JOURNAL OF ANTHROPOLOGICAL RESEARCH (Formerly SouthwesternJournal of Anthropology) VOLUME50 * NUMBER3 * FALL * 1994

FORAGINGRETURNS OF !KUNGADULTS AND CHILDREN:WHY DIDN'T !KUNG CHILDRENFORAGE?

NicholasBlurton Jones Departmentsof Education,Anthropology, and Psychiatry, University of California, LosAngeles, CA 90024

KristenHawkes Departmentof Anthropology,University of Utah,Salt Lake City, UT 84112

PatriciaDraper Departmentof HumanDevelopment, Pennsylvania State University, UniversityPark, PA 16802

Childrenof thehunting and gathering !Kung San seldomforaged, especially during the longdry season. In contrast,children of Hadza foragers in Tanzaniaoften forage, in both wet and dryseasons. Because we haveargued that the economicdependence of !Kung childrenhas importantconsequences, we musttry to understandwhy they did notforage. Experimentaldata on foragingby !Kungadults and childrenshow that children would havehad to walkfar from dryseason camps to acquiremuch food. Interviewssuggest that !Kungchildren risk getting lost if theywander unsupervised into the bush. Thus, foragingwithout adult companywas a poor optionfor !Kungchildren. Foraging with adultsmight have been a betterstrategy. We calculate the benefitsto a !Kungmother if heroldest child accompanied her to thenut groves.Because of thehigh processing costs, a child'swork time was mostprofitably spent at homecracking nuts.

MANYSCHOOLS OF anthropology attach importance to the subsistence infra- structureof a population.While the subsistenceecology of the !Kungof north- western Botswanastood for two decadesas one of the best described(Lee 1968, 1969, 1979) andas a landmarkin quantitativeanthropology, it may still be possible,even thoughthe circumstancesof !Kunglives havechanged greatly,

(Journalof Anthropological Research, vol. 50, 1994)

217

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions 218 JOURNALOF ANTHROPOLOGICALRESEARCH to supplementthis earlierdescriptive material. We present some figureson the rates at which!Kung women acquired food in 1988 in habitatsand communitiesdescribed by Lee in 1965 (Lee 1979). The datawere gathered with a specificquestion in mind.We hadargued that for the !Kung,the work involvedin feedingchildren constrained women's lives and indirectlylimited theirfertility (Blurton Jones and Sibly 1978; see alsoHarpending 1994; Blurton Jones 1994). But we had seen childrenforage successfully among the Hadza hunter-gatherersof the woodedsavanna of northernTanzania (Blurton Jones, Hawkes,and O'Connell1989). So it seemed essentialto findout why !Kung childrenwere not persuadedor permittedto forage and thus relieve their mothersof at least a partof theirburden. Somegeneral patterns in the cross-culturalvariation in childwork have been described,leading to the widelyheld view that while childrenin agricultural societies do muchuseful work, childrenin huntingand gatheringsocieties do little or none (Barry,Child, and 1959). Thus we tend to take it for grantedthat children of huntersand gatherers do no work,gratefully receiving the sustenance,care, and attentionoften lavishedon them by adults (see, e.g., Briggs1970; Konner 1976). But this orthodoxywas strikinglychallenged by the Hadza.This exception,even if only a single case, remindsus that explanationby categorizationcan never be satisfactory.What is it abouthunter- gathererlife that leaves childrenin some populationsdoing no work?What is it aboutother populations that results in children foraging? Why do somehunter- gathererchildren forage more andothers less? In fieldworkon the Hadza,we have approachedthe questionof children's workand its relationshipto fertility,child care, andthe relationsbetween the sexes from the perspectiveof behavioralecology. This perspective(though far removedfrom Western culture's commonsense concepts of causes of be- havior)has been highlyproductive in researchon hunter-gathererand other "traditional"societies (Smithand Winterhalder1992). From this perspective we expect parentsto allocateresources between care of offspringand the generationof more offspringso as to maximizethe numberof the parents' descendantsin subsequentgenerations (see, e.g., Lessels 1991; Smithand Fretwell1974). Thus we sharewith earlier perspectives the expectationthat if childrenwork, they are easier to feed and keep alive, and parentswill divert more of the availableresources away from child care andtoward bearing more children.Furthermore, we expectthat this potential advantage to the reproductive success of parentswould mean that wheneverchildren could feed themselves safely,they would be encouragedto do so. Thuswhen they do not workand are activelydiscouraged from trying, we must expect there to be some practical reason.Seen fromthe view of the child,in a contextwhere even the leisurely !Kungare described as showing"some mild undernutrition" (Truswell and Hansen 1976:171) and where the fieldworkernever sees a childrefuse an offer of food, it would not be surprisingif childrenhelped themselves to any food that was availablewithout too much effort or danger. On these groundsagain, we should as well ask, Why don't they forage?as Why do they forage?

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions FORAGINGRETURNS OF !KUNGADULTS AND CHILDREN 219 Manyauthors have claimedthat where childrencan do usefulwork, fertility is higher,and where they cannot,it is lower (e.g., Caldwell1982). Blurton Joneset al. (1992)and Blurton Jones, Hawkes,and Draper (1994) claimed that several reporteddifferences in the reproductionand behaviorof !Kungand Hadzaadults can be predictedfrom the differencein apparentcosts of diverting resourcesfrom feeding children toward bearing more children. If we are claim- ing to explainso muchfrom children's foraging opportunities, we are obliged to show why Hadzachildren forage so frequentlyand why !Kungchildren forageso seldom.Until we do this, our particularbehavioral ecological edifice for explainingdifferences between the Hadzaand the !Kungis builton sand. In this paperwe reportdata gathered in mid dry season 1988 to examine the benefitsthat might arise from !Kung children foraging, first, unaccompanied by adults,and, second, in the companyof adults.We aimed at a simpledisproof of the hypothesisthat the differencebetween the time!Kung and Hadza children spendforaging can be explainedby the costs andbenefits of foragingin their respectivehabitats. If we foundthat !Kungchildren could have obtainedmuch foodnear camp, suffered minimal risks fromthe quest for food, or couldhave increasedtheir food intake by accompanyingmother on foragingtrips, yet still did not forage, our hypothesiswould have falleninto immediatetrouble. We will show that it survivesthese initialtests. Lee (1979:71)proposes a completelydifferent view of !Kungchildren's foraging,which we discussin some detailin BlurtonJones,Hawkes, and Draper (1994). He suggests that childrenforage when the producer:dependentratio is low and that most of the time, in most camps, childrenrepresent an "un- utilizedreserve of laborpower."

BACKGROUND:THE !KUNGAND THE HADZA

The !Kung,living in northwesternBotswana, have been the subjects of manycultural, archaeological, biological, and demographicstudies (e.g., Lee 1979; Shostak1981; Yellen1977; Howell1979). The Hadzaof northernTan- zaniahave been describedby Obst (1912); Woodburn(e.g., 1968a, 1968b, 1988); O'Connell,Hawkes, and BlurtonJones (1988, 1991, 1992); Hawkes, O'Connell,and BlurtonJones (1989, 1991);BlurtonJones, Hawkes, and O'Con- nell (1989);and Blurton Jones et al. (1992). The !Kungand Hadza live 2,500 km apartand are culturallydistinct. Claimsof links between Hadzaneand Khoisanlanguages by Bleek (1931) andothers are not uniformlyaccepted by modernlinguists and are the subjectof ongoinginvestigation (Sands, Maddie- son, andLadefoged 1993); Hadzane remains one of the few languageswhose affinitiesare unknown.Nevertheless, there are manysimilarities between the two populations.People in both populationslive, or until recently lived, in small,mobile bands composed of aboutfifty people, most relatedby kinship or marriage.Neither group kept domesticstock or raisedcrops; men hunted with bow andpoison arrows, and women gathered, with occasional help from men. Manyof the same plantand animal genera or species were exploitedby

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions 220 JOURNALOF ANTHROPOLOGICALRESEARCH eachpopulation. Both groups lived in dry savannah and made frequent moves, eitherin searchof foodand water or to resolvedisputes. Both groups live at a verylow population density (!Kung, 0.042-0.08/km2; Hadza, 0.24-0.3/km2 [BlurtonJones et al. 1992])with usually a substantialdistance between neigh- boringcamps. In both,infants are carriedby mothermost of the dayand suckledfrequently. Diarrhea is a commoncause of deathof youngchildren in bothsocieties, and many adults and children contract malaria. Children interact withchildren of a widerange of ages, havingrather few exactage mates in a camp.In both cases, the samespecies of predatorsare quite abundant in the bush,making foraging by childrena potentiallyrisky enterprise. The Hadza and!Kung (while they were living as mobileforagers) each depended on stand- ing surfacewater in the rainyseason, and both faced reduced numbers of wateringplaces in the dryseason. Whilesimilar in the aboverespects, important differences also exist, and we arguethat ecological differences account for differences in the returnson children'slabor. Conditions are generallyless harshfor the Hadza.Water sourcesare morenumerous and near to foodsources, the dailymaximum temperaturesare lower, resources such as baobabtrees are much more abun- dantin Hadzacountry (up to 32 trees/kmin2),butthere are no mongongonut trees.The completed fertility for the two populations is different: Hadza women average6.2 births(Blurton Jones et al. 1992),whereas postreproductive !Kung womenreported 4.7 births(Howell 1979). The average stature of !Kungmen andwomen is 161cm and 150 cm, respectively, and their respective average weightis 47.9kg and40.08 kg. Thefigures for Hadza are: men 161 cm, 54.3 kg;women 150 cm, 48 kg (Hiernauxand Hartono 1980). The greater robust- ness of the Hadza(especially the women,as discussedby Hiernauxand Har- tono)seems to indicatemore favorable conditions. Sincethe 1970s,!Kung in the Dobearea have linked their fortunes more andmore closely to Hereroand Tswana herd owners who have settled the area.They rely less onbush foods, sometimes grow small gardens, and some- timesreceive government food aid. Furthermore, the numberof livestockin thearea has increased steadily, and some are owned by !Kung. The degradation of the environmentaround contemporary settlements is striking,especially to thosewho had visited the areain the latesixties or earlyseventies. Solway andLee (1990) suggest that the degradation extends far afield, which questions the relevanceof researchin the 1988environment to understandingforaging in the 1950sand 1960s. This question took high priority in ourdata collection andis examinedat lengthin the Appendix. Sincethe sixties,the Hadzahave been subjected to a seriesof attemptsto makethem settle andforsake the foraginglifestyle. These episodeswere mostlyshort lived, and at no timehave all Hadzaleft the bush.But on the otherhand, there are probablyno Hadzawithout experience of settlementlife. ContemporaryHadza seem to switcheasily and instantly between settlement and bush but requireinducements such as plentifulfree food to make them move to a settlement.

