwith Harner's(1977a, 1977b)publicationson Aztec humansacrificeand
cannibalism.Harnersuggests their causes were relatedto protein scarcity as a
consequenceof high populationpressurein a distinctiveecologicalsituation.Ortiz
de Montellano(1978, 1990) rejects Harner'sargumentsbased on Aztec dietary
information.Price (1978), Hassig (1990), and Isaac (1983) suggest that human
sacrifice was an epiphenomenonreflectinggeopoliticaldynamics, political and
military instabilities, demographicconditions, and economic production and
distributlonnetworks.But these rejectionsof the ecological hypothesisdo not
convincinglysupportthe alternativesproposedas the causesof humansacrificeand
cannibalism.Althoughsome reportsof cannibalismare doubtful
institutionalized
(e.g., see Arens 1979), evidencefor both humansacrificeand cannibalismis well
(c.f.
andclinicalliterature2
historical,bioarchaeological,
documentedinethnographic,
TurnerandTurner1995). Cannibalismin some nonhumanprimategroups(Goodall
1977) suggeststhatit may play a role in ecologicalandsocial adaptations.
This articlereportscross-culturalanalyseson previouslypublisheddatasets of
Sample(SCCS)(MurdockandWhite1969)to assessthe
the StandardCross-Cultural
role of ecologicalfactors,religiousconditions,and social complexityvariablesin
predictinghumansacrifice.The focusis on legitimatehumansacrificecarriedoutby
religiousleadersas normativesocialactivities.Thewide rangeof measuresexamined
include:social complexityvariables;agriculturalpotential,meatprotein,domestic
animals, and total foods; food storage adequacy;threat of famine; population
(assessedthroughwarfarefor landand
circumscription
pressure;andenvironmental
resources).The relationshipof religiousandsocial complexityconditionsto human
sacrificeareassessedto illustratetherelevanceof othersocialfactorsto the incidence
investigations.
of humansacrifice,andto suggestdirectionsfor lFurther
285
ETHNOLOGYvol. 37 no. 3, Summer 1998, pp. 285-98.
ETHNOLOGY, c/o Deparanentof Anthropology,The University of Pittsburgh,PittsburghPA 15260 USA
Copyright 1998 The University of Pittsburgh.All rights reserved.
This content downloaded from 129.252.86.83 on Mon, 10 Mar 2014 16:32:37 PM
All use subject to JSTOR Terms and Conditions
286
ETHNOLOGY
HYPOTHESIS
THEECOLOGICAL
Harner (1977a, 1977b) rejects anthropologicaltheories that reflect Aztec
explanationsfor humansacriElce;i.e., thatsacriflcewas requiredby their religion
andgods. Basedon studiesof populationpressure,Harner(1970) suggeststhatthe
unparalleledscope of Aztec human sacrifice and cannibalismresulted from
factors which createdprotein shortagesand population
demographic-ecological
pressure:unfavorableagriculturalconditions,seasonalcrop failures, the lack of
domesticatedherbivores,the depletionof wild game in the region, food scarcity,
causedby limitationson theexpansionof
circumscription
famine,andenvironmental
by poorfarmingland,
agriculture.Thefertilecentralvalleyof Mexicois surrounded
expansion.While
precludingagricultural
circumscription
creatingan environmental
lack of necessary
their
proteins,
vegetable
complementary
provide
beans
and
corn
fattyacids andthe seasonalscarcityof thesefoods madethese sourcesinsufficient.
This leads Harnerto hypothesizethatcannibalismprovideda significantsourceof
protein.The Aztecsdidnot ordinarilyeatpeoplefromtheirownpolity,butpracticed
warfare,calledflowery wars,"as ritualsto obtainsacrificialvictims.Theconquered
territoriesnearbywere a convenientsourceof victims.Althoughthe consumptionof
humanflesh was reservedfor the elite, theclasssystemallowedfor upwardmobility
of greatwarriors,so warsto obtainvictimsfor sacrificeweresupportedby a hungry
populationwhichdesiredbothprestigeandproteinin theformof humanflesh. Great
warriorscouldalso receivetherightto consumehumanflesh, whichwas sharedwith
lineagemembers.
Harner'shypothesisof cannibalismbasedon chronic
Price (1978) characterizes
fromfamineandlack
distinguished
insuifficiently
proteinshortagesas unsubstantiated,
largelyconsumedby
were
sacriiSlces
human
that
fact
the
by
undermined
of food, and
Montellano(1978)
de
Ortiz
the elite classthatalreadyhadeasy accessto othermeat.
presents nutritionalresourcedata and dietary informationchallengingHarner's
hypothesis.By documentinga wide rangeof meatproteinsavailable,the enormous
food tribute they received, the stores of food maintained,and the intensive
agriculturaltechniquesemployed,he arguesfor an adequatediet for the Aztec
population.Otherfactorsmitigatingagainstthe ecologicalhypothesisinclude: 1)
of the total
conqueringnew landsfor agricultureand tribute;2) the insigniElcance
of humansacrifice
humanproteinavailablefromsacrifice;and3) the preponderance
duringthe annualperiods of food abundancefrom harvests,ratherthan during
of
periodsof proteinscarcity.Ortizde Montellanosuggeststhatthe preponderance
and
of
thanks
za
gesture
it
was
humansacrificeduringharvestperiodsindicatedthat
reciprocityto the godst (1978:614)ratherthanfilling a needfor protein.
