Behavioral inferences from Early Stone artifact assemblages: an experimental model

Nicholas

Behavioral inferences from Early Stone artifact assemblages: an experimental model*

Toth

Dept. ofrlnthrojology, Indiana University, Bloomington, Indiana 47405? L’.S.d.

A methodological approach for assessing the nature and palaeographic distribution of early stone artifact assemblages is presented, modeled after an approach originally used for fauna1 analysis. By combining experimental replicativc studies with careful analysis of Palaeolithic archaeological occurrences, it is potentially possible to reconstruct entire technological systems and to assess what stages of lithic reduction may be preferentially represented at high density artifact concentrations that we normally call and what stages of reduction are found in lower archaeological “sites”, density (“off site”) scatters. Based upon the results of this approach, alternative explanations for certain sta,qes of lithic reduction being preferentially represented at a number of Plio-Pleistocene archaeological sites at Koobi Fora, Kenya are considered and evaluated with regard to early hominid organizational patterns. It appears that hominid stone technology was a relatively complex system by 1.9 to 1.4 million years B.P.. involving significant transport and carrying of stone artifacts representin? various stases of reduction and suggesting more foresight and plannmg than observed among extant nonhuman primates. The application of this approach to other Palaeolithic occurrences should enable anthropologists to obtain a better understanding of the organizational patterns of tool-making hominids throughout the course of human evolution.

Received 6 February 1987 Revision received 10 November 1987 and accepted 12 November 1987 Publication

date

May

1988

Rq~wordst Stone technology. experimental archaeology, hominid evolution. Oldowan. Ache&an, cognition. organizational skills, palaeoanthropology

Journal

of Human

Evolution (I 987)

16, 763-787

Introduction Lithic

technology

represented

is probably

the most complete

in the prehistoric

of taphonomic/diagenetic

record.

agencies

of any proto-human

behavioral

Unlike bones, which are susceptible that can profoundly

system

to a wide range

bias a fauna1 assemblage,

stone

artifacts are relatively indestructible in most prehistoric contexts (unless exposed to the elements, unburied, for long periods of time), and obviously prehistoric hominids were the principal agents lithic materials.

that collected,

Aside from hominid artifact

systems

behavior,

transported,

manufactured,

used, and discarded

these

then, the only other major forces acting upon these stone

are the effects of geological

or biological

agencies

which may affect the

spatial relationships of these materials. For example, stone artifacts may be swept downstream by floods and buried; or deflation, root action, trampling, termites and burrowing

rodents may affect the vertical dispersion

ofstone

artifacts over time. Yet, in the

early stone age, it is likely that most, if not all, of the stages of lithic reduction represented hominid

within

territorial

a given

sedimentary

basin

ranges are both confined

if the sources

should be

of raw material

and the

within that basin.

Ifone views a study area, such as Koobi Fora in northern Kenya, as a “closed system” in that raw materials for artifacts were acquired in the same sedimentary basin (and territorial range) where stone artifacts were made, used, and discarded, then a sound * Paper presented at the symposium Brrkrlev, C.4, in April 1986 in honor 0047-2484/87/07/80763

+ 25 $03.00/O

“The Longest Record: of ProfessorJ. Dcsmond

The Human Clark.

Career

in Africa”,

0

1987 Academic

held at

Press

Limited

understanding

of the lithic

technologies

should enable an investigator (“stages”) of the reduction

archaeological sites, and h\. surface represented in low-density “otf-site” clues about the organizational their cognitive abilities. This

study

occurrences

involves

combined

careful

experimental

survev

archaroloqical

into

the nature

occurrcnccs

of

early

the

trchnolo,qirs

attempt to understand the hominid assemblages found at these sites.

prehistoric

reduction

Kenya.