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions FORAGINGRETURNS OF !KUNGADULTS AND CHILDREN 221 Foragingby Children Sincethe comparisonof Hadzaand !Kungchildren's foraging motivated this fieldwork,we summarizethe evidencefor a differencein Table 1. The best

TABLE 1 Months with Data on Foraging by !Kung and Hadza Children

J F M A M J J A S O N D

f ...... :i::ii:ii~i::i!:.!:::!:.. /: !...... :::::::::::::::::::::::::: ...... ,...... , -.....,...... t B..~ ...~ ...... : 3 1 3 3 1 ?...... , ..?.?.l~i ? j ? ...... i !. ii i i iK~ ~~ii ...... ii~ . . ~~l . .i . i! . . !i. .! .ii . . !i!i: i~:....:~ . ... . i::~...... i:. .. ::i ...... H :..L.J...... :.....".Hadza t~ •i•'..:.:...... ,,."."'"''"".".'""."." 5 5 5 az .... f".....-fQ...... i~ i...... •...... i.'...... i...... :i....:::~ ..8.....ii ....::.ii...... =...... : . . .. . :. :.. : ...... :. : :. : :. : ...... : .: . . . :.:.. . : : :: : : :: : : .:::: .::.:.:::.. ::. . : .: . : :: : : : : :: : :: : : :: : : : :: : : : : ...... :::: .. .:.. . ::: .:.:...... :. .:..: .:.:.:.:= ...... : :. . . .:. ..:...... : . ..: :...... :.:: : : : : :: : ..7. ..i]]i : ] ] ]i ] ::::.:: :: ::: ..:::: ::: :: ...... ~iiii~:: ::: ::. ~ii iiiiiiiiiii!E ..::::::: :: ::!!iiiii~ : :: ::: ::!!!ii :. ii~~:!:i~ ::;i-:~: ~ii!:~:i: ~iii~~ : .. ..:-::.:-.-:.: . :.:... . .:..:.::.: .:. .:.::-:-.:.: :.:l :.:. ::: .:.::::: :::: :.::::: :-:,::. •.:l :.:. .:- ..- ..- . ....:: .E :'. .': ..: . .:'. ..: •.:•••:•••: ••• •• :•:.• ;•::•:• ••:•:•:•:• •:.: ••: :• ;•• :-::-:.:••• -:. •:• -;.: •:•: :. ;. •;••••: ::..::. . .: . ::: ::.. . . .::::. . :::: ::::. : ...... :...... ::::.:...... :.. :.:.. .. : : : :: : : : : : ...... :. .: . . .. :.:....:...... ::. . .: :: : ::::.: :. :: :: :: : :: :: :::: : : : : -.:.:..: :

Note: Numbersin bodyof tablerefer to notes 2-8 below.Rainy months are shaded. 1. !Kungrainfall data from Yellen and Lee 1976 (Figure1.1); annualrainfall 239 mm, 597 mm, 378 mm (mean405 mm);some fallsin Apriland May. Hadza rain seasonality from Woodburn 1964 and personalobservation. Rainfall is heaviestin December-Januaryand May. Annual amountsare not well documented. 2. Draper1976 (Table9.1). Childout of campin 8 of 76 spot observationsof girls (7 gathering withwomen; 6 of these under3 years old). Childout of campin 18 of 93 spot observations of boys(6 wereboys over 11 outwith men; 7 werewith women, mostly still carried by mother). 3. Draperand Cashdan1988. Girlsspent 0.6 minutesper 10-minuteobservation gathering or collectingwater or processingfood; boys spent 0.2 minutes. 4. Lee 1968, 1979 (Table9.3). Childrenforaged on 4 out of 244 child-days. 5. From 1985-86 Hawkesand O'Connellspot observationsand follows of adults.During June, nuclearfamilies pursue honey almost daily. Hawkes, O'Connell, and BlurtonJones (1989) report childrenand women collecting berries during October, March, and April. 6. August-October1986 (BlurtonJones, Hawkes,and O'Connell1989). Childrenforaged or processedfood in 50%of the one-hourfollows. Children aged 5-10 broughthome 729 kcal/ day;aged 11-15 broughthome 1,526 kcal/day. Baobab fruit and makalita roots. Berriesripened in mid-November. 7. September-November1988 (Hawkes,O'Connell, and BlurtonJones n.d.). On 47%of days childrenaged 6 andover accompaniedadults on day-longtrips to forageon Salvadorapersica andCordia gharaf berries at distancesup to 5 kmfrom camp (31% of daysin September,52% in October,42% in November).Berries were collectedat a rate of 1,045 kcal/hr.in patchfor S. persica,2,089 kcal/hr.for C. gharaf. 8. January-March1989 (fieldwork by BlurtonJones). Children foraged or processedfood in 45% of one-hourfollows. Children aged 5-10 broughthome 660 kcal/day;aged 10-16 broughthome 1,673 kcal/day.Makalita and panjuko roots, baobabfruit, nestling weaver birds. A few Cordia berrieswere ripe.

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions 222 JOURNALOF ANTHROPOLOGICALRESEARCH evidencethat !Kungchildren seldom foraged is givenby Draper'sten months of quantitativeobservations on childrenaged 2-14 (Draper1976; Draperand Cashdan1988). We must note that her observationson forager!Kung were fromthe /Du/daarea, where there are fewer mongongonuts (elsewherethe stapleplant food), but her conclusionis supportedby numerouscomments in the ethnographies.The 1988 interviewswith !Kungadults (Blurton Jones, Hawkes,and Draper 1994) confirmed this and other points with little prompting: !Kungchildren only foragedwhen food was so close that they couldbe kept in sight; usuallyfood is far away fromcamps, whichare near water. Adults said they scaredchildren into stayingin campand awayfrom going in search of foodbecause the bushwas dangerous.They saidthat takingchildren along on gatheringtrips "spoiledthe work,"explaining that childrengot tired and thirstyand held the adultsback. !Kungchildren were more likelyto foragein the rainyseason, whentemporary water could be foundduring a day'souting. Further,during the rainyseason, entire camps relocated away from permanent water;children could then forage near home and not lackwater. Nevertheless, duringmost of the year and in most localities,!Kung children did very little foragingand were dependentupon adults for theirfood until they were in their mid-teens. Our field data on the Hadza(Table 1) are supportedby most previous observers(the usual comment is, "womenand children set offto gatherberries, roots or baobabfruit" [Obst 1912]). More recently,Hadza informants in all Hadzaregions reported that childrenforage successfully. Hadza children (and women)seem to have a particularfood in mindeach time they leave camp, and dominantfoods differwith season, location,and year. We have data on Hadzachildren foraging with and withoutadult company and on several re- sources. Becauseof the manybrief and highly variable seasons of the Hadza environment(for example, Salvadora berries ripened in September1988, Oc- tober 1982, andNovember 1986), complete information on Hadzaforaging will take many years to accumulate,but it shouldbe clear that Hadzachildren foragemore than!Kung children.

DerivingHypotheses about !Kung Children's Foraging We have shown that observersand informantsagree that !Kungchildren did littlegathering work, andthus they were almostentirely dependent upon adultsfor food. Adults further assert thatthey activelydissuaded children from gathering(Blurton Jones, Hawkes, and Draper1994). These data, do not, however,give the economicvalue, real or potential,of children'swork. The questionis, Would!Kung have been better off (in the calculusof costs and benefits)if childrenworked? Cost-benefit analyses rely on arguingthat what individualsdon't do gives worse outcomesthan what they do do! The question, What would happenif they did somethingelse? can best be answeredby experiment. We proposethat alternativesto encouragingchildren to stay in camp,such as allowingthem to foragewithout adults or invitingthem to come to the

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions FORAGINGRETURNS OF !KUNGADULTS AND CHILDREN 223 grove, yieldedtoo little gainin food to outweighthe costs entailed.Thus we must assess consequencesof such alternativestrategies. We consider (1) childrenforaging unaccompanied by adults,as Hadzachildren often do, and(2) childrenaccompanying adults to the mongongonut groves on trips similarin distanceto the trips Hadzachildren take with adultsto berrypatches. Differencesbetween the flat, featureless,scrub-covered countryside in which the !Kunglive andthe undulating,landmarked country of the Hadzaled us to expect that unaccompanied!Kung children may be unableto find their way home. In BlurtonJones, Hawkes,and Draper (1994), we reportinterviews of !Kungadults that showedthe hazardsof unaccompaniedchildren getting lost in the bush. In a recent series of interviewswith Hadzaadults, very few ex- pressed the view thatchildren could get lost; most indignantlydenied that this couldhappen. Lee (1979)and Yellen and Lee (1976)described the distribution of plantcommunities in the Dobe areaas a series of parallelbands stretching east-west across the countryside(see Figure1, below).The most productive habitatswere far frompermanent water, andeven these becamedepleted as the dry season progressed.Thus we proposedthat unaccompaniedchildren in the Dobe areafaced an unfavorabletrade-off between the likelygains in food

dune molapo dune flats molapo dune ...... '...... flats

hardpan water course hardpan

Figure1. SchematicMap of the DobeArea Illustrating Parallel Distribution of Habitatsand Resources Mongongonuts grow on stabilizeddunes ("dune"), Grewia berries on "flats."Molapo refersto low areasthat hold water in the rainyseason but are comprisedof hard unproductivesoil in the dry."Hardpan" is the thinsoil over the calcretebase of the valleysthat sometimes channel substantial water during rain and underground water in the dryseason. The nine permanent water sources in the Dobearea are found where the calcretehas been penetrated by humanor otheraction.

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions 224 JOURNALOF ANTHROPOLOGICALRESEARCH intakeand the costs arisingfrom the distancethey had to travelto get significant foodreturns. The fartherthey went, the longerthey were exposedto the risk of losingtheir way, exhaustionand thirst, or encounterswith dangerousani- mals. We proposedto measurefood returnsat differentdistances from the shallow"valley" that holds permanent water (Experiment 1). These measure- ments coulddisprove our hypothesis-food may be plentifulnearby. We proposethat childrenwho foragein the companyof adultshave no risk of gettinglost andnegligible risk from predators. !Kung adults complained that childrenget tiredand thirsty and slow theirjourneys, but Hadzachildren often go 5-6 km each way withadults. We measuredthe rate at whichchildren and adultscrack mongongo nuts so thatwe couldinvestigate the economicsof nut collectingand processing(Experiment 2). We comparethe rate at whicha womanand child would make food available for consumptionif a childover age 8 accompaniedits motherto the nut grove with the rate achievedif the child stayedhome andcracked nuts fromtime to time. Processingmongongo nuts is quitetime consuming,and they cannotbe eaten untilthey are cracked.Lee reportsthat children over age 8 cracknuts for themselves and younger siblings. Sincemongongo nuts are stillused by !Kung,it was safe to assumethat today's childrenwould be as experiencedand competent at this artas in bygonetimes. The experiencedethnographer will doubt that a series of experimentscon- ductedin one field season can informus aboutlonger term conditions.We went to greatlengths to checkthe relevanceof our measuresto the context thatinterests us: the !Kungand their dry seasonhabitat up to the earlyseven- ties as describedby Lee andby Howell(1979). In addition to localand temporal variation,the environmentmay have been so modifiedby the increasedpop- ulationof livestock(Solway and Lee 1990)that our results would be misleading. We describeour effortsto deal with these issues of validityin the Appendix. Experiment1 assumesthat the experimentalforaging returns reflect availability of food. To test this, we surveyedfood availablein a series of quadratsat differentdistances from the valley(see Appendix:Experiment 3).