RELIGIOUSAND SOCIALHYPOTHESES
theAztecpracticeof humansacrifice
Ortizde Montellano(1978, 1990)attributes
to theirbelief thatthe gods requiredit. This cosmologicalhypothesisis obviously
This content downloaded from 129.252.86.83 on Mon, 10 Mar 2014 16:32:37 PM
All use subject to JSTOR Terms and Conditions
AZTECHUMANSACRIFICE:
THE ECOLOGICAL
HYPOTHESIS 287
ttue but uninformative. Religious justifications for humarlsacriflce and cannibalism
do not alone explain such beliefs arld behavior. The individual Aztec warrior's
motivation for participation in warfare for sacrificial victims included the social
benefits accrued. But such obvious linkages do not inform us of the reasons human
sacrifice was adopted as a mechanism for allocating prestige and mobility. The
motivationsfor self-sacrificewere undoubtedlyreinforcedby religious beliefs thatthe
victims would go to heaven. But why such beliefs would be adopted remains
unaddressed. Their beliefs may explain their behavior, but leave unanswered the
question of why the beliefs and practices were adopted, and fail to identify the
psychosocial functions of human sacrifice.
A dominantperspective in social arlthropology,influenced by Emile Durkheim,
is that religious beliefs and practices derive from and are reflective of social
conditions (Bourguignon 1976; BourguignonandEvascu 1977; Davis 1971; Swanson
1960; Peregrine 1996; Winkelman 1986 1990, 1992)* If so, the determinantsof
human sacrifice should be sought in the social conditions which struct:urereligious
beliefs. The perspective that determinantsof human sacrifice are found in the social
conditions which structurereligious beliefs is illustratedin assessments of the broader
economic networks and political conditions associated with Aztec sacriElce and
warfare (Price 1978; Hassig 1990; Isaac 1983). These reject human sacrifice as a
motivatorfor Aztec flowery wars (presumptivelyfor sacrificial victims), and thereby
dismiss the importance of human sacrifice within the broader context of state
activities, warfare, economic systems, and class stratification.
Price (1978) characterizescannibalismas a stylistic trait and an epiphenomenon
which had little effect on economic intensification,warfare, political expaxlsion,and
social stratification.Price suggests that the explanationof Aztec human sacrifice and
cannibalism is to be found in economic networks of production and distribution,
patterns of state redistribution,political and military instabilities, and pacifications
of the nobility Her model postulates that human sacrifice reinforced existing
stratificationand political power, but she acknowledgesthe lack of evidence for such
effects in her work. Isaac (1983) hypothesizesthat humansacrifice served ideological
functions in Aztec society by uniting the class interests of nobles and upwardly
aspiring lower-class warriorssbut he fails to explain why human sacrifice would be
adopted for such integrativepurposes.
Hassig (1990) links Aztec humansacrifice to their precariouseconomic position
and consequent social and political adaptation.Their flowery wars are analyzed as
a strategyof empire-buildingwhich enabledgreaterexpansionby allowing dependent
areasto be self-administeredyet pay tribute, ratherthan investing higher political and
administrativecosts to directly administerthe conqueredareas. Hassig characterizes
the flowery wars as a deliberatestrategyto wear down strongerenemies, ratherthan
as an institutionalizedprocedurefor obtaining sacrificial captives. Price (1978) and
Isaac (1983) similarly suggest that the flowery wars reflected the shifting balances of
power in the geopolitical dynamics of the Aztec Triple Alliance and their inability to
conquer the Tlaxcala-PueblanValley. Price suggests that the inconclusive natllre of
This content downloaded from 129.252.86.83 on Mon, 10 Mar 2014 16:32:37 PM
All use subject to JSTOR Terms and Conditions
288
ETHNOLOGY
militaryoperationsmadehumansacrificean ideologicalmechanismto explainaway
the failure of militaryconquestunderthe guise of ritualpursuits.Moteuczoma's
characterization
of the wars as ritualsfor obtainingcaptivesandprovidingmilitary
trainingfor soldiersis seen as a strategieruseto directthe Spaniards'attentionaway
from Aztec politicalandmilitaryweaknesses(Isaac1983).
While there are many economic, social, and politicalexplanationsfor Aztec
humansacrifice,no convincingevidenceis offeredas to why they shouldhave had
a role in theseprocesses.Whyshouldthesesocialconditionsleadto humansacrifice?
Theflowerywarsdidserveas a population-control
mechanism-reducingpopulation
pressuresin the Valleyof Mexicoby cullingyoungmalesthroughmilitaryfatalities
(Prlce1978)-and undoubtedly
relievedpressureon the landtenuresystem.Butwhy
use humansacrificeas a justificationandmechanism?
If populationor otherecological,economic,class, or politicalfactorsareto be
demonstrated
as causesof religiouspractices(e.g. humansacrifice),they shouldbe
shown to have systematicrelationsin a cross-culturalsample. Argumentsabout
systemic causes of humansacrificeand cannibalismbased on case materialsare
unconvincingon methodologicalgrounds.The failureto establishrelationshipsof
humansacriflcewith specificecologicalor socialvariablesleavesexplanationof the
behavioropento characterization
as socialpathology.Butif humansacrificeis to be
explainedas an ecological,economicssocial, or politicalphenomenon,it mustbe
demonstrated
throughsystematiccross-cultural
analysis.
METHODS
This studyused a subsample3(Winkelman1986, I990, 1992; Winkelmanand
White1987)fromthe StandardCross-Cultural
Sample(SCCS)(MurdockandWhite
1969) and its datasets for a systematiccross-cultural
assessmentof ecologicaland
social predictorsof humansacrifice.The subsampleincludedsocietiesof the major
regionsof the worldandcovereda timespanfrom 1750B C. to the presentcentury.