and 0;iv.c insights

at

sins thcsc

rcsponsiblc

into

archaeological

It is an approach

archarological

patterns

arc

y-icld important

of Plio-l’leistocc.ll~

rcprcscntcd

behavioral

ofstonc should

hominids

in northern

of

stages

observed

tool-making

East Turkana,

analysis

what

‘Ihe patterns

skills of early

Fora,

replication

at early

to ascertain

areas.

investigations

from Koobi

reprcscntcd

to ascertain by experimentation and cxcav,ation what aspects of‘ stone arc prcf‘crcntially represented or missing at given

that has

with

detailed

occurrcnccs.

in an

ti)r the stone

artifact

Background The Plio-Pleistocene Turkana, northern between

approximately

a sedimentary

basin

by the proto-Lake time

period)

Leakey,

1.9 and 1.4 million

Turkana

1978; McDougall

stream

courses

towards

indicate

flowing

stream

courses

Leakey

& Leakey,

and montanc

forest

1978; Vicens,

25 Plio-Pleistocene

have been excavated Bunn

Turkana.

Edal., 1980; Harris,

ofthc

within

and on the west during

this

1976: Lcakev

&

e! al., 1985). study

arca during

much ot

dominated

by- grassland plains disscctcd byfrom the volcanic highlands direction,

southwest

Gallery

forest

rvould

in the volcanic

havr

highlands

bcrn

characteristic

along

1975, 1978:

(Behrensmey,er,

1979, 1980).

sites from

and analysed

& Glradow.

I98 1; Brown

reconstructions

a landscape

in a generally

wcrc deposited

highlands

to a large riv,cr sy tcm prriodically 1980: Hurlord

e! al., 1980; McDougall,

the proto-Lake

Some

rcducrd

1975; Gleadokv,

and palaeoccological

Plio-Pleistoccnc

years B.P. ‘I’hcse residues

on the cast by the volcanic

(possibly

ef al..

Fora arca of‘ East horn the Koobi rcsiducs of early hominid populations

occurrcnccs behavioral

that was founded

(Curtis

Palaeogeographic the

archaeological Kenya, rcprrsent

Koobi

to date

Fora

(Bunn,

1978. 1983; Harris

between

1.9 and

1.4 million

1981. 1982, 1983; Bunn

8r Bishop.

1976; Harris

years

oi Kroll,

& Isaac,

old 1986;

1976; Isaac,

197&z and 6, 1978~ and b, 1980, 198 1a and 6, 1982, 1983, 1984; Isaac & Harris, 1978; Isaac. Harris, & Grader, 1976; Isaac, Leakey & Behrensmcycr, 1971: Kaufulu, 1983; Keeley & Toth,

1981; Kroll,

nd.;

a and b, 1987; Toth The vast majority

Kroll

& Isaac,

1984; Lrakcy.

1986). of these sites have been assigned

1970; Schick,

1986: Toth,

1982. 1985

& Schick.

to the Oldowan

Industrial

Complex

(or Mode I technology). These sites arc characterized by percussors (hammerstones) of cobbles or pebbles (using thr Wentworth scale to drfinc particle size). simple cores manufactured from cobbles, pebbles (such cores arc conventionally assigned to typological categories such as choppers, polyhedrons, discoids, core scrapers, etc.); smaller retouched pieces made on flakes or fragments (conventionally called flake scrapers); and associated debitage

(whole

flakes,

split and snapped

flakes,

and miscellaneous

fragments

(SW Figure

1). One excavated site, FxJj 63, represents an early Acheulean or Mode II technology. characterized by a range of typically Oldowan forms plus the addition of’ new c.lemcnts

765

a

0’

b

5cm

I. A range ofOldowan forms and their traditional classificatory designations. (a) Hammerstone, (b) subspheroid, (c) bifacial chopper, (d) polyhedron, (c) discoid, (r) flake scraper, (g) flake, (h) core scraper. (Parts (c)-(h) after Barbara Isaac). Drawn by J. Ogden. Figure

considered picks/handaxes,

to be

the

hallmark

or cleavers

(Clark,

of the

Acheulean

Industrial

Complex,

1969, 1970, 1974, 1975; Howell

1982). (In the early Ache&an, the distinction between is not as clear as it is in the later Acheulean.) Although notably Frank yet been successful, some researchers, 1984)) believe the site to be approximately with the Chari/Karari tuff complexes.