EXPERIMENT1: FORAGINGEXCURSIONS

Wedid not wantto findout whereor howmuch !Kung are foragingnowadays but to measurewhat children and adults could have acquiredif they hadgone differentdistances from dry season camps.The distributionof the plantcom- munitiesand habitatsin the Dobe area is a series of bands,running roughly east to west androughly parallel to two mainvalleys in whichmost of the nine permanentwater sources are found (Figure 1). It seemedappropriate to sample this countrysidewith a north-southtransect of the dune-molapobands to the northand west of Dobe, usingthe roadalong the fencedborder with . A valuableopportunity to controlfor effectsof livestockwas presentedby this barrier.The borderfence, which!Kung are allowedto cross, butbeyond which Bantu and their livestock had been prohibitedfor some years, formedan immenseexperimental enclosure. Livestock have destroyedthe habitatin the

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions FORAGINGRETURNS OF !KUNG ADULTS AND CHILDREN 225 immediatearea of the now-permanentsettlements. These settlements are locatedat permanentwater in the two mainvalleys, referred to by Lee as the !Kangwavalley and the /Xai/xaivalley. These shallowvalleys have a rocky floorbeneath which water stays undergroundall year-round.But the wateris onlyaccessible where the rockis brokenthrough by humanor other agency. By makingour observations near the valleybut a littleaway from settlements, we hopedwe were samplinghabitat representative of dry season campsites but outside the zone of destructionaround contemporary settlements. We drovepeople by truckto variouspoints, let them foragethere, andrecorded whatthey collected. Draperspread the news that we wouldlike to take womenand childrento the bush to observe them gatheringfood. The response was enthusiastic,a majorproblem being to sharethis privilege evenly among the !Kungsettlements locatedat andnear Dobe. In all,fifty adults and teenagers and fifty-one children andinfants participated in our foragingexcursions. Between five andthirteen adult women and teenagers, plus children,were collectedup eachday for a tripby truckto the bush.We measured the distance alongthe borderfence by the truckodometer. The locationat whichwe stopped for peopleto foragewas sometimesdetermined by theirrequests and some- times by our wish to samplethe environment.On most days we went to one locationby popularrequest and one locationof our choice. Each time we stopped,we asked people to gather,and we gave each persona plasticbag into whichto put the foodand bring it backto us to be identifiedand weighed. We sent people to forageon a specifiedside of the border.We were helped by ourfield assistant Timon Mbatara, who spokeEnglish and !Kung in addition to his Hereromother tongue. After people had been foragingfor fortyminutes to an hour,we calledthem back by soundingthe carhorn and shouting. During these foragingsessions, peoplewalked away from the carbut did not get much more thanhalf a mile away. After weighing,the food was given back to the personwho collectedit. Peopleinvariably took the food home with them. Groupsof peopleforaged for twenty-onesessions on the Botswanaside and thirteensessions on the Namibiaside. Some sites were visitedmore than once (e.g., the farthestnut groves). The locationswere reasonablywell distributed by distance,between 1.5 km south and 17.5 km northof our pointof origin at the watercourse,and by altitude,habitat, and soil type (dune,flats, molapo, andhardpan). Excursions to otherlocations, including a series around/Xai/xai, are describedin the Appendixand suggest thatour main transect represented a substantialportion of the environment. The calorievalue of each person'scollection of foodwas calculatedfrom its weightand the figuresin Table2. We have been unableto finda caloricvalue fortree resin(gum) collected and eaten by !Kung.Small amounts were collected during our excursions, especially where other foods were scarce. The mean "on-site" returns at each distance (calorie value of the food col- lected per hour gathering at the experimental location) are shown in Table 3. The yields of each adult and teen are marked in Figure 2. (If !Kung children

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions 226 JOURNALOF ANTHROPOLOGICAL RESEARCH TABLE2 Calorie Values of Main Foods Taken in July-August 1988

Name Kcalories per Kcaloriesper 100 gm eaten gm collected Mongongonut 654 0.949 Marulanut 642 0.949 (1) Tsin bean 544 3.808

!Gwa berry 293 1.46

N/n berry 227 1.13

Other berry -- 1.3 (2) /Tan root 47 0.47 Water root 31 0.31 Sha root 79 0.79

Otherroot 63 0.63 (3) Bitter melon 21 0.105 (4)

Note: Column1 fromLee 1979 (TableD.1). Column2 calculatedfrom edible portion as fraction of amountcollected (see Lee 1979:Chapters6 and7, AppendixD). Plantnames as in Lee 1979. 1. We haveno dataon the edibleamount of marulanuts per weightcollected. It is probablyless thanfor mongongonuts, andprocessing time is probablylonger; but we haveconservatively estimatedit as the same as for mongongonuts. 2. Otherberries are assumedto be halfwaybetween the figuresfor !gwaand n/n. Few other berrieswere taken,so errorswill not be large. 3. Otherroots are assumedto be halfwaybetween the figuresfor sha and/tan, assumingwater root to have an unusuallyhigh water content. We haveassumed that littleroot is discarded, whichslightly inflates kcal/gm collected. 4. We used estimatesof amountsof meloneaten by our experimentalsubjects; substantial rind is alwaysdiscarded.

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions FORAGINGRETURNS OF !KUNGADULTS AND CHILDREN 227 TABLE3 "On-Site" Returns from Experimental Foraging Excursions

Km north Habitat N Botswana Namibia N Namibia from water kcal/hr kcal/hr course

-1 Molapo 8 37.5 7 34.4

1 Molapo 9 54.7 7 5.8

2 Flats 5 131.0 5 115.4

3 Flats 13 295.9 9 353.3

4.1 Flats 12 346.3 8 0

5.7 - 5.9 Dune 22 330 8 253

7 Molapo 5 694.2 5 748.4

9 - 9.5 Flats 17 482 12 629.5

9.6 - 9.9 Flats 7 655.0 11 561

11 Flats 9 374.8 9 274.6

13.3 - 13.9 Dune 7 1596 13 3055

14.4 Flats 9 621.4 9 560.1

15.8 Dune 11 5219.9 0

17.2 Dune 7 4015 0

17.9 Dune 5 1962 5 722

18.6 Dune 11 1769.2 0 -

18.9 Flats 9 597.2 0

Note: Returnsare expressedin meankilocalories per hourper person.N = numberof adults andlate teens in sample.

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions 228 JOURNALOF ANTHROPOLOGICALRESEARCH

4500

4000 1 * K c 3500 a 3000 / * h 0 r 2500 *0 o 2000 0 n 1500 s S0 " S t 1000 ?, o0

500 .? * 00 o1 o -2 0 4 6 8 .010 12 . 14 16 18 , 20 Km north of valley * Botswana side o Namibia side Figure2. "On-site"Returns Obtained by Adultsand Teenagers during Experimental ForagingSessions Eachpoint is one person'shaul at one locationon one occasion.Values greater than 4,000are rounded down to around4,000-4,500 to aidscaling the diagram. only forage a little in the wet season, their performancein the experiment mightreflect lack of expertise,not an unbountifulenvironment, so we ignored our data on acquisitionrates of !Kungchildren. This biases againstshowing the resultsthat we expect.)These initialfigures exclude processing time. They were convertedto returnsfor hypothetical foraging trips in whichpeople would have walkedfrom the valley (at km zero) and subsequentlyspent time proc- essing the food.To minimizethe effectof journeytime, we assumedmaximum possibledaylight trip durations (twelve hours), from which we subtractedjour- ney time (round-tripdistance + 4 kph walkingspeed) to give time in patch duringwhich resources were acquiredat the rates we observedin eachlocality. We subtractednothing for energyused in walking.We used Lee's (1979:198) data on processingmongongo nuts (andour datafrom Experiment2, which closely matchLee's). This nut-processingtime is extensive (0.833 hoursper kg of mongongonuts collectedplus time for roastingthe nuts) andis probably more for marulanuts. To judgefrom Lee's accounts,processing time is mod- erate for berriesand fairlybrief for roots. We treatedtime spent processing roots andberries as zero. The results, foragingreturns with travel and processing time included,are shownin Table4 andFigure 3. These returnsare measuredin a way closely comparableto the returnrates reportedin BlurtonJones, Hawkes,and O'Con- nell(1989) for Hadza children (which were obtainedfrom follows of spontaneous

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions FORAGINGRETURNS OF !KUNG ADULTS AND CHILDREN 229 TABLE4 Kilocalories per Hour of Work, Including Travel and Processing Time

Kcal/hr Totalkcal includingtravel collectiblein and Km Habitat Kcal/hron site daylighthours processing' - 1 Molapo 36 414 34 1 Molapo 33 379 32 2 Flats 123 1353 113 3 Flats 319 3349 279 4.1 Flats 346 3443 287 5.9 Dune 311 2815 189 (727)2 7 Molapo 721 6128 511 9.5 Flats 555 4024 335 11 Flats 350 2275 189 13 Dune 2325 12787 542 14.4 Flats 591 2837 236 17 Dune 2738 9583 462 18.9 Flats 597 1522 127

Note: Calculatedfrom mean calories per hour"on-site" at each distancesampled. 1. Calculationof foragingreturns (including time to walkto patchand return and processing time). Calculationfor nuts (processingtime 0.833 hoursper kilogramplus 0.33 hoursfor roasting loadof nuts; collectingtime maximized by allowing12-hour excursions): a. Caloriesper hourin patch b. Gramscollected per hour c. Distancein kilometers(odometer or map) d. Time to patchand back at 4 kph[(c x 2) + 4] e. Time in patchif all daylighthours are spent foraging(12 - d) f. Kilogramscollected (e x b) g. Time to processthis manykilograms of nuts (f x 0.833) h. Totaltime gatheringand processing (12 + g) + 0.33 to roastwhole load i. Caloriesobtained in the 12-hourexcursion (a x e) j. Caloriesper hour(i + h) Calculationfor other foods (berriesand roots; processingtime ignored,but collectingtime maximizedby allowing12-hour excursions): a. Caloriesper hourin patch c. Distancein kilometers(odometer or map) d. Time to walkto patchand back at 4 kph[(c x 2) - 4] e. Time in patchif all daylighthours are spent foraging(12 - d) i. Caloriesobtained in the 12-hourexcursion (a x e) j. Caloriesper hour(i + h) 2. Intentionallyinflated figure obtained by usingthe "on-site"returns recorded in the dunesat 13 and 17 kilometers.