Assessmentsof humansacriflcewere derivedfrom a studywhich utilizedformal
analysisof codedvariablesbasedon descriptivedataderivedfromethnographies
(see
Winkelman1992; Winkelmanand White 1987 for methodology). Culturally
recognizedmagico-religiouspractitionerstatuseshad been previouslyassessedon
social,political,religious,medical,cosmologicalandritualcharacteristics,
including
formsof sacrifice.The presentstudycompileddatafromthe individualmagicoreligiouspractitionerstatusesto determinethe societalincidenceof humansacrifice.
HumanSacrzfce Variable
The present researchassesses ritual humansacriE1ceratherthan sacrificial
cannibalism.As therearedifferentformsof humansacrifice(e.g* socialcontrolvs.
socialpathology)(TurnerandTurner1995),thepresentstudydistinguisheslegitimate
human sacriElce(a form of normativebehavior, involving propitiation)from
This content downloaded from 129.252.86.83 on Mon, 10 Mar 2014 16:32:37 PM
All use subject to JSTOR Terms and Conditions
AZTECHUMANSACRIFICE:
THE ECOLOGICAL
HYPOTHESIS 289
malevolenthumansacrifice(immoralpracticesattributedto sorcerersandwitches).
These malevolentbeliefs (e.g., consumingbody parts, or the soul or spirit of the
victim) apparentlyreflect attributionsratherthan actualbehaviorand are not the
concern of this study. Legitimatehuman sacrifice is typically associatedwith
magico-religious
practitioners
calledpriests.Priestsareelitesociopoliticalfunctionaries who also have centralroles in communalreligious activitiesassociatedwith
propitiationandcalendricalagricultural
rituals.
Seven societies of the sample had humansacrifice (Ovimbundu,Ibo, Kafa,
Roman,Marquesans,Atayal,andAztec). The Romanswere the only case in which
the practicesof humansacrificewere not carriedout by priests. In Rome, human
sacrificewas carriedout by a practitionercalled by terms translatedas witch,
sorcerer,and necromancer.In formalcross-culturalanalysis,this practitionerwas
classifiedas a shaman/healer,
not a witch/sorcerer(Winkelman1986, 1992). This
practitioner'suse of humansacrificeinvolvedeffortsto contactand influencethe
spiritworld, makingit a formof propitiation.AnalysesdroppingRomansfromthe
sampleproducedslightlyweakercorrelations,suggestingits appropriateinclusion
here.
Ecologicaland Social Variables
The ecologicalhypothesisof humansacrifice,whichpostulateshumansto be a
food resource utilized under adverse conditions, requiresassessing the factors
affectingfood availability(agricultural
potential,rainfall,naturaldisasters,risks of
famine, adequacyof storage, trade networksfor food etc.). These ecological
predictorswere derivedfrom previouslypublishedSCCS data sets on agriculture
suitability(landslope suitabilityof soil, climate,andannualrainfall)(Pryor1986);
subsistenceeconomyandsupportivepractices(includingintercommunity
tradeas a
food source,foodpreservation
andstorage,food surpluses,andsubsistencevariables
on contributionsof agriculture,hunting,fishing,gathering,anddomesticanimalsto
the food supply)(Murdockand Morrow 1970);politicalorganization(Tudenand
Marshall1972);politicalparticipation
andpeace,includingdecision-making
processes
andpoliticalfission (Ember,Russett and Ember1993);measuresof war for land
and resourcesand unpredictableresourceproblems,includingthreat of famine,
weatheror pest disasters,andchronicresourceproblems(EmberandEmber1992);
culturalcomplexity(MurdockandProvost1973);settlemetltpatternsandcommunity
organization(Murdockand Wilson 1972); and religious variables(Whyte 1978;
Murdock1967;WinkelmanandWhite 1987;Winkelman1992).
RESULTS
Datawere analyzedwithSYSTAT(1992) utilizingthe TablessCorrelation,and
Multiple Regressionprograms.Measuresof culturalcomplexity (Murdockand
Provost1973)wereusedto identifygeneralsocietalconditionsassociatedwithhuman
This content downloaded from 129.252.86.83 on Mon, 10 Mar 2014 16:32:37 PM
All use subject to JSTOR Terms and Conditions
290
ETHNOLOGY
sacrif1ce.These societiesare characterized
by a varietyof socialconditionsandare
typicallyat the mid-rangeof levels of culturalcomplexity.None of the original
culturalcomplexitymeasures4(five-pointordinalscales) is signiElcantly
correlated
with humansacrifice;the measuresdo not linearlydistinguishsocietieswith human
sacriE1ce
fromthosewithout.All sevencasesof humansacrif1cearefoundin societies
witha highrelianceuponagriculture
andwithsedentaryandrelativelypermanent
and
fixedresidencepatterns.All of thesocietiesin thesamplewithhumansacrificerelied
upon domesticatedanimals(bovines or smaller domesticatedanimals)for food
sources,but they were not pastoralsocietiesanddid not heavilyuse milk products.
Six of the cases were at the highestlevels of populationdensity.