1.2 million

& Clark,

notably

large

1963; Isaac,

thinner handaxes and thicker picks stratigraphic correlations have not Brown (personal communication, years old, roughly

contemporaneous

766

K.

TOTH

.\t Irast three hominid forms appear to he represented B.P. at Koohi

boisri:

( 1 ) AustralopithrcuJ

vvidcnce from other sitra suygrsts that this form

approximately 1.0 million )-cars H.P. (2) Homo habiliJ, found b!, at least populations

hctcvcrn I.9 and 14 million years

Fora:

being contemporaneous

1.9 million

! ears

rstinct

goes

b)

possihl)

with some relict

with Homo erer/u at approximately

I.5 million )-cars

old.

B.P.

(3) Homo erectu, known from approximatcl>-

1.5 million years B.P. at Koobi Fora (and as

carly as 1% million years B.P. from W’cst ‘I’urkanai Although occurrences

it is not certain which hominid at Koohi

(Ho\vcll. 1978).

fbrms wcrc rcsponsihlc

Fora. it is likeI). that the more qcneralizcd

ti)r the carlp stone age genus Homo populations

wcrc more dependent

upon tlakcd stone tools than NYU.~i1~~tmlopithecu.,.and certainly

on this technological

tradition

‘l’hc Plio-Plcistocrne for a thorough ( I)

The

populations,

occurrences

early

stone

technologies

so that information

evolutionary Plio-Pleistocene hominids

wrr(

obtained

implications.

simple, so that reconstruction experiments

In

t)y

prc-anatomicall>-

their behavioral

addition,

since

the

in Koohi Fora sediments, of archaeological patterns

patterns

modern could ha\-<

fijssil

remains

of

the spatial patterning

sites could he examined:

by the archaeological

of the reduction

asscmhlagcs

of lithic materials

were quitr

and rrplicati\.c

could he quite accuratr;

(3) All of the raw materials water-worn

produced

about

fossils relative to the placement rcprcsentcd

to hr ideal

for the lollo~vi~lg reasons:

are represented

(2) The stone tcchnologirs

carr).

of’the robust australopithccincs.

from the Koohi Fora arca wrrc considered

tcchnoloLgical assessment

important

of hominid

after the extinction

clasts

(“pchhlcs”.

used h!, thr earl). hominids “cohhles”)

from the streams

basin. It was thus easy to identify weathered the stages of flaking of the raw materials (4) Only one major raw material a fairly homogcncous,

at KootJi

Fora

in the ancient

started

as

scdimcntar)

cortical areas on cores and debitage.

so that

could hc easil). assessed:

type was used for stone artifacts at all Koohi Fora sites:

medium-grained

lava (of primarily

basaltic

and trachy-andesitic

composition). Because of this. the experimental rcplicativc program could concentrate on this one rock type, and technological or functional dit‘fcrcncrs hrtwccn various rock types did

not

have

chart/chalcedony,

to

he

a ma.jor

concern

(other

and quartz were represented

raw

materials,

such

in much smaller numbers

as

ignimbritr.

at some sites):

(5) The raw materials that were used by the Plio-Pleistocene hominids are still availahlc in modern streams in the Koohi Fora region toda),. so that rcp1icatL.r and functional experimental studies could be realistically accomplished: (6) The preservation of scvrral of the Koohi Fora sites was excellent, so that information from bone modification patterns, microwear analysis. and con,joining studies could bc incorporated into this stud),; (7) The Koobi Fora archaeological sites samplrd a range of palacogcographic environments that included river channel, river floodplain, and deltaic environments as well as a colluvial deposit, so that the effects of’gcological forces could be assessed; (8) The Koohi Fora archaeological assemblages wrrr excavated under the direction

of

BEHAVIORAL

Glynn

Isaac

INFERENCES

and J. W. K. Harris

prehistoric

materials.