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions 230 JOURNALOF ANTHROPOLOGICALRESEARCH

800

700

600

K 500 c a I 400 r 300 2001

100

-1 1 2 3 4.1 5.9 7 9.5 11 13 14.4 17 18.9 Kmfrom water course Figure3. Returnsin CaloriesAcquired per Hour of Foragingand Processing for HypotheticalForaging Trips to DifferentDistances North of !KangwaValley Datafrom Table 4. foragingexcursions). In Figure3 we inflateone !Kungdata point--returns for tripsto the Dobenut grove-by usingin-patch rates frommore distant groves, lest Dobe is more heavilydepleted these days becauseso manypeople have donkeysto carrynuts. There is a cleartrend for "on-site"returns to increase with distancefrom the valley.Many more calorieswere collectedper hour fromdunes (nut groves) than from flats. Returns including travel and processing time showedan increasein returnswith distancefor the first 7 km or so but decliningreturns as journeysof over 20 km round-tripare involved.

Conclusionfrom Experiment 1 Food was gatheredat a greaterrate at locationsfar from the valley, the onlysource of dry season water.There was no consistentdifference between the Botswanaside and the Namibiaside of the border(see Appendix).The rate of returnfrom foraging excursions near the valleywas very low. When traveland processing time is included,the returnrate nowhereexceeded 730 kcal/hr.

EXPERIMENT2: PROCESSINGMONGONGO NUTS

We wantto knowwhy !Kungchildren seldom accompanied adults to the nut groves fromdry season camps.Would there have been a materialadvantage or disadvantageif childrenwent to the nutgroves? Would a womanmake edible caloriesavailable to her childrenat a greaterrate if she took her 8-year-old

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions FORAGINGRETURNS OF !KUNGADULTS AND CHILDREN 231 withher to the mongongonut grove or left it hometo cracknuts for itself and youngersiblings? The mongongonut is well knownto anthropologistsas the stapleplant food of the !Kung.Less celebratedis the cost of this food,especially its processing, describedin detailby Lee (1979). The nut has an outer layer,edible early in the season, but the mainnutritional benefit comes fromthe kernel;to reach this the nutmust be crackedopen. Lee (1979)reports that children over about 8 crackmost of the nuts thatthey eat, andthey havebeen observedgathering nuts duringthe rainyseason when familieswere campedin nut groves. But accountsseem to agree that childrendid not accompanywomen to the nut groves duringthe dry season. Sincewe have observedHadza children making trips of 5-6 km to berrygroves accompaniedby adults,we need to examine the possibleconsequences of !Kungchildren accompanying adults on trips to the nut grove duringdry season. Experiment1 gave dataon childrenpicking up nutsin the grove. InExperiment 2 we triedto measurehow rapidly children, teenagers, and adultscracked mongongo nuts. We use the results andother figuresfrom Lee (1979) to calculatethe rate at whichedible calories can be madeavailable to children.

Proceduresand Results We timedthirty-one people cracking mongongo nuts: eight adults, six teens (estimatedage 14-17), andseventeen children (aged 5-13, mostly9-13). Nuts were collected,usually by our assistants,sometimes also by us. They were roastedby our!Kung assistants, as is usuallydone to preparethem for cracking (Lee 1979).We counted out 20 nutsand weighed them. They were thenhanded to the subject,who was seated with suitablerocks that hadbeen selected by our !Kungassistants. The subjectwas timedcracking the nuts untilhe or she hadfinished. Two smallboys were givenonly 5 nuts. One was stoppedwhen it was clearthat, even thoughencouraged by his mother,he couldnot safely crackany. The other took nearlyeight minutesto crackhis 5 nuts.'Both of these subjectsare omittedfrom the childfigures discussed below. The dis- cardedshell and nut flesh kept for consumptionwere weighed. Six adults crackedon average314 nuts/hr.;the six teens, 241 nuts/hr.;and seventeen children,120 nuts/hr.Boys andgirls crackedat the same rate. Two adults(an old, but active, manand N/isa Kxau, our energeticassistant in Experiment3) were given 100 nuts to crack,and they workedmuch more slowlythan the adultsand teens given only 20 nuts (180/hr.versus 314/hr.). The rates for teens and adultsgiven 20 nuts to crackare comfortablyclose to Lee's "200-300nuts per hour"(Lee 1979:198),but the slowerpace of the adultsgiven 100 nuts to cracksuggests that we shouldnot expect anyoneto processlarge numbers of nutsat the highestspeed. Lee's instance(1979:198) of a womancracking 558 nuts in 156 minutesis a rate of 215/hr.Below we use 200 nuts/hr.,or 1.2 kg/hr.for adults. Some of the childrenwere not requiredto processthe innershell (it seemed to be difficultfor them). A subsamplewas asked to do this. They processed

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions 232 JOURNALOF ANTHROPOLOGICALRESEARCH only 100 nuts/hr.(631 g/hr., 599 kcal/hr.).We also doubtwhether children wouldkeep up the speed that we recordedfor more thana few minutesat a time. A reasonableestimate for children'snut processingseems to be 500 g/ hr. If the childrenare rankedby relativeage, there appearsto be a gradual increasewith age to the teen rates above 1 kg/hr. In the calculationsreported below, we use the followingvalues. Mother carries12 kg of nuts home fromthe nut grove. One kg of unprocessednuts yields949 ediblecalories after processing. The round-tripto a nutgrove takes three hours. Adultsgather nuts at just under10 kg per hour (12 kg in 1.25 hours),a muchhigher rate than we observedin 1988. Lee (1979:193)estimates that collectorscan gather20-30 fruits/minute,collecting "a few kilogramsof wholefruits" "every 20 minutesor so"and adding these smallloads to a central pile. He estimatesthat three to five such smallloads make a fullbackload of "10-15 kg of wholefruits." Our calculations show that the rate at whichnuts arepicked up has littleeffect on the caloriesacquired per hour of foodgathering andpreparation. Nuts are normallyroasted before cracking. We assume, based on our timedobservations, that a 12 kg loadtakes 0.33 hours to roast. We assume that childrendo not have to roast their own nuts; they are roasted alongwith mother's.We assumethat childrenpick up nuts at the adultrate.

TwoPossible Nut-Harvesting Strategies Wouldwe expect an 8-year-oldchild, who can crack mongongo nuts at about 500 g/hr., to go to the grove to help mothergather nuts or stay home?A !Kungchild who broughthome about2.5 kg of nuts (a little more than the loadof berrieswe observedHadza children carrying) and processed them itself wouldbe almostself-sufficient in plantfoods, andits mothercould use her load to feed youngerchildren. Children over 8 mightbe able to carrymore than 2.5 kg, andbelow we explorethe payofffrom carrying loads up to 10 kg. But a childwho stays home and cracksnuts saves its mothersome time. Which strategyworks out best? Do !Kungparents and children follow this strategy? Withinour paradigm,"best" refers to the reproductive-economicinterests of individuals.Mother gains by maximizingthe rate at whichshe makesfood availablefor consumptionby herself and her children.The childreceives a reproductive-economicbenefit from increasingits own food intake and (ac- cordingto kin selection)from increasing the food intakeof youngersiblings andmother in so faras this enablesmother to providemore sibs. The interests of motherand child do not completelycoincide (Trivers 1974); the child'sfitness benefitfrom aid to sibs willbe less thanmother's benefit. We mightgenerally expect the childto workless thanwould maximize mother's gain, but in the presentinstance, more workmay directlyhelp the childas well as mother. Thus we assess "teamreturns," the benefitsin kilocaloriesmade available for the childrento eat, andthe costs in time. Wedo not addtogether mother's work time and child'swork time but computethe time that elapses to make foodavailable: time spent on the journey,picking up nuts, roastingthem, and crackingthem. We call this "elapsedtime"-the worktime from mother setting

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions FORAGINGRETURNS OF !KUNGADULTS AND CHILDREN 233 off to the nut grove untilthe last nut is crackedand available to eat. We look to see whichstrategy makes food available to childrenmost rapidly,measured as kilocaloriesper hourof elapsedtime. Lee (1968, 1979)reports that women went to the grovesroughly every two to three days, not daily.Our calculations fit well with these observations:no one couldcrack a fullload of nuts on the same day as she gatheredthem. It takes anotherone to two days, unless peoplework very harddawn to dusk, andwe assumethey have otherworth- whileactivities to pursue.Our models also show thatincreasing the weightof nuts motherbrings home increasesmother's kilocalories per hourvery little once the loadspass 12-15 kg, close to Lee's reportednormal loads of 10-15 kg (Lee 1979:193).This impliesthat the few kilogramsa childadds may not be very useful--the nuts still have to be cracked. Our calculationsare illustratedin Table5. A childwho goes to the grove and carriesback some nuts (a "carrier")increases the team's income-the total weight of nuts broughthome. A childwho stays home and cracksnuts (a "stay-athome") saves mother time and reduces the costs-the team's elapsedtime. We used two very similarmodels to examinethis trade-off.Our firstmodel considers only the approximatelyfive hoursit wouldtake for a child to go to the groveand back. We assume that carriers and stay-at-homes behave in the same way the rest of the time (andour conclusionis not affectedby whetherwe assume childrendo any crackingduring the time when mother and childare both at home). Duringthe five hours, the carrierchild brings home a specifiedweight of nuts. This weightis addedto the weightmother broughthome (always12 kg in these calculations).The stay-at-homecracks nuts at its specifiedrate for zero to five hours.This saves mothersome time, calculatedas the gramsthe childcracks in anhour divided by the gramsmother cracksin an hour,multiplied by the hoursduring which the childcracks. This savingis subtractedfrom the time motherneeds to crackher load of nuts. (Followingthe ethnographiesand our informants,we assume that mother leaves childrenat home with some uncrackednuts.) Returns,in kilocalories per hour, are calculatedfrom mother load plus childload (convertedto kilo- calories),divided by elapsedtime, the time mothertakes on the tripplus the time she needs to crack(which is shorterif she had a stay-at-homehelper, longerif there are more nuts). It turns out that at the values tested (1-10 kg carriedby a child;stay-at- homecracks for 0-5 hr. at 500 g/hr., 750 g/hr.,or 1,000 g/hr.),the time saved by the stay-at-homeincreases the teamreturn rate more than the extracalories broughthome by the carrier.We calculatedthe hoursof crackingneeded for the stay-at-hometo matchthe returnrates achievedby a carrierwho brings 1-10 kg home. A childwho cracksat 500 g/hr. (our estimatedrate for 8-10 year olds) must crackfor 3.23 hoursto matchone who carrieshome 5 kg of nuts. Onlyif the carriercan bringhome 10 kg does the childmake better use of the 5 hoursspent on the tripthan a stay-at-homecould. As the childgrows up, it canpresumably carry heavier loads and could bring homemore. Intuitionwould suggest thatthere shouldthus come an age when

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions 234 JOURNALOF ANTHROPOLOGICAL RESEARCH TABLE5 "Team Returns" from Mongongo Nuts

Mother Carrier Team 1 Stay-at- Team 2 home

Child work 5 5 hr

Journey 1.5 1.5 1.5 0 1.5 out

Collect 1.25 1.25 1.25 0 1.25

Load (kg) 12 5 17 0 12 Journey 1.5 1.5 1.5 0 1.5 home Roast nuts 0.33 0 0.33 0 0.33 Hr Child 5 - 5 cracks 4.25=0.75

C saves (1) .42 x .75 .42 x 5 =.31 = 2.1 M crack 10 14.2 -.31 10- 2.1 = 13.89 = 7.9

Team time 18.44 12.48 (2) Kcal 11388 4745 16133 0 11388

Kcal/hr 875 912

Note: Calculationsto see whethermother gets morehelp from taking her childto the nut grove or fromleaving it at home. Team1 comprisesmother and a childwho goes to the grove to carrynuts. Team2 comprisesmother and a childwho stays homeand cracks nuts for the same amountof timeas the childin team 1 works. Bothchildren crack at 500 gm/hr. Bothmothers crack at 1,200 gm/hr.Team 2 gainsthe most kilocaloriesper hourof work. 1. Childsaves. Time savedby childcracking nuts = child'scracking rate + mother'scracking rate x hoursfor whichchild cracks; e.g., 500 + 1,200 x 5 = .42 x 5 = 2.1 hours.This representsthe timemother would have spent crackingthe nuts thatthe childcracked. 2. Teamtime. This is the time fromthe beginningof the journeyuntil the last nut is readyfor eating.It is notmother's work time plus child's work time.