AgricultaralPotential
Societieswith humansacrificehavea majorrelianceuponagricultureandin all
but one agriculturecontributedmore than any other food source. Gathering
contributedless than 10 per cent of the food supplyin all cases exceptone, but all
reliedupon huntingandgatheringto some extent.Morethanhalf of the cases had
good or betterlandslope, and all had fair to good land;five of seven cases were
ratedas havingexcellentclimate,and all had high levels of averagerainfall(600
mm, most over 1,000 mm). NeitherPryor's(1986) agricultural
potentialmeasures,
nor recodesfocusedon poor climate,land, or agricultural
potential,nor summary
measuresof agriculturalsuitabilityhadsignificantcorrelationswith humansacrifice
(Table 1). All of the societies with humansacrifice, however, had low or fair
agriculturepotential(as opposedto good or excellent),andall werein the mid-range
of measuresof agricultural
potential.
Table 1: Correlationof Agricultural
Suitabilitywith HumanSacrifice
IndependentVariables
Land Slope
Suitability of Soil
Climate and Rainfall
Total Suitability
Lowest Sliitability
Spearman'srho=
p
-.09
.11
.13
.04
-.07
(one-tailed)
ns
ns
ns
ns
ns
Food Supply and Famine Threat
Measuresof the availabilityof a rangeof food suppliesandresources(Murdock
andMorrow1970;EmberandEmber1992)were not significantlycorrelatedwith
humansacrifice(Table2). All of the societieswith humansacrificehad meansof
food storage,but flve of the seven experiencedseasonalor annualvariationin the
food supply. Threeof the seven importedfood throughintercommunity
trade,but
four had inadequatefood suppliesto lastthroughdifficulttimes. The majorityof
This content downloaded from 129.252.86.83 on Mon, 10 Mar 2014 16:32:37 PM
All use subject to JSTOR Terms and Conditions
AZTECHUMANSACRIFICE:
THE ECOLOGICAL
HYPOTHESIS 291
Table 2: Correlationof Pamine and Food Supply Variables with Human Sacrifice
Independent
Vanables
Spearman's
rho=
p< (one-tailed)
Huntingand Fishing
DomesticAnimalFood
Food Surplusvia Storage
StorageAdequacy
Season/Annual
Variation
FamineThreat
WeatherThreat
ChronicResourceProblems
-.22
.07
.03
.00
.10
-.02
-.25
-.15
ns
ns
ns
ns
ns
ns
ns
ns
ComputedVariables:
Sllmma ry Food Threat
Risk of Famine(CompositeMeasure)
TotalMeatProtein
TotalFood
Food Problems
-. 18
-.26
.06
-.07
-.10
ns
ns
ns
ns
ns
societies with humansacriflcehad low threatsof famine, food problemsdue to
weatheror pests, or chronicresourceproblems(one exceptionon each variable),
with only one case of high risk of famine(Aztec).The food shortagedata(Ember
andEmber1992) has limitedusefulnessbecauseof numerouscases of missingdata
or unreliablecodings for the societieswith humansacrifice. Over half of all the
ratingsfor these societies were unlcnownor lacked sufficientrater reliabilityto
report.Giventheuncertainty
aboutfoodreliabilityindicatedby the missingdata,this
informationwas used as a partof an indexon food uncertainty.Caseswith missing
data were placed on a scale (SummaryFood Threat)between no/low resource
problemsand moderateresourceproblems.Humansacrifice had no significant
relationshipto this recodenorto a compositefood resourcescale basedon the sum
of the recodesof resourcevariables(Table2). A similarscale was computedon
faminerisk, utilizingMurdockandMorrow's(1970) variablesfor food supplyand
storageadequacyin combination
withtheEmbers'variables.Thiscompositemeasure
of famine risk was nonsigniElcantly
associatedwith humansacrifice(Spearman's
rho=-.26), but in the oppositedirectionpredictedby the ecologicalhypothesis.
SammaryFood Ecology Measures
Themeasuresreportedabovewerereducedin categories,weighted,andsummed
intothreemajorscales assessingfood resources:totalmeatprotein(contributions
of
hunting,fishing, domesticanimals,largeherbivores);total food (meatproteinplus
food import, total agriculturalpotential,and food storage adequacy);and food
problems(threatof famine,weather,resourceproblems,faminerisk,poorland,poor
This content downloaded from 129.252.86.83 on Mon, 10 Mar 2014 16:32:37 PM
All use subject to JSTOR Terms and Conditions
292
ETHNOLOGY
climate, and poor agriculture potential). None of these summary scales had a
significant correlation with human sacriflce, thus providing little evidence in favor
of the ecological hypothesis. The nonsignificantnegative correlations with weather
threat, resource problems, and risk of famine suggest the opposite relationship
between food supply and human sacrifice of that predicted by the ecological
hypothesis.
Popalation Density and Population Pressure
Murdock and Provost (1973) provide a measureof populationdensity rankedat
five levels, from less than one person per squaremile (psm) through 100 psm. The
societies with human sacrifice all have higher levels of population density, with all
but one society with human sacrifice at 26+ psm. There was a significant Pearson
correlationof r= .40 (p < .01) with a binary recode (population < /= vs. > 26 psm).
Population pressure measures were derived by dividing population density by food
resource measures and the inverse of famine risk and food problem variables. These
assessments emphasize the pressure created by large populations with serious food
resource limitations or high famine risks. Human sacrifice was significantly and
positively associated with the population pressure measures based upon total food
supplies (r-.49, p<.OO1), adequacy of food storage (r=.S1, p<.OO1), and meat
protein (r= .29, p < .05), while the populationpressuremeasuresbased on a summary
of food problems and resourcesSand famine risk were positively but nonsignificantly
correlated with human sacrifice (Table 3).