concern. As previously theoretically

For

noted,

this

using

reason,

modern major

an interesting

represents

a relatively

FROM EARLY

aspect

“closed

techniques,

biases

from

of the

system”

767

TOOLS

with

excellent

recovery

selective

recovery

were

Koobi

Fora

with regard

study

area

of

not a

is that

to lithic technology:

it

based

on documentation on the territorial ranges of modern hunter-gatherers as well as non-human primates, it is likely that early hominids who lived there most likely spent most of their

lives which

hominids

appear

from

local

outcrops

the sedimentary to have

channels

rather

in the volcanic

this sedimentary activities

have

“stone

basin.

of their

age visiting

If stone

on the eastern

well-rounded,

exploiting

highlands,

more

and made,

Although

their

traces

side of the lake. The tool-using

stream-worn angular

used,

ranges,

the lithic

materials

of tool-making

palaeogeographic

patterns

cobbles

blocks

lithic

foraging

and

exploited

of rock

materials

patterns, behavior

to by Glynn

within

and seasonal

by these

tool-using

(referred

and pebbles

characteristic

and discarded

territorial

understood,

left diagnostic

symptomatic

their

than

are not yet well

should

basin,

acquired

hominids which

Isaac

are

( 19816) as

cards”).

artifacts

were

sporadically

distributed

across

this paleolandscape,

a range

of

geological forces may have affected their final resting place in the prehistoric sediments. For example, artifacts in stream channels may have been swept further downstream; artifacts

along stream

disturbance, streams

and

topography

incorporated

bioturbational

action been

forces,

ash-falls could

as well as lacustrine

the floodplain,

channel could

deposits.

and soil processes,

been

deposition

by depositional

affected

basin by both

and erosion,

transgressions,

Artifacts

on

by trampling

and ultimately tilled

buried

in. Artifacts

the sedimentary

higher or other

by alluvial

distributed forces

in lake margins

by dunes,

with little

by meandering

deposited

buried

and artifacts

covered

flooding

or even dissected

have been initially

as the depositional

have

by lacustrine stages

the

systems

or by vegetation

environments

regressive

along

into

stream

or volcanic

buried

may have been buried

downstreams

between

sediments deltaic

floodplains

or swept

or suffered

in

of stream could have

erosion

during

of the lake.

Yet, taking into account some reworking of artifactual materials by these agencies of disturbance and sedimentation, as well as the fact that a given stratigraphic unit may be sampling

significant

amounts

that one can discern paleogeographic

setting

communication;

Isaac

environment with large

appears amounts

sedimentary

sytem

recovered

major

oftime, patterns

at areas & Harris,

e.g. perhaps of the original

like Koobi

Fora

1975; Frank

tens ofthousands distribution (Isaac,

Brown,

from

by survey

major

eruptions.

and excavation

Thus,

it is still likely

artifacts

communication,

during

periods

the

periodically

1987). The

of site formation, brought

we may view the technological

as reasonably

within

1980, 1981a and b, 1984, personal

personal

to have been primarily depositional of tuffaceous volcanic material

ofyears,

of stone

representative

of the ancient

into

the

materials patterns.