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions FORAGINGRETURNS OF !KUNGADULTS AND CHILDREN 235 childrenbegin to accompanymother to the grove. Butour data on nutcracking by teenagers show us that as childrenage, they also get better at cracking nuts. The resultis thatchildren who crackat 750 g/hr.need onlywork 2 hours to matcha childwho carries5 kg and for 3.33 hours to matcha childwho bringshome 10 kg fromits 5-hour tripto the grove. If they crackat 1,000 g/ hr., they need work only 2.5 hoursto matcha 10 kg load. This leads to the surprisingsuggestion that there shouldnever come a time whena child,even a teenager,accompanies mother to the nut grove. We are remindedof Lee's (1979:265)comment that "adolescentgirls do not start regularsubsistence work untilmarriage." After marriage,they presumablybegin to have repro- ductivegoals of theirown, andthey sharetheir mother's goals less thanwhen they were single. The sameconclusions arise from our second, basically similar model in which we calculatethe consequencesof childrenworking from 5 to 15 hourson each loadof nuts. We assess the wholeperiod from mother departing until the last nut is cracked.(In this modelit does not matterwhether mother leaves nuts for childrento use whileshe goes to the grove.) One childgoes to the grove andbrings back a loadof nuts, whichis addedto the teamload, but the duration of the tripuses up some of its totalwork time; the remainderis spentcracking nuts. The stay-at-homespends all its work time cracking.Cracking saves mothertime, calculatedas in the first model. The savingis subtractedfrom the time motherwould need to crackthe loadof nuts. At longwork durations andfast crackingrates, the mother'scracking time plus triptime is less than the child'swork time. At this point, elapsedtime must be set to the child's work time. At most values, this results in a lower returnrate (becausetoo muchof the crackingis beingdone at the child'sslow rate instead of at mother's faster rate). Thus there is an optimalduration of childwork beyondwhich returnsdecline (Figure 4). This optimumis at a higherreturn rate andshorter durationfor stay-at-homesthan for carriers-8-10 hours in a two-daynut cycle, 4-5 hoursper day. As in the firstmodel, it looksas if leavingthe childhome is the best strategy. This secondmodel adds the conclusionthat when the childstays at home, the team gets a greaterincrease in team returnsfrom each addedhour of child work thanwhen the childgoes to the grove. It also supportsthe suggestion thatteenagers should stay homeand that growing strength will not bringabout a time when the childshould switch to the carrierstrategy. Conversely, both modelssuggest that if youngerchildren crack nuts at a slower rate, but can carry, say, 5 kg of nuts, they wouldbe expected to go to the grove. How wouldwe explainwhy childrenaged 6-8 did not go to the grove? Ouradult !Kunginformants emphasized children's thirst and exhaustion. In the following we assumethis exhaustion translates into slower walking and thus into a longer time for trips to the grove. If the childslows motherdown on her tripsto the grove, we can calculate the consequencesby addingto mother'stravel time in the spreadsheetfor

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions 236 JOURNALOF ANTHROPOLOGICALRESEARCH Team Returns Child carries 5 kg 1300

T 1200 e . m 1100 ...... Ke...... K 1 0 c 0...00 , a " a a ...... I 9 0 0 ...... h : r 800

700 5 6 7 8 9 10 11 12 13 14 15 Time C works

o 500 Carry -- 500 Home - 750 Carry 750 Home -- Figure4. Kcal/hr.Made Available to Children Kcal/hr.are plotted as a functionof the hourswhich a childover 8 worksat eachof two strategies:accompanies mother to nutgrove and carries 5 kg nutshome (dotted lines);stays home and cracks nuts for self and persuasive siblings (solid lines). A pair of linesis shownfor children who crack nuts at 500g/hr., and a pairfor children who crackat 750g/hr. (low teenage rate). eithermodel. Adding half an hour(a mere 15 minutesto eachleg of the round- trip)lowers the teamreturn rate by 25-30 kcal/hr.when the childgoes to the grove. Addingan hourlowers the returnrate by around50 kcal/hr.Gains in team returnper hour of childwork fallfurther behind the rates achievedby leavingthe childhome. Even a childwho can only crackat 400 or 300 g/hr. but can carry5 or 2.5 kg does better to stay home if he slows motherdown by about20 minuteson the round-trip.Thus the informantsmay have pointed to exactlythe aspectof nuteconomics which tips the balancein favorof leaving youngchildren home. We concludethat there is little supportfor the notionthat !Kungmothers or childrenwould have benefitedfrom children accompanying adults to the nut grove.The reasonhas much to do withthe relativelylengthy processing needed to makemongongo nuts edible. DISCUSSION

The Validityof theExperiments The Appendixdeals with several importantquestions about validity. Live- stock have destroyedbush foods arounddry season campsites, and Solway andLee (1990:118)assert that the habitathas been degradedenough to "se- riouslyundermine the foragingoption." Could our data on food availability

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions FORAGINGRETURNS OF !KUNG ADULTS AND CHILDREN 237 reflectavailability in previousdecades? Was local and annual variation so great that we couldnot generalizefrom our experiments?These are well-known limitationsof short-runobservations. The groundnear present-day settlements was almostbare, but we collected dataaway from settlements, sometimes near the valley,sometimes very far fromit, and on each side of the borderfence. We expectedto findeffects of livestockin our quadrats(Appendix: Experiment 3) and that these woulddi- minishgradually to distancesof 5-10 kmfrom the valley.Similarly, we reasoned that foragingreturns from the Namibianside of the borderfence (wherelive- stock were prohibited)would be muchricher than on the Botswanaside. We were wrong on both counts. The yields in Namibiawere no higherthan in Botswana.Quadrats with tracks of livestockdid not consistentlyyield less food thanquadrats without tracks. The negativeeffects of livestockwere localized to the area withinone kilometeror so of settlements.This outcomeis sur- prising,but the evidenceseems strong. Returnsfrom different localities support the generalpicture obtained from ourmain transect. Our simulation showed that variation in densityof mongongo nuts lyingon the groundwould have little effect on the caloriesobtained per hour of gatheringand processingnuts. The density of nuts on the ground affectsthe amountof time takento collecta load, whichis a smallproportion of the total gatheringand processing time. Traveland crackingcomprise the mainportion of the time requiredto move nuts fromthe grove to the mouths of a womanand her children.Our findings are, surprisingly,quite robust.

ChildrenForaging Unaccompanied by Adults Accordingto the literature,!Kung children seldom foraged. So first we asked, Whatwould happenif !Kungchildren tried to forage withoutadult companyfrom dry season campsin a habitatlike that aroundthe !Kangwa watercourse?We collecteddata on foragingby !Kungadults and children whom we tookby carto a varietyof locations.We showedthat, just as ourinformants told us, more food is foundfar fromwatercourses. Our quadrat data closely matchedthe plantcommunities and soil types describedby Lee (1979) and matchedour foragingdata. Ourforagers had not ignoredlarge and important sources of food. The soil types, drainagepatterns, and underlyinggeology seem to be the reason for the distributionof food plantsand the distanceof the richerpatches from permanent water. We have alreadyreported (Blurton Jones, Hawkes,and Draper1994) the evidence that !Kungchildren who leave campunaccompanied by adultsrun substantialrisks of gettinglost and, once lost, of dyingof exposure. We as- sumedthat the longera childunaccompanied by adultsspent out of campand the furtherit went, the greaterits risk of gettinglost. Tracksrecorded in our quadratsshowed that risks of encounterswith predatorsdid not increaseas a childwent furtherfrom the watercourse(but chancesof an encounterwith a predatorwould increase with time spent out of camp).The datapresented here on foragingreturns and availabilityof food show that the gains from

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions 238 JOURNALOF ANTHROPOLOGICALRESEARCH exposureto these risks wouldhave been very small,unless children travelled far (8 km round-tripsto productiveberry flats and 12 km round-tripsto nut groves). We concludethat foraging unaccompanied by adultswas not likelyto have been a rewardingactivity for !Kungchildren. The costs and the benefitsfrom children'sunaccompanied foraging seem muchmore favorable among the Hadza.Blurton Jones, Hawkes, and O'Connell (1989)report Hadza children aged 5-10 obtain629 kcal/hr.,and children aged 10-15 obtain1,014 kcal/hr.by collectingand processing baobab fruit, mostly fromwithin 200 m of camp. For !Kungto achievesimilar return rates, they wouldhave to travelas far as the Dobe nut grove, 5.5-6 km each way, much furtherthan we haveobserved unaccompanied Hadza children travel. A recent series of interviewssuggests that Hadzachildren suffer little risk of getting lost. In manyhours of followingHadza children in the bush, we have seen none show any uncertaintyabout the route home (they guideus!), nor have we seen their parentspanic like !Kungparents when the whereaboutsof a childis unknown.