Table3: Correlationof PopulationPressurewith HumanSacnfice
IndependentVanables
Population Density Recode
Population Pressure Based on Storage Adequacy
Population Pressure Based on Meat Protein
Population Pressure Based on Total Food
Popiation Pressure Based on Famine Risk
Population Pressure Based on Food Problems
Environmental
Pearson's r=
.40
.51
29
49
20
.22
p < (one-tailed)
.01
.000
05
.001
ns
ns
Circamscription
One cause of Aztec cannibalismpostulatedby Harner is environmental
circumscription.The lack of availabilityof good agricultllralland in nearbyareas
furtherexacerbatedfood shortages The measuresof Ember and Ember (1992)
provideproxy assessmentsof environmental
circumscription
in their measuresof
warfarefor the seizureof resourcesandthe use of landobtainedin warfare.These
originalvariableswere recodedto binaryvariablesreflectingthe absenee/presence
This content downloaded from 129.252.86.83 on Mon, 10 Mar 2014 16:32:37 PM
All use subject to JSTOR Terms and Conditions
AZTECHUMANSACRIFICE:
THE ECOLOGICAL
HYPOTHESIS 293
Table4: Correlationsof Environmental
Circumscription
with HumanSacnfice
Independent
Vanables
InternalWarfor Land
ExternalWarfor Land
OverallWarfor Land
InternalWarfor Resources
ExternalWarfor Resources
OverallWarfor Resources
OverallWarfor LandandResources
Spearmansrho-
p
.44
.20
.30
.37
.17
.30
.35
(one-tailed)
.007
ns
.04
.02
ns
.05
.02
of such conditions, and the individualsummarymeasuresfor overall land and
resourceseizurethroughwarwere summedinto an overallmeasureof war for land
and resources.The individualsummarymeasuresof internalwarfarefor land and
resourcesas well as combinedmeasures,werepositivelyandsignificantlyassociated
with humansacrifice(Table4). The measureof localpoliticalfissionwas also used
as an assessmentof (a lack of) environmental
circumscription.
Whiletherewas no
significantassociationwith local politicalfission, six of the seven societies with
humansacrificewerewithoutlocalpoliticalfission,consistentwiththe hypothesisof
environmental
circumscription.
Bothpopulation
pressure(basedon storageadequacy)
andthe overallmeasureof warfor landandresourceshaveindependently
significant
contributionsto the predictionof humansacrifice (populationpressurer=.51,
p<.001; overall war for land and resourcesr=.35, p<.02; multiple R=.62,
p < .006). Butlimitedexplainedvariance(38 percent)indicatesthatotherfactorsare
also responsiblefor humansacrifice.
Thehypothesisof generalsocialexplanations
of humansacrif1ceis not, however,
supportedby the present research.Inclusionin the regressionequationof ten
measuresof culturalcomplexity(MurdockandProvost1973;see note 4), including
variablesfoundpredictingmagico-religious
practicesin previousstudies(Winkelman
1986, 1992) (fixity of residence, agricultureSpolitical integration,and social
stratification),providedno significantincreasesin explainedvariancein human
sacrificebeyondthataccountedfor by the ecologicalvariables(populationpressure
andenvironmental
circumscription).
This rejectsthe notionthathumansacrificeis
a generaladaptation
to conditionsof increasingsocietalcomplexity.Butthe relatively
limitedexplanationof varianceby theecologicalmeasures(38 percent)indicatesthat
thereare still substantialdeterminants
to be identif1ed.
The presentstudy'sf1ndingsillustratethe relationof sacriElcein general,and
human sacrifice in particular to socioeconomicconditions:the absence of all
sacriiiceonly in huntingandgatheringsocieties;humansacrificefoundin complex
ratherthansimpleagriculxralsocietiessbutnot in pastoralsocieties;andthe lackof
a linearassociationof humansacrificewith measuresof complexity,with human
sacrificeoccurringin the mid-rangeof culturalcomplexity.Societieswith human
sacrificealwayshad priests,but neverhad shamans(as determinedin Winkelman
This content downloaded from 129.252.86.83 on Mon, 10 Mar 2014 16:32:37 PM
All use subject to JSTOR Terms and Conditions
294
ETHNOLOGY
1986 and 1992). HumansacriElceas a religiousactivityis relatedto otherreligious
andsocial conditionsO
Furtherassessmentof the relationshipbetweenhumansacrificeand religious
beliefs was based on Murdock's(1967) variableon high gods being presentand
important,and Whyte's(1978) dataon the relativestatusof males and femalesin
religiousbeliefs. Manyof the societieswith humansacrificehad religiousbeliefs
characterized
by the relativeequalityof maleandfemaleroles, andsix of sevenhad
high gods absentor unimportant.