Methodology After an extensive literature analysing the technologies approach

used

was

directly

review oflithic studies, it was decided that a novel approach to at these early prehistoric sites could be interesting. The influenced

by Raymond

Dart’s

pioneering

analysis

of the

768

N. TOTE1

fauna1 materials study,

Dart’s

from the South methodology

(1) Estimate breccias

elements). (2) Based given

the minimum

by using

expected

if all parts

Dart

were

ofindividuals,

bones,

case,

predict

phalanges.

mandibular

the number

ribs,

ofa

vertebrae,

etc.)

represented.

frequency

of a given

any discrepancies

that while

and ulna)

phalanges)

were

argued,

were

certain

types

clement

between

significance

some

elements

(such

tended

to bc well-represented,

found

in very

due

small

to the

the predicted

tools

elements,

(such

and collecting

behavior,

approach

frequency

and actual

many

the first prehistorian

to utilize

apply

to

this

strategy

actually frequencies

patterned

making

and

scoops)

out of the

place of habitation.

mammalian

The adaptive

carcasses.

as strategy

(see White,

1952),

and

using

to assist Australopithecus

studies;

localities

Dart

daggers,

taphonomic

hominid

ribs,

discrepancies,

was primarily

and processimg

proximal

as vertrbrac,

selectively clubs,

at their

humerus,

(such

subsequent

such

early

scrapers,

them

distal

elements

These

technology

i.e. killing

influenced

other

australopithccines

as saws,

of this “oesteodontokeratic”

in its predatory

as the mandible,

numbers.

to the Makapansgat

of bone

well-represented

Dart’s

limb

in the excavated

(in Dart’s

at the site.

found

radius

1953, 1957). In that

rcprescnted

elements

number

scapula,

predicted

represented at the site. (4) Attempt to explain of elements

(Dart,

hovids

prescrvrd

minimum

p&es,

of all animals

the

of individual

commonly

this estimated

(skulls,

(3) Compare

cave of Makapansgat

number

the most

upon

element

African

was as follows:

thus

although Dart

he was not

was the first to

introduce

this

logic

to

palaeoanthropology. Although subsequent actualistic research has demonstrated that much ot’thc patterning of fauna1 element composition seen in the early South African cave deposits could bc explained and

in terms

probably

behavior,

Dart’s

methodology Fora,

technologies and curation, one important Dart’s

study

approach

of the stone approach

was attempted, element

& Cruz-Uribc, to analysing form

artifact

(Toth,

of this study,

in assessing

prehistoric

by most

nature

acquisition,

of the

the stone

was made

artifact

assemblage

was

localities early

manufacture,

1982, 1985 a and b, 1987; Kecley an attempt

materials

a

today.

at Plio-Pleistocene the

raw material

fauna1

fauna1 analysts

assemblages

to understanding

considering

use, and discard

methodology

(Klein

is still used in some

a holistic

and accumulation (Brain, 1967a and 6, 1981) 1984) rather than early hominid

destruction

effects

statistical

that

In the present Koobi

of carnivore

diagenetic

& Toth.

to employ composition

the basic

at

hominid transport 1981). As logic of

of Koobi

Fora

sites. While it is obvious that flaked stone materials arc not governed by the genetic coding and rules ofgrowth and development as arc animals, it was suspected that the manufacture of certain types of core forms would yield predictable patterns of debitage. And because stones, unlike bones, are relatively indestructible and not as susceptible to natural agencies as previously noted, it should theoretically be easier to observe the system of prehistoric Iithic technology. Ifone makes the assumption

that there is a general

of cores of debitage represented at most early stone confirmed by conjoining studies), then comparisons

causal

relationship

between

the types

age sites (which may be partially of predicted and actual debitage

BEHAVIORAL

INFERENCES

patterns may yield valuable clues regarding and geographical

distribution

The methodology (1) Identify (2)

lithic materials.

used in this study was as follows:

all core forms and retouched

Predict,

based

pieces from which flakes had been detached to “minimum number of individuals”);

counterpart

on experimental

data,

the proportions

of different

below) that had been struck from the cores and retouched

the analogous

counterpart

to “element

(3) Compare the expected from the prehistoric sites; (4) Attempt different

769

TOOLS

hominid behavior patterns affecting the nature

of prehistoric

(this would be the analogous (described

FROM EARLY

assemblage

populations

to explain any significant

discrepancies

occurrences

a range of stratigraphic

between

Member”

sites),

at Koobi positions,

no streams

the actual proportions

of

sample Fora,

seven were chosen

technological

containing

for this study,

norms, and paleogeographic

contexts. These are summarized in Table 1. For the sites dated to approximately 1.9 million years, FxJj the “Lower

excavated

from the cores present at each site.