ChildrenForaging in the Companyof Adults Foragingin the companyof adultsmay be an entirelydifferent matter. The presenceof adultsmust reducethe risks of childrengetting lost and preyed upon,and if Hadzachildren aged 7 canmake their 10 kmround-trips to berries, we cannotassume that !Kung children were incapableof a tripto the nutgrove. Whydidn't !Kung children go to the nut groves with the women?We used datafrom experimental sessions in whichchildren cracked mongongo nuts and datafrom Lee (1979)to calculatethe consequencesof two strategiesfor mother and children.We computedthe caloriesbrought home and the time takento collect and process them to see whichstrategy would make edible calories availableto a woman'schildren at the greatest rate. We used this measure becauseit can be expectedto determinethe growthand survivorshipof chil- dren. Mongongonuts cannotbe eaten untilthey are cracked,and the cost of crackingthem has a large influenceon the returnrate obtainedfrom them. Ourcalculations showed that mother and child gain more per hourof childwork if the childstays at home. Indeed,even if the unsupervisedchild cracks nuts for muchless thanthe durationof the trip, the mother-childteam do better if the childstays at home (see Figure4). It seems reasonableto expect a child to cracknuts wheneverit gets hungryand perhaps occasionally to cracksome more when naggedby a youngersibling. Surprisingly,younger children who are less efficientat crackingnuts would help mothermore by goingto carrynuts. But in the interviews,!Kung adults commentedthat if childrengo intothe bushwith adults, they get tired,thirsty, and"spoil the work"by pleadingto be takenhome. !Kungadults regard children as a liabilityon tripsinto the bush. We showedthat even a slightincrease in journeytime imposedby a tiredand thirsty child would tip the balanceclearly in favorof leavingthe childat home. Yet Hadzachildren accompany adults on trips to berrygroves that were at least as distantand long lasting(10 hours)

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions FORAGINGRETURNS OF !KUNGADULTS AND CHILDREN 239 as !Kungtrips to nutgroves. Whydon't they get tired?Would holding mothers backbe sucha penalty?Important differences in the environmentmay explain the situation.On the Hadzatrips to berrygroves, there is muchshade, water is availablealong the way and near the berry groves, and the Salvadoraand Cordiaberries themselves containmuch water. In contrast, on dry season journeysto nut groves in !Kungcountry, there is neithershade nor water.At the groves there is just enoughshade to take a rest in but no water.All water must be carried.Daytime temperatures are very high (100 percent of days above 33'C in Septemberand October,highs 33-45TC,i.e., 92-100'F [Lee 1979:104-6];globe thermometerreading at Tsumkwe,50 km west of Dobe, September-December1981, mode 116-1200F [BlurtonJones, Hawkes, and O'Connell1989]). Environmental physiology literature (e.g., Drinkwateret al. 1977; Drinkwaterand Horvath 1979; Squire1990) suggests that childrenare less capableof work in hot environmentsthan adults. But the surprisingly complexeconomics of foragingmay also playa significantrole. The Hadzachild obtains very highreturns in the berrygroves (1,220-1,440 kcal/hr.[Hawkes, O'Connell, and BlurtonJones n.d.]). The processingcosts are very low (the childcan eat the berriesas it picksthem), and our second model, constructedto analyzemongongo nut strategies, shows that when processingcosts are very small, accompanyingmother to the grove should pay,even if the childwalks more slowly than adults. The verylimited processing of berries,e.g., mixingSalvadora berries with water and baobab pith, maybe little affectedby the quantityof berries (likeroasting mongongo nuts, which seemed to take about the same time whateverthe quantityof nuts). This furtherdecreases the valueof an assistantprocessor and increases the value of an assistantgatherer. A finalpuzzle concerns times andlocations (such as /Du/da,where most of Draper'sdata were gathered)where mongongonuts are not widelyavailable andwomen forage on a varietyof berries(such as Grewiaretinervis) and roots. Why don'tchildren go with them?These are resourceswith low processing costs, like the distantberries that Hadza children exploit with adult company. Grewiaberries are abundantall over the Africansavanna and have quitelong seasons; they shouldhave been an importantresource for savannaforagers at manyplaces and times. But we knowtoo little abouttheir economics (food value, processingtimes) to add an economicargument to our informants' opinionthat childrenlack the staminato aidtheir food quest.

Explainingthe Difference between !Kung and Hadza Children'sForaging Opportunities Wesuggest that important reasons for the differencein foraging opportunities for childrenin the two localitiesare the spatialdistribution of water andfood andthe amountand nature of processingneeded by the availablefoods. In the Dobe area, the richestfood sourcestend to be farfrom dry season water (as describedby our !Kunginformants, evidenced by ourdata, and implied by Lee [1979:175,191-92, 203]). The Dobearea has only nine permanent water holes.

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions 240 JOURNALOF ANTHROPOLOGICAL RESEARCH We havenot countedthe waterholes in Hadzacountry but doubtthat we ever wouldobtain a completecount. We alreadyknow many that we andour Hadza hosts use routinely.Most are far too smallto supportlivestock or a large complexof camps,but there are some dozensclose togetherall over ourstudy area. If Hadzachildren should deplete their resources near camp, it wouldbe easy to move to anotherwater hole. Our effort to model the economicsof mongongonuts is most notablefor the complexitythat it uncovers.In discussing!Kung and Hadza strategies for exploitingplant foods, we hadto considerthe degreeto whichindividuals share reproductiveinterests, traveltime andthe costlinessof the travel,acquisition time (e.g., simplypicking food fromground or bush, roots in loose soil that are easily and quicklydug up versus roots that take long and heavy work), processingtime, and the degree to whichprocessing time depends on the amountof foodbeing processed (e. g., crackingeach nut versus roastinga pile of any size). We findthat we lacksufficient information on these variablesto attemptto calculatestrategies for !Kunguse of roots or for the use of Grewia berries by childrenin either population.An interestingelaboration of !Kung mongongonut use is that womensometimes crack nuts in the grove before returninghome. Ourlimited exploration of this issue suggests that it willnot requireus to changeour conclusionsabout children's foraging. The detailed economicsof food gatheringand processingseem likely to accountfor the frequencyand duration of !Kungwomen's foraging trips, the nonparticipation of childrenand teenagers,and severalof the differencesbetween !Kungand Hadzasubsistence.

Appendix Validity of the Experiments Field experimentsare unusualin anthropology,and it is importantto discuss their validity.Much of oureffort in the fieldwas directedtoward assessing validity-the relevanceof ourshort series of experimentsin onefield season to thetime and region thatinterests us. Sincemany authors have portrayed !Kung economics as relatively comfortable,and Solway and Lee (1990) described the recent environment as seriously degradedby livestock,we wereespecially alert for erroneously low estimates of the returnsobtained from foraging. We surveyed a seriesof quadratsin the Dobearea (Experiment3) and conductedforaging experiments in the /Xai/xaiarea (Experiment 4). We havealready shown that foraging returns on the cattle-freeNamibia side of the borderfence were no differentfrom those on the Botswanaside and describedhow we collectedno datafrom the obviouslydevastated country close to contemporary settlements. Experiment3: QuadratSurvey We wantedto show whetherour experimentalforagers were neglectingany rich foodsources (nowadays many !Kung own livestockor workfor herdowners, and they sometimesreceive food from government distributions, so perhapsour foragerswere not reallytrying). We wantedto understandwhy morefood was obtainedfurther from the valley.Was it due to depletionby peopleor livestock(as indicatedby presenceof

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions FORAGINGRETURNS OF !KUNG ADULTS AND CHILDREN 241 tracksin a quadrat),or to the locationof differentplant communities (dune, molapo, flats),or the seasonalityof berries?As therewere no permanentvillages north of Dobe andno permanentwater sources within almost 100 kilometers,we expectedthat signs of use by peopleand domestic stock woulddecrease as we drovenorth in searchof foraginglocations. At each of twelve pointsalong the Botswanaside of the border,beginning 1.5 km northof the valleystarting point and working north as far as 17.5 km, we surveyed three 15 m2quadrats. The distanceswere chosento representhardpan, molapo, flats, and dunehabitats (Lee 1979:93,fig. 4.3). Once the truckwas stoppedat our chosen distance(by the readingon the odometer),the exact locationof each quadratwas selectedby throwinga colorfullyflagged stick as faras possibleeast fromthe roadside. The fallenstick marked the southwestcorner of the firstquadrat. To locatethe second quadrat,we hurledthe stick east fromthe southeastcorner of the first quadrat.The same proceduremarked the cornerof the thirdquadrat. The aimwas to obviateany tendencyto choose quadratsin a microhabitatthat would support our expectations. Onfinding the markerstick, one of the three !Kungmen who helpedus as trackers walkedthe area, reportingon all tracksseen in the approximatearea of the quadrat. The tracksrepresent much of the passagethrough the quadratin the roughlythree monthssince the end of the rain. Manytracks were reportedas "inthe rains"and "sincethe rains,"none frombefore the rainyseason. Presumablythe tracksof larger animalslast longerthan those of smallanimals, since smalltracks may be filledmore rapidlyby windblownsand. We recordedwhether human tracks were foundand which species of wild animals,especially predators, and livestockhad left their spoor since the rains.Then we markedthe borderof the quadratwith string, using a compassand tape to makethe areaas nearlysquare and oriented north-south as possible.Next N/ isa Kxau(famed for her energyand hardwork) collected all the plantfoods that she couldfind in the quadrat.Her instructionwas to get all the food of all kindsthat she couldfind, howeverlong it took her. These foods were then identifiedand weighed. Then she walkedthe quadratwith one observerand pointed out andidentified all the berry bushes, fruitingand not fruiting.Her reputationas an efficientand energetic gathererwas confirmedwhenever she participatedin the experimentalforaging ex- cursions.The team was completedby our interpreter,Timon Mbatara. The procedurein nut groves was slightlydifferent. Individual trees were chosenby followinga compass-determinedpath until we came undera tree. Tracksin the area coveredby the tree were examined,and food foundbeneath the tree was collected. The areas were quite similarin size to our 15 m' quadrats.More thanthree trees were examinedin each grove, but the figures in Table 6 (below) are proratedto representthree trees, roughlyequivalent to the three quadratsreported for other locations.

Results Foodavailability, habitat, and distancefrom the valley. Table 6 showsthe totalcalorie valueof the foodfound in each quadrat(except tree gumfor whichwe haveno calorie value). The patternof availabilityof food followedthe patternof in-patchforaging returns.Available food increasedwith distancefrom the valley (correlationof .3452 between caloriesand kilometers,p <.006, N = 53) and was greaterin nut groves (dunes)(t = 3.0, p <.001, N = 53) and less in molaposthan on flats (t = 3.4, p <.003, N = 27). Furtherstatistical analysis showed that the locationof differenthabitats accounted

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions 242 JOURNALOF ANTHROPOLOGICAL RESEARCH forthe increasein foodwith distance from the valley.The correlationbetween distance and caloriesis reducedif examinedwithin a habitattype. If dunesare removedfrom the sample,the correlationfalls to .1193 (p = .277, N = 27). If dunesand molapos are removedand the quadratsfrom flats are examinedalone, the correlationis -.0606 (p = .397). For dunesonly, the correlationremains (r = .3474, p = .041, N = 26). Thusflats yielded the same amountof foodwherever they were, andmolapos yielded theirsmall amount of foodhowever far from the valleythey were. But dunes (where the nut groves are found)yielded more food the furtheraway we travellednorth from Dobe grove. This situationmay be evidencefor humandepletion of the Dobe grove. These results suggest that the reasonforaging yields increase with distancefrom the valleyis primarilythe distributionof habitattypes and,secondarily, the depletionof the nearestnut grove. TABLE6 Calories Obtained from Each Set of Three 15-Meter Quadrats

Km north Habitat Total yield Kcal from Kcal from Kcal from from water (Kcal) berries Roots nuts course