The unimportance
of high gods in these complex
societies with humansacrificeis inconsistentwith generalfindingsof high gods
associatedwithsocialcomplexity(Swanson1960;Davis 1971;Peregrine1996).This
suggeststhatsincehumansacrificewas associatedwithsocietieswithouthierarchically integratedreligioussystems,humansacrificemaycreatereligiousintegration,as
suggestedby Isaac(1983)andPrice(1978). MurdockandWhyte'soriginalvariables
wererecodedintobinaryvariablesrepresenting
the lackof highgods andthe relative
equalityof male andfemalereligiousfigures(gods and spirits,mythicalfounders,
and witches). Some of these individualmeasureswere positivelyand significantly
correlatedwith humansacrifice(Table5). A summarymeasureof low religious
hierarchy(computedby summationof the binary recodes) had the strongest
correlation(Spearman'srho=.49, p<.001). Inclusionof this summaryreligious
measureof low religious hierarchyin multipleregressionalong with the two
ecologicalmeasures(population
pressureandenvironmental
circumscription
[overall
war for landandresources])accountedfor significantadditionalvariancein human
sacrifice(multipleR=.72, R2=.S1, F ratio=14.4, 3, 41 df, p<.000, 13 per cent
increase; 51 per cent total varianceexplainedwith inclusion of the religious
measure).The low religious hierarchymeasureenteredthe equationbefore the
environmental
circumscription
measure,suggestingits strongerpredictivepower.
Table 5: Correlationof Religious Variableswith Human Sacrifice
IndependentVariables
High Gods (present vs. unimportant/absent)
Gods and Spiritss Both Male and Female
Mythical Founders, Both Male and Female
Witches, Both Male and Female
Low Religious Cohesion (computed)
Spearman'srho=
.16
.42
.30
.32
.49
p< (one-tailed)
ns
.004
.05
.04
.001
AZTECHUMANSACRIFICEIN CROSS-CULTURAL
PERSPECTIVE
The cross-culturalanalyses support the general hypothesis of ecological
contributions
to theexplanation
of humansacrifice.Someof theecologicalconditions
Harnerhypothesizesas leadingto Aztec cannibalismare correlatedwith human
sacrificecross-culturally.In contrastto Harner'shypothesis,however,all societies
with humansacriElcehaddomesticated
food animals,butthe reductionof wild game
This content downloaded from 129.252.86.83 on Mon, 10 Mar 2014 16:32:37 PM
All use subject to JSTOR Terms and Conditions
AZTECHUMANSACRIFICE:
THE ECOLOGICAL
HYPOTHESIS 295
is suggested by these agriculturalsocieties with high populationdensity still
dependinguponhuntingandfishingin theoveralldiet.Measurements
of cropfailures
andfaminesarenotsignificantlyassociatedwithhumansacrifice,andthe nonsignificant correlationswith resourceproblemsand faminerisks are the oppositeof the
ecologicalprediction.The Aztecs,however,werehigh on manyof thesemeasures.
The signiElcantcorrelationsof humansacrificewith populationpressureand
environmental
circumscription
assessments(warfarefor larldandresources)support
the ecological hypothesis. Prices argumentthat Harner's hypothesis is not
sufficientlydistinguishedfrom famineor lack of food is addressedin this crossculturalanalysis.The lack of correlationof humansacrificewith famine, resource
adequacy,and food threatsindicatesthat the causes of humansacrificeare not a
directfunctionof food availabilityper se. Societieswith humansacrificeappearto
havethe oppositerelationshipto food supply,takingstepsto assurefood adequacy
throughfood storageand importation.The significantpredictionof humansacrifice
by specific ecologicalvariablesclarifiesthe generalecologicalconditionsHarner
hypothesizesas leadingto Aztec cannibalism.The ecological effects on human
sacrificeare not directlyfromthe availabilityof resources(food scarcity),but from
populationpressure.The significantlystrongerpredictionof humansacrificeby
populationpressure,ratherthanpopulationdensity,indicatesthatthe causesarenot
the sheer numberof people per se, but ratherthe stressfulconditionsthat density
createsunderresourcescarcityandunreliability.
But ecologicalfactorsalone are insufficient,as well over half of the variance
remainsto be explained.Inadditionto theseecologicalinfluences,measuresof a lack
of religioushierarchyare also significantlyassociatedwith the incidenceof human
sacrifice. Religiousbeliefs do reflect societalconditionswhich apparentlyhave a
significanteffect in motivatingthe adoptionof humansacrifice.The correlations
alonedo notrevealthepsychosocialfunctionsof the religiousactivities.Theysuggest
thathumansacrificeis a mechanismfor achievinga formof religiousintegrationin
societieslackingintegrativehierarchical
systemsof belief.Themotivationsfor human
sacrificeandfor cannibalismareillustratedby the broaderpsychocultural
dynamics
of sacrificewith respectto intergroupsocial relations.The psychosocialeffects of
cannibalismare suggestedby the Aztecs' dynamicsof in-groupand out-group
relationshipswith respectto consumptionfor sacrifice- whetherone is consumed
(victim)or co-consumer.Since the Aztecsdid not consumemembersof their own
group,unitingwith them in wars for sacrificialvictims and cannibalisticpractices
was a mechanismfor establishingin-groupmembership.The functionsof sacriflce
in creatingan integrationof peopleare illustratedby its servingas a mechanismfor
permittingclass mobilityof the warriorandhis relatives.This extensionof the right
to consumehumanflesh to family memberssignals a very basic level of social
*
fi
nc uslveness.
These systematiccorrelationsof humansacrificewith ecologicaland religious
conditionsillustratethatsuchpracticesshouldbe understoodin the contextof social
determinantsof behavior,ratherthanas forms of social pathologyand aberrance.
This content downloaded from 129.252.86.83 on Mon, 10 Mar 2014 16:32:37 PM
All use subject to JSTOR Terms and Conditions
296
ETHNOLOGY
in theAmericanSouthwestled
TurnerandTurner's(1995)assessmentof cannibalism
them to reject explanationsbased on social pathology.They (Turnerand Turner
1995:13) point insteadto violent social control, possibly initiatedby socially
of humansacrificewithpopulationpressure
t Thecorrelation
pathologicalindividuals.
indicates that populationdynamics are importantcontributoryfactors to the
developmentof thesepractices.Thesesocialdynamicsproducinghumansacrificeare
not, however,a generalfunctionof increasingsocialcomplexity.