The archaeological representing

types

composition”);

of flakes with the actual populations

types of flakes and those predicted

Out of the 25 excavated

flake

pieces (this would be

1, 3, and 10 (formerly

stone

pebbles

called

or cobbles

of an

adequately large size were nearby, and transport of at least several kilometers is postulated (these distances of raw material transport are similar to those seen at Olduvai Gorge (Davis,

1978;

Isaac

Leakey,

1971;

Potts,

& Harris, 1982,

1978;

Isaac,

Harris

& Crader,

1976; Hay,

1984a and 6) . At the sites dated

1971,

to approximately

1976; l-4-1.5

million years B.P., FxJj 50, 18 GL, and 33 (f ormerly called “Upper Member” sites) sources suitable to produce the artifacts found at most of the excavated

of raw materials occurrences In Area

may have been much closer (Harris, 103 at Koobi

environments material

without

Fora,

stone artifacts

source is estimated

curation of lithic materials (P. Williamson,

personal

for the lack of association

have

1981 1 1982,

or the hominid use ofnon-lithic communication,

1979; Toth of the bones and stones.

been

found

in lake

1983).

Since

the nearest

foundation

margin raw

away, either careful hominid

materials,

& Woods,

of the predictive

such as mollusc shells

in prep.) could account

model

experiments

All of the cores and retouched

pieces from all Plio-Pleistocene

studied to gain a general appreciation stone. This analysis considered: (a) flake (b) (c)

(Bunn,

1978).

bones

to have been several kilometers

Experimental Generic

cut-marked

sites at Koobi Fora were first

for the ranges of technological

the original form (“blank”) used for artifact manufacture fragment, or indeterminate); the flaking pattern (unifacial, bifacial, polyfacial); the extent of flaking (light, moderate, heavy); and

(d) the morphological

end-product

(using conventional

patterns

of flaking

(cobble or pebble, flake or

typological

nomenclature,

such

s.

‘1’0’1‘11

BEHAVIORAL

as unifacial

chopper,

bifacial

INFERENCES

chopper,

FROM EARLY

TOOLS

polyhedron,

core

discoid,

771 scraper,

flake scraper,

pick or handaxe). While certain

every

prehistoric

recurrent

Based

upon

this initial

all of the major

different

“reduction

early

criteria

and

outlined

ccre

modes”,

Acheulean

form

a comprehensive

forms

(original

is a unique

can be discerned

and retouched or technological

artifact

above

retouched

patterns

analysis,

replicate and

core

technological

were

experimental forms means

program

using

pattern,

different

extent

experiments

the 29 “generic”

reduction

Oldowan

and

forms

retouched

experiment whole

(averaging

all debitage

numbered

sequentially

above

excavated

these,

Koobi

cores

at each

Deviations

data,

and

that

For and

each

using core

individual

whole

flakes

all experimental

comprised

were

cores

and

providing

the

the archaeological

i.e., a model

sample, ifall

as to the types offlakes

form within

evident

differences.