1.5 Molapo 14 14 0 0

3.5 Flats 106 0 106 0

4.5 Flats 319 163 156 0

5.9 Dune 625 0 0 625

7.0 Molapo 17 0 17 0

9.0 Flats 406 406 0 0

9.5 Flats 1111 1111 0 0

11.0 Flats 467 329 138 0

13.0 Dune 4398 0 0 4398

14.4 Flats 118 118 0 0

16.0 Flats 146 133 13 0

17.5 Dune 3173 0 0 3173

Berrybushes. Table 6 showsthe caloriesobtained from berries in eachset of quadrats. We also tabulatedthe numberof berrybushes andthe yieldof the dominantspecies of bushin kilocaloriesper bushfor each quadrat.This showeda predominanceof n/n bushes near the valley(just as Lee [1979:160]reported) and of !gwabetween 9 km and 14.4 km (flatsnorth of Dobe nut grove andChipi grove). But the datashow that n/n yieldedalmost no fruitat this time while !gwa were quite productive.The rank

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions FORAGINGRETURNS OF !KUNG ADULTS AND CHILDREN 243 orderof yieldat the timeof our observations(!gwa [19.75 cals/bush]> /tore [9.9 cals/ bush] > kamako[0.04 cals/bush]> n/n [0.03 cals/bush]> zoma)resembles Lee's (1979:484)sequence of fruiting(latest fruiting to earliest:kamako > !gwa > /tore > n/n > zoma).N/n may have been a useablefood sourcefor childrenat the beginning of the dry season, but evidentlythey do not last long, unlikethe faraway!gwa. Our data from /Xai/xai(Experiment 4) suggest that even !gwa near a valleymay not be very productivelater than July. Depletionof resourcesby people. Nuts in the Dobe grove were strikinglysparse, and 1988 was a year of heavy rain after manyvery dry years. As Lee (1979:189-90) reportedfor other such years, the mongongocrop was poor. AlthoughLee (1969, 1979:175,and elsewhere) writes of depletionof grovesclose to camp,we areconcerned that the sparsenesswe witnessedmight result fromthe combinationof a poor year with the greater ownershipof donkeysby !Kungin 1988 comparedto the 1960s. Donkeysare used to carrynuts, and their use mightaccelerate the depletionof the Dobe grove. Tracksof peoplewere foundunder five of the ten trees examinedin the Dobe nut grove, the nearestnut grove (andin only two other of the remainingforty- three quadrats).For this reason, we used the intentionallyinflated figure for returns fromtrips to Dobe grove (727 kcal/hr.)so that the "on-site"returns (calorie value of food gatheredper hourat the grove) are takento be as highas for the more distant groves (see Table 4, above). The almost completeabsence of tracks of people in quadratselsewhere gives littlesupport for the view thatthe low foragingreturns near the valley were due to depletion.With the modelof mongongonut economics,it is easy to show that no matterhow rapidlynuts are pickedup (whichdepends on how densely the nuts lie on the ground;we tried valuesfrom 1 to 20 kg/hr.),the return ratescannot rise muchabove those we havediscussed. Annual variation in the mongongo crop seems to be muchless of a threatto our argumentthan intuitionwould have suggested. Depletionof resourcesfar afieldby cattle. We recorded tracks of livestockin oursurvey quadratsand sought associations with amount of foodyielded. Associations suffer prob- lems of interpretationnot applicableto the "experimental"test. For instance,cattle may be attractedby grass and leave their trackswhere there is more grass. More grass may be foundwhere there is less othervegetation. A standof productive!gwa berrybushes leaves littlebut bareground beneath and betweenthe bushes. A t-test showedno differencebetween calories obtained from quadrats containing cattle tracks and those withno tracks.Analyses of varianceexamining distance and livestock gave no evidencefor an effect of livestock.Quadrats with tracks still showeda correlation of caloriesand distance (r = .4183,p = .008, N = 33). So didquadrats with no tracks (r = .3736, p = .052, N = 20). Whenhabitat types were included,there was a significantmain effect of dunesand none of livestock.But whendunes were excluded fromthe dataand a t-test performedto comparequadrats with cattle tracks with quadrats with none, a significantdifference was found.Quadrats in flats andmolapo with cattle tracksyielded less food thanquadrats with no tracks.This is the only indicationwe couldobtain that cattlemight have affectedfood availability,and as mentionedabove, this associationcould result from cattle seekinggrass and thus avoidinglarge stands of berrybushes. Severalof our trips were to distancesfar out of reach of dry season day trips on foot, a result of our persistent,but unsuccessful,efforts to show that higherreturns couldbe obtainedin remoteundisturbed habitat. We thinkit is impossibleto arguethat

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions 244 JOURNALOF ANTHROPOLOGICALRESEARCH the low returnsin some of the very remotesites we visitedwhere there was no sign of livestockwere, nonetheless,a resultof recentlyincreased numbers of livestock. Predators.Tracks of predatorswere foundin eleven out of fifty-threequadrats and were not associatedwith distance. If predatorsare a risk to children,they are a risk that is equallypresent near or far from valleys. The tracks, while representingan impressivearray of wildlife,were an accumulationsince the last rains about three monthsbefore data collection. In our thirtyforaging and quadrat days walkingabout in the bush, often very far fromsettlements, we encounteredonly one puffadder and one leopard.

Conclusionfrom Experiment 3 The quadratdata revealedno massivefood supplythat our experimentalforagers failedto exploit.Habitat type andberry season explain,but do not remove,the effect of distance.Richer habitats were fartherfrom the valley, and later fruitingberries (lastinglonger into the dry season)were also fartherfrom the valley.Though we can thus offer some explanationfor the distanceof food, the decisionsof foragersmust take into accountthe distanceat whichthe food actuallycan be found,regardless of the reasonfor its beingso faraway. According to ourdata, this inconvenient relationship betweenfood-bearing habitats and permanent water is not a resultof recentlyincreased humanuse of the land.

Experiment4: ForagingExperiments at /Xailxai Perhapsthere is so muchlocal variation in habitatthat generalizationis impossible. Weconducted foraging experiments around /Xai/xai, included in Yellenand Lee's (1976) definitionof the Dobearea and the site of muchresearch by Wilmsen(1989 and earlier). The permanentwater at /Xai/xaiis now a complexof wells clusteredin the valley, whichruns west to east throughthe villageand then curvesaway to the southeastto Drotsky'sCave and beyond.To the west, the valleybecomes wider and shallower, muchas does the !Kangwavalley as it approachesthe bordernear Dobe. Ourforaging tripswere aimedto locatehigh-yielding patches and to gainan impressionof the area around/Xai/xai, not to providea second transect. Some locationswere selected by !Kungparticipants, others by us. We list the resultsin note 2, givingdirection and distance of the foraginglocations fromthe valleyand from the village,the numberof peopleinvolved, the habitat,and the meanyields. Meankilocalories per hourcollected "on-site" were 719. This figure is not markedlydifferent from that aroundDobe (695 kcal/hr.).Returns calculated to includetravel time are also similarto those aroundDobe. As in the maintransect near Dobe, the biggestyields were in a mongongogrove. The best berryyields were again farfrom the valley.The valleyitself was almostcompletely lacking in bushfoods (though not lackinggrass). Thus our foragingexcursions around/Xai/xai and elsewhere (N!oon!oodimaand nut groves at Cheracheraha)gave similarreturns to thosepresented for ourmain transect. It wouldbe hardto arguethat we collecteddata in an untypically poor partof the Dobe area. However,Lee (1979:160)reports substantial berry groves just to the northof /Xai/ xai, whichhad apparently disappeared by 1988:"At /Xai/xai vast grovesof !gwabushes (Grewiaretinervis, B9) extendback from the panfor almost1 km on the northside." Thisis an importantobservation. If these !gwabushes near /Xai/xai were as productive as the !gwabushes 9 km northof Dobe andthose 5 km southand 14 km east of /Xai/

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions FORAGINGRETURNS OF !KUNG ADULTS AND CHILDREN 245 xai, we wouldhave expected everyone, including children, to haveexploited them, but our quadratdata show that the presenceof berrybushes is no guaranteeof berriesin the dry season. In 1988 we saw dense standsof Grewiato the southand east of /Xai/ xai, muchnearer the valleythan in the Dobe area. The valleyhad steeper banks,and berrybushes began within a few hundredyards of the bottom.But the returnsobtained fromforaging in these nearbyberry bushes were very, very low. Wedo not knowwhy; soil type or greaternumbers of birdsor insectsnear wet seasonwater might be factors. Perhapsthe season for fruitingis earliernear valleys, and thus by mid dry season, most fruithas gone. Thus if the berrygroves close to /Xai/xaiwere as unproductive as the berrygroves at Site A nearthe valleybut farfrom the village(see note 2), they wouldnot have lastedlong and might have been littleused by children.

Comparisonwith Lee's Early Data Whilecontemporary effects of cattlewere not evidentaway from settlements, there couldstill have been a longerterm degradation of the entireenvironment due to grazing and other overuse, as assertedby Solwayand Lee (1990). Lee reportedno dataon acquisitionrates or densitiesof foods,but in his 1979 AppendixD he reportsthat /tan roots "mayweigh up to 10 kg, but the more typicalsizes are 1 to 2 kg, and smaller specimensweigh as littleas 30 g" (Lee 1979:485).Twenty /tan roots thatour foragers acquiredweighed 195-2,700 g, mean 1,025 g. The mean diggingtime was seven minutes,which at Lee's47 cal/100g gives2,886 kcal/hr. upon encounter. Lee (1979:486) also commentsthat sha root "edibleparts rangein weightfrom 10 to 75 g, with an averageweight of 44 g." The mean of 231 pieces we weighedindividually was 64 g (14-97 g). These pieces are extractedin strings,and we calculateda returnrate upon encounterof 3,043 kcal/hr.No wonderour foragersstopped even for a singleplant. But these figuresgive no indicationthat roots were any smallerin 1988 than in the 1960s. Lee (1979:484)illustrates the collectionof !gwaand n/n berries(Grewia retinervis andG. flava), sayingthat "5 to 10 litersof the berryare typicallygathered by a single personin a few hours."The berries"follow a ripeningsequence that extends the Grewia season from Decemberto July and beyond."So we were collectingdata late in the berryseason. In berrygroves, our foragerscollected 434 g/hr.;women of middleage andup collected507 g/hr. Lee says a literweighed 311 g. If "afew hours"is at least three, then 5 liters = 1,555 g in three hours,a rate of 518 g/hr.If a few hoursmeans as muchas five hours,then 10 liters = 3,100 g in five hours,which is 620 g/hr.These figuresare in the highend of ourrange (166-861 g/hr.). They are low by Hadzastandards (Grewiabicolor at 440-1,000 g/hr. [Hawkes,O'Connell, and Blurton Jones 1989]). Hawkesand O'Connell(1982) calculatedreturns for a 6-hour mongongonut trip usingfigures from Lee (1968, 1979) as 670 caloriesper hour,a little lower thanthe 727 that we calculatehere for hypotheticaltrips to the Dobe nut grove. This close agreementsuggests that returnsfor mongongonuts have not significantlydecreased over the interveningyears. Ourimpressions of conditionsnear settlements could have led us to the sameextreme assertionthat Solway and Lee (1990)make. But we see nothingin the aboveanalyses to suggest thatdegradation affected the countrysideas a whole.The degradationnear watermay, nonetheless, achieve the effectSolway and Lee (1990:118)claim: "inability of theirland to supporta foragingmode of production."But anywho claim that change in the widerenvironment invalidates our experimentsmust presentthe data.