CONCLUSIONS
Harner'soriginalhypothesisof ecologicalcausesof Aztechumansacrificeand
cannibalismreceivespartialbut significantsupportfrom the presentcross-cultural
of humansacrificethatremain
analysis.Thereare, however,additionaldeterminants
social, andpoliticalfactors
economic
of
systemic
to be identified.The hypotheses
involvedin the complexof Aztecwarfare,humansacriElce,andcannibalismsuggest
directionsfor futureworkto identifyadditionalfactorsthatpredisposesocietiesto
humansacrifice.While discussionsof Aztec humansacrificeandcannibalismhave
typicallyemphasizedthese as uniquefeaturesof Aztec society, the cross-cultural
analysespresentedhere suggestthatthe Aztecs typify patternsof humansacrifice
found cross-culturally.But the Aztecs might have been particularlymotivatedto
consumehumanflesh as a consequeneeof ecologicaland populationpressures.In
eomparisonto other societies with humansacrifice,the Aztecs were extremein
severalmeasures:the only humansacrificesocietyin this samplewith a highrisk of
famine; the highest on several measuresof populationpressure;in the highest
categoryof populationdensity(over 500 personsper squaremile); andthe highest
levels of overallwarfarefor land and resources.So while Aztec humansacrifice
conformsto typical cross-culturalpatterns,the magnitudeof their sacrifice and
cannibalismmay reflecttheirextremeconditionson manyecologicalvariables.
NOTES
1. I tharlkDavid Jacobs for his encouragementto pursue this research project. I also thank him and
Doug White, Michael Harner, ChristyTurner, Carol Ember, and Rob Le Veille for helpful suggestions,
and Cindy Winkelman for her assistance with data analysis.
2. Acosta Saignes 1950; Benson and Boone 1984; Bergmann 1992; Boal 1982; Brown 1991; Gohain
1977; Gonzalez Torres 1985; Green 1975; Hamerton-Kelly1987; Hogg 1966; Hughes 1991; Levenson
1993; Lewis 1986; Loeb 1964; Macoby 1982; Roman Berrelleza and Alberto 1990; Sagan 1974;
Sugiyama 1995; Taahill 1975; Tierney 1989.
3. A 25 per cent stratiEledrandom45-society subsample of the SCCS was used in the present study.
Societies in the subsampleinclude: Africa: Nama Hottentot,SKungBushmen, Ovimbundu,Ibo; CircumMediterranean:Wolof, Fulani, Pur, KafaflAmhara, Tuareg, Babylonians, Romans, Kurd; Eurasia:
Samoyed, Toda, Kazak, Garo, Vietnamese, Semang, Tanala,Japanese,Chukohee;InsularPaciElc:Iban,
Alor, Kirnam, Lesu, Pentecost, Marquesans, Trukese, Atayal; North America: Montagnias, Kaska,
Twana, Paiute, Hidatsa, Creek, Zuni, Aztec; South America: Bribri, Callinago, Saramacca Jivaro,
Tupinamba, Cayua, Mapuche. Societies with legitimate human sacrifice are: Ovimbundu, Ibo, Kafa,
Romaxl,Marquesans, Atayal, and Aztec.
This content downloaded from 129.252.86.83 on Mon, 10 Mar 2014 16:32:37 PM
All use subject to JSTOR Terms and Conditions
AZTECHUMANSACRIFICE:
THE ECOLOGICAL
HYPOTHESIS 297
4. Variablesincludedare: writing and records, fixity of residence, agriculture,urbanization,
technologicalspecialization,landtransport,
money,populationdensity,politicalintegration,andsocial
stratification
.
BIBLIOGRAPHY
AcostaSaignes,M. 1950. Tlacaxipeualiztli;
un complejomesoamericano
entrelos caribes.Caracas.
Arens,W. 1979. The Man-Eating
Myth:Anthropology
andAnthropophagy.
New York.
Benson,E., andE. H. Boone(eds.). 1984. RitualHumanSacrificein Mesoamerica.WashingtonDC.
Bergmann,M. S. 1992. Inthe Shadowof Moloch:TheSacrificeof ChildrenandIts Impacton Western
Religions.New York.
Boal, B. M. 1982. The Konds:HumanSacrificeand ReligiousChange.Wilts.
Bourguignon,E. 1976. Possession.San Francisco.
Bourguignon,
E., andT. Evascu.1977. AlteredStatesof Consciousness
withina GeneralEvolutionary
Perspective:A Holocultural
Analysis.BehaviorScienceResearch12(3):197-216.
Brown,S. S. 1991. Late Carthaginian
ChildSacrificeand SacrificialMonumentsin TheirMediterraneanContext.Sheffield.
Davis, W. 1971. Societal Complexityand the Nature of PrimitiveMan's Conceptionof the
Supernatural.
Ph.D. dissertation,Universityof NorthCarolina.ChapelHill.
Ember,C., andM. Ember.1992.Warfare,Aggression,andResourceProblems:Cross-Cultural
Codes.
BehaviorScienceResearch26(1-4):169-226.
Ember,C., B. Russett,andM. Ember.1993. PoliticalParticipation
andPeace:Cross-Cultural
Codes.
Cross-Cultural
Research27(1&2):97-145.