63, the quarrying Since

could

no evidence

be noted

made

to produce

at the site.

then be examined

(It should pieces

stages

for expected

had occurred

flake populations

From

for the set of

was generated

core reduction

of cores/retouched

of debitage.

expected

of the Then,

expected

an assemblage.

flake population

for the observed

of FxJj

in the comparisons

were made

and the actual

sought

bifaces

conducted

the data generated

core/retouched

sites with a high number

the unifaces

analysis,

as to the entire

the expected

explanations

this approach,

mode).

collected,

assemblages

predictions each

of the flake population

and possible

included

Fara

were generated

between

size was

were

and early Acheulean

off. Subsequently,

attribute

site in the archaeological

characteritics

per

of the

and morphological

cores and retouched pieces were assigned to whichever matched most closely the technology of the specimen.

in manufacturing

predictions

screen

to

for this study.

upon the experimental

to be produced

typical

experiments

struck

to detailed

sample for this investigation, 29 generic reduction modes based

5 mm

predictions

For the seven

four

as they were

flakes were subjected

empirically-derived

for producing

was designed

combinations

offlaking,

end product) (for details, see Toth, 1982). One hundred and seventeen individual replicative modes

nonetheless

Fora technologies.

from Kaobi Fora sites. Twenty-nine to produce all of the major Oldowan

proposed,

form, flaking

specimen,

in the Koobi

that with

on flakes, these

such as

flakes

of such quarrying

is not

is exhibited

at this early Acheulean site, and it appears that either unmodified large flakes, roughed-out picks/handaxes, or finished products were brought to this locality. A complete assessment of such

technological

Although

certain

as the problem heavily nonetheless

systems,

however,

methodological

of estimating

worked

down

to

the original a small,

the sizes and shapes

from a site provide

valuable

of cortex

have

of unflaked

to include

size and shape core

pebbles

platform

(butt)

of a piece form

and

range

stages.)

(Toth,

1982)) such

some

other

partial,

has been

polyhedrons),

cores, and debitage

of raw material

be made. in the study, total,

quarrying of rock that

(e.g.

and cobbles,

the possible

estimate can probably flakes were considered

on the striking

these

arise with this approach

non-cortical

clues regarding

hominids, so that a reasonable Six different types of whole absence

would

problems

based

available

to

on presence

or

or no cortex

on the

dorsal surface. Flakes greater or equal to 2.0 cm in maximum dimension were considered (Figure 2). This system, although developed independently, is very similar to that developed by Tavoso (1972) and employed by Villa (1983) in her analysis of the Lower Palaeolithic lithic assemblages from Terra Amata, Comparisons of the predicted and actual flake

France. type populations,

as well as absolute

772

N.

TOTH

FLAKE

Figure 2. The six flake types non-cortical platforms.

numbers

used in this study.

TYPES

Types

I-III

h,avc cortical

platfbrms

of flakes, are shown in Figures 3 and 4 and in Table

ihutts),

types

IV-VI

2. As can be seen, with the

exception of site FxJj 50, the actual samples of flake numbers are less than predicted. In fact, only 79% of the cores/retouched pieces from FxJj 50 were replicated due to the atypical nature of a number of cores from that site (unifacial no experimental

analog.

If those cores had been considered,

outnumbered

the archaeological

replications).

The overall discrepancy

discoids on cobbles) predicted

that had

flakes would have

sample for this site as was the case in all the other site between

actual and expected

numbers

of flakes at

the site could be due to the fact that either some of the flakes had been winnowed away by water

action

represented

before

Examination population

final

burial,

at the prehistoric

or that

only

certain

sites. This will be discussed

of the proportions

of flake

types

I-VI

shows clearly that, again with the exception

stages

of core

reduction

were

in more detail below. in the predicted

and

actual

of FxJj 50, the archaeological

sites

have a much larger percentage of non-cortical flakes (Type VI) than predicted. (Again, if the unifacial discoids mentioned above had been included in the simulation, higher predicted

numbers

the predicted

of flake types I-III

and actual populations

would have created a greater discrepancy for this site.) The overall larger proportions

type VI at the sites than in the predictions

between of flake

suggest that the earlier stages of flaking these

cores, which would produce many cortical flakes, were not present at the site. Experiments have shown that even if larger, (discoids, core scrapers,

cortical

flakes were reduced down into other core forms

etc.), this would still produce a second generation

many of which would be cortical, and not drastically Again, this will be discussed below.

ofsmaller

affect the proportions

flakes,

of flake types.