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions 246 JOURNALOF ANTHROPOLOGICAL RESEARCH NOTES

1. Weare grateful to the Officeof the President,Republic of Botswana,for permission to do researchin Botswana;to the NationalScience Foundation for support;to /"Ashe Kumsa,/Oma Bau, and !Xoma K'au for assistance in ourquadrat survey; and to Timon Mbatarafor interpretingand other assistancein the field. Hawkesand Blurton Jones wouldlike to thankboth Patricia Draper and Henry Harpending for theirhospitality and advicein the field.Harpending and Draper were supportedby a grantfrom the National Instituteof Aging. 2. We reportthree ways to calculatekilocalories per hour:(a) kcal/hr.spent "on- site"where food was sought;(b) kcal/hr.spent on-site, plus time spentprocessing the food and time spent walkingto site fromthe nearestpoint in the valley;(c) kcal/hr. spenton-site, plustime spentprocessing the foodand time spent walkingto site from the present-dayvillage of /Xai/xai. Site A: 0-2 km fromvalley, 16 km southeastof village,nine people, molapo-flats habitat,(a) 81 kcal/hr.on-site, equivalentto (b) 77 kcal/hr.including travel from valley andprocessing, (c) 27 kcal/hr.including travel from village and processing. Site B: 2 km, 9 km southeast,eight people, mongongo nut grove, (a) 1,699 kcal/hr. on-site, equivalentto (b) 658 and(c) 550 kcal/hr. Site C: 3.5 km, 12.4 km northwest,ten people, flats, (a) 1,136 kcal/hr.on-site, equivalentto (b) 970 and(c) 549 kcal/hr. Site D: 4.25 km, 14 km east, ninepeople, flats, (a) 459 kcal/hr.on-site, equivalent to (b) 378 and (c) 191 kcal/hr. Site E: 5.2 km, 5.2 south, five people, flats, (a) 541 kcal/hr.on-site, equivalentto (b) 424 and (c) 424 kcal/hr. Site F: 6.2 km, 15.6 km northwest,seventeen people, flats, (a) 401 kcal/hr.on-site, equivalentto (b) 297 and(c) 140 kcal/hr.

REFERENCES CITED

Barry,H., I. Child,and M. Bacon,1959, The Relationof ChildTraining to Subsistence Economy.American Anthropologist 61:51-63. Bleek, D., 1931, The Hadzapior Watindigaof TanganyikaTerritory. Africa 4:273- 86. BlurtonJones, N., 1994, A Replyto Dr. Harpending.American Journal of Physical Anthropology93:391-97. BlurtonJones, N., K. Hawkes,and P. Draper,1994, Differencesbetween Hadza and !KungChildren's Work: Original Affluence or PracticalReason? Pp. 189-215 in Issues in Hunter-GathererResearch (ed. by E.S. Burch).Oxford: Berg. BlurtonJones, N., K. Hawkes,and J.F. O'Connell,1989, Modellingand Measuring Costs of Childrenin Two ForagingSocieties. Pp. 367-90 in ComparativeSocioecology (ed. by V. Standenand R. Foley). Oxford:Blackwell. Blurton,Jones, N.G., and R.M. Sibly, 1978, Testing Adaptivenessof Culturally DeterminedBehaviour: Do BushmanWomen Maximize Their Reproductive Success by SpacingBirths Widely and Foraging Seldom? Pp. 135-58 in HumanBehaviour and Adaptation.Society for the Studyof HumanBiology Symposium 18. London:Taylor andFrancis. BlurtonJones,N., L. Smith,J.F. O'Connell,K. Hawkes,and C.L. Kamuzora,1992,

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions FORAGINGRETURNS OF !KUNG ADULTS AND CHILDREN 247 Demographyof the Hadza:An Increasingand High Density Populationof Savanna Foragers.American Journal of PhysicalAnthropology 89:149-81. Briggs,J., 1970, Never in Anger.Cambridge, Mass.: HarvardUniversity Press. Caldwell,J.C., 1982, Theoryof FertilityDecline. New Yorkand London: Academic Press. Draper,P., 1976, Socialand Economic Constraints on ChildLife amongthe !Kung. Pp. 199-217 in KalahariHunter-Gatherers (ed. by R. Lee andI. DeVore).Cambridge, Mass.: HarvardUniversity Press. Draper,P., andE. Cashdan,1988, TechnologicalChange and Child Behavior among the !Kung.Ethnology 27:339-65. Drinkwater,B.L., and S.M. Horvath,1979, Heat Toleranceand Aging.Medicine andScience in Sports11:49-55. Drinkwater,B.L., I.C. Kupprat,J.E. Denton,J.L. Crist, andS.M. Horvath,1977, Responseof PrepubertalGirls and CollegeWomen to Workin the Heat. Journalof AppliedPhysiology, Environment, Exercise Physiology 43:1046-53. Harpending,H.C., 1994, Infertilityand ForagerDemography. American Journal of PhysicalAnthropology 93:385-90. Hawkes,K., andJ.F. O'Connell,1982, AffluentHunters? Some Commentsin Light of the AlyawaraCase. AmericanAnthropologist 83:622-26. Hawkes,K., J.F. O'Connell,and N. BlurtonJones, 1989,Hardworking Hadza Grand- mothers.Pp. 341-66 in ComparativeSocioecology (ed. by V. Standenand R. Foley). Oxford:Blackwell. Hawkes,K., J.F. O'Connell,and N. BlurtonJones, 1991, HuntingIncome Patterns amongthe Hadza:Big Game, CommonGoods, ForagingGoals and the Evolutionof the HumanDiet. PhilosophicalTransactions of the RoyalSociety, London, Series B, 334:243-51. Hawkes,K., J.F. O'Connell,and N. BlurtonJones, n.d., HadzaChildren's Foraging. Unpub.ms. in authors'possession. Hiernaux,J., andD.B. Hartono,1980, PhysicalMeasurements of the AdultHadza of Tanzania.Annals of HumanBiology 7:339-46. Howell,N., 1979,Demography of the DobeArea !Kung. New York:Academic Press. Konner,M.J., 1976, MaternalCare, InfantBehavior and Developmentamong the !Kung.Pp. 218-45 in KalahariHunter-Gatherers (ed. by R.B. Lee and I. DeVore). Cambridge,Mass.: HarvardUniversity Press. Lee, R.B., 1968, WhatHunters Do for a Living,or, How to MakeOut on Scarce Resources.Pp. 30-48 in Manthe Hunter(ed. by R.B. Lee andI. DeVore).Chicago: Aldine. Lee, R.B., 1969, !KungBushman Subsistence: An Input-OutputAnalysis. Pp. 47- 79 in Environmentand CulturalBehavior (ed. by A.P. Vayda).Garden City, N.Y.: NaturalHistory Press. Lee, R.B., 1979, The !KungSan. Cambridge,Eng.: CambridgeUniversity Press. Lessels, C.M., 1991, The Evolutionof Life Histories, Pp. 32-68 in Behavioral Ecology(ed. by J.R. Krebsand N.B. Davies). Oxford:Blackwell Scientific. Obst, E., 1912, Von Mkalamains Landder Wakindiga.Mitteilungen der Geogra- phischenGeselleschaft in Hamburg26:3-45. O'Connell,J.F., K. Hawkes,and N. BlurtonJones, 1988, HadzaScavenging: Impli- cationsfor Plio-PleistoceneHominid Subsistence. Current Anthropology 29:356-63. O'Connell,J.F., K. Hawkes,and N. BlurtonJones, 1991, Distributionof Refuse- ProducingActivities at HadzaResidential Base Camps:Implications for Analysisof

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions 248 JOURNALOF ANTHROPOLOGICAL RESEARCH ArchaeologicalSite Structure.Pp. 61-66 in The Interpretationof ArchaeologicalSpatial Patterning(ed. by E.M. Krolland T.D. Price).New York:Plenum Press. O'Connell,J.F., K. Hawkes,and N. BlurtonJones, 1992,Patterns in the Distribution, Site Structureand Assemblage Composition of HadzaKill-Butchering Sites. Journalof ArchaeologicalScience 19:319-45. Sands,B., I. Maddieson,and P. Ladefoged,1993, The PhoneticStructures of Hadza. UCLAWorking Papers in Phonetics84:67-87. Shostak,M., 1981,Nisa: The Lifeand Words of a !KungWoman. Cambridge, Mass.: HarvardUniversity Press. Smith,C.C., andS.D. Fretwell,1974, The OptimalBalance between Size andNum- ber of Offspring.The AmericanNaturalist 108:499-506. Smith, E.A., and B. Winterhalder,eds., 1992, EvolutionaryEcology and Human Behavior.New York:Aldine de Gruyter. Solway,J.S., andR.B. Lee, 1990, Foragers:Genuine or Spurious?Current Anthro- pology31:109-46. Squire, D.L., 1990, Heat Illness: Fluidand ElectrolyteIssues for Pediatricand AdolescentAthletes. Pediatric Clinics of NorthAmerica 37:1085-109. Trivers,R.L., 1974, Parent-OffspringConflict. American Zoologist 14:249-64. Truswell,A.S., andJ.D.L. Hansen,1976, MedicalResearch among the !Kung.Pp. 166-94 in KalahariHunter-Gatherers (ed. by R.B. Lee and I. DeVore).Cambridge, Mass.: HarvardUniversity Press. Wilmsen,E., 1989, LandFilled with Flies: A PoliticalEconomy of the Kalahari. Chicago:University of ChicagoPress. Woodburn,J. C., 1964, The SocialOrganisation of the Hadzaof NorthTanganyika. Ph.D. diss., Universityof Cambridge,Cambridge, Eng. Woodburn,J.C., 1968a,An Introductionto HadzaEcology. Pp. 49-55 in Manthe Hunter(ed. by R.B. Lee andI. DeVore).Chicago: Aldine. Woodburn,J. C., 1968b,Stability and Flexibility in HadzaResidential Groupings. Pp. 103-10 in Manthe Hunter(ed. by R.B. Lee andI. DeVore).Chicago: Aldine. Woodburn,J.C., 1988, AfricanHunter-Gatherer Social Organisation: Is It Best Un- derstoodas a Productof Encapsulation?Pp. 31-64 in Huntersand Gatherers (ed. by T. Ingold,D. Riches,and J. C. Woodburn).New York:Berg. Yellen,J.E., 1977, ArchaeologicalApproaches to the Present: Modelsfor Recon- structingthe Past. New York:Academic Press. Yellen,J.E., and R.B. Lee, 1976, The Dobe-/Du/daEnvironment. Pp. 27-46 in KalahariHunter-Gatherers (ed. by R.B. Lee andI. DeVore).Cambridge, Mass.: Har- vardUniversity Press.

This content downloaded from 155.97.85.93 on Mon, 11 May 2015 20:24:07 UTC All use subject to JSTOR Terms and Conditions