Gohain,B. C. 1977. HumanSacnficeandHead-Hunting
in North-Eastern
lndia. Gauhati.
GonzalezTorres,Y. 1985. E1sacriElcio
humanoentrelos mexicas.Mexico.
Goodall,J. 1977. InfantKillingand Cannibalism
in Free-LivingChimpanzeesof the GombeStream
Reserve.AnimalBehaviorMonographs1:161-311.
Green,A. R. W. 1975. The Role of HumanSacrificein the AncientNearEast. Missoula.
Hamerton-Kelly,
R. (ed.) 1987. ViolentOrigins:WalterBurkert,ReneGirardandJonathanZ. Smith
on RitualKillingand CulturalFormation.Stanford.
Harner,M. 1970. PopulationPressureandthe SocialEvolutionof Agriculturalists.
SouthwestJournal
of Anthropology26(1):67-86.
1977a. The EcologicalBasisfor AztecSacrifice.AmericanEthnologist4:117-35.
1977b.The Enigmaof Aztec Sacrifice.NaturalHistory86(4):47-51.
Hassig, R. 1990. AztecWarfare.HistoryToday40:17-24.
Hogg, G. 1966. Cannibalism
andHumanSacrifice.New York.
Hughes,D. D. 1991. HumanSacrificein AncientGreece.New York.
Isaac, B. 1983. The Aztec "PloweryWar":A GeopoliticalExplanation.Journalof Anthropological
Research39:415-32.
Levenson,J. 1993. The Deathand Resurrectionof the BelovedSon: The Transformation
of Child
Sacnfilcein Judaismand Christianity.
New Haven.
Lewis, I. M. 1986. Religionin Context:Cultsand Charisma.New York.
Loeb, E. M. 1964. The BloodSacrificeComplex.New York.
Maccoby,H. 1982. The SacredExecutioner:HumanSacrificeandthe Legacyof Guilt.New York.
Murdock,G. P. 1967. Ethnographic
Atlas.Pittsburgh.
Murdock,G. P., andD. Morrow.1970.SubsistenceEconomyandSupportivePractices:Cross-Cultural
Codes 1. Ethnology9:302-30.
Murdock,G. P., and C. Provost.1973. Measurement
of CulturalComplexity.Ethnology12:379-92.
Murdock,G. P., and D. White. 1969. StandardCross-Cultural
Sample.Ethnology8:329-69.
Murdock,G. P., andS. Wilson.1972. SettlementPatternsandCommunityOrganization:
Cross-Cultural Codes 3. Ethnology11:254-95.
This content downloaded from 129.252.86.83 on Mon, 10 Mar 2014 16:32:37 PM
All use subject to JSTOR Terms and Conditions
298
ETHNOLOGY
Ortiz de Montellano, B. 1978. Aztec Cannibalism:An Ecological Necessity? Science 200:611-17.
1990. Aztec Medicine, Health, and Nutrition. New Brunswick.
Peregrine, P. 1996. The Birth of the Gods Revisited: A PartialReplication of Guy Swanson's (1960)
Cross-CulturalStudy of Religion. Cross-CulturalResearch 30(1):84-112.
Price, B. 1978. Demystification, Enriddlement,and Aztec Cannibalism:A Materialist Rejoinder to
Harner. American Ethrlologist5(1):98-115.
Pryor, P. 1986. The Adoption of Agriculture: Some Theoretical and Empirical Evidence. American
Anthropologist 88:879-97.
Roman Berrelleza, R., and J. Alberto. 1990. Sacrificio de ninos en el Templo Mayor. Mexico DP.
Sagan, E. 1974. Camibalism: Human Aggression and CulturalForm. New York.
Sugiyama, S. 1995. Mass HulmanSacrifice and Symbolism of the Feathered Serpent Pyramid in
Teotihuacan, Mexico. Tempe.
Swanson, G. 1960. Birth of the Gods. Ann Arbor.
SYSTAT for Windows. Version 5 Edition. 1992. Evanston, IL.
Tamahill, R. 1975. Flesh and Blood: A History of the Cannibal Complex. New York.
Tierney, P. 1989. The Highest Altar: The Story of Human Sacrifice. New York.
Tuden, A., and C. Marshall. 1972. Political Organization:Cross-CulturalCodes. Ethnology 11:436-64.
Turner, C., and J. Turner. 1995. Camibalism in the Prehistoric American Southwest: Occurrence,
Taphonomy, Explanationand Suggestions for StandardizedWorld Definitions. Anthropological Science 103(1):1-22.
Winkelman, M. 1986. Magico-Religious PractitionerTypes and Socioeconomic Conditions. Behavioral
Science Research 20(1-4):17-46.
1990. Shaman and Other "Magico-Religious" Healers: A Cross-CulturalStudy of Their
Origins, Nature, and Social Transformations.Ethos 18:308-52.
1992. Shamans, Priests and Witches. A Cross-Cultural Study of Magico-Religious
Practitioners. Tempe.
Winkelman, M., and D. White. 1987. A Cross-CulturalStudy of Magico-Religious Practitionersand
ASC: Data Base. HRAP Research Series in QuantitativeCross-culturalData, Vol. 3, eds. D.
Levinson and R. Wagner. New York.
Whyte, M. K. 1978. Cross-CulturalCodes Dealing with the Relative Status of Women. Ethnology
17:21 1-37.
This content downloaded from 129.252.86.83 on Mon, 10 Mar 2014 16:32:37 PM
All use subject to JSTOR Terms and Conditions