A blind test was given to test the accuracy of this “generic” replicative approach (Toth, 1982). Steve Edwards, a Berkeley palaeobotanist and skilled flintknapper, privately selected lava cobbles from a large sample of diverse sizes and shapes for flaking, and submitted only the flaked cores and retouched pieces (a sample size of 30 specimens), and none of the debitage. From only this sample, predictions of what flake types should have been represented were generated, and then compared to the actual population offlakes that Edwards produced. Considering the fact that the types of cores and retouched pieces were not necessarily typical of those produced by Koobi Fora hominids, and had to be assigned to one of my 29 reduction modes with greater dilliculty (in fact, one new mode had to be

BEHAVIORAL

INFERENCES

FLAKE

%

FROM EARLY

773

TOOLS

PREDICTIONS

100

50

01 FxJj

1

SITE

FxJj REPL

3

SITE

REPL.

% 100-j

50 1

FxJj

10

SITE

REPL

96 100

50

n

n

0,l FxJj

FxJi

50

SITE

REPL

1SGL

SITE

REPL.

% 100

50

0 I FxJi SITE

33

FxJj REPL

SITE

63 REPL

Figure 3. The breakdown of the flake types from the archaeological sites (greater or equal to two cm) compared to the expected percentages based on simulations from the replicative experiments. Note that type VI (non-cortical) flakes occur in greater than expected numbers at most archaeological sites.

774

N. TOTH

FLAKES % 100

cl n

PREDICTED

ACTUAL

50

i

0 1

FxJj

Table 2.

FxJ

1

j13

FxJj

10

FxJjSO

FxJjlElGL

FxJj33

FxJi63

Comparisons of flake types (greater or equal to two centimeters in maximum dimension) between the archaeological sites and predictions based on simulating flake populations from the replicative experimental modes. Proportions (row) are shown in parentheses

F.XJ] I 0 i(l4JO) 2 (0~0-1)

Site

Simulation

I fi-.t~lJ 7 Site

Simulation FxJ;

0 (0~00)

0 [lW1)(1) ” ( 04N31

I (047)

IO

t (04Jll ,i (rl.rli)

Site

Simulation I;slJ 50 Site Simulation L:Th IRGI.

Site Simulation E-x.6 33 Site Simulation

0 (0.001 39(04%I

FTJ] %Y Sitr Simulation

? (041) 3 (041 I

defined), reasonably

nonetheless similar

comparisons

of the predicted

(SW Fig-m-r 5 and Table

and actual

flake populations

w-r<

3).

Because of the evidence that Koobi Fora site FxJj 50 \vas rclativcl) undisturbed h) geological forces (Bunn et al., 1980; Kroll & Isaac, 1984: Schick, 1986), plus thr fact that this site had a relatively high percentage of conjoining piccrs that had been rrfittrd h)

BEHAVIORAL

INFERENCES

FROM

EARLY

TYPES

775

TOOLS

NUMBERS

% 100

q TOTH n EDWARDS II Ill IV

I

50

v

VI

..:. ic dlL

01

.:i: cdl

EDWARDS

TOTH

Figure 5. Bar graphs showing the results of the blind test. Predicted (Toth) are compared to the actual test sample (Edwards).

Table 3.

and flake type percentages

The results of the blind test. Predicted numbers and flake type percentages are compared the actual test sample. Note the overestimation of total numbers and of flake Type VI

Predicted Actual

I

II

III

IV

V

VI

Total

9 (0.04) 7 (0.04)

18 (0.08) 15 (0.08)

40 (0.17) 26 (0.15)

20 (0.08) 17 (0.09)

97 (0.41) 93 (0.52)

55 (0.23) 21 (0.12)

239 179

project members,

it was decided that a core-by-core

more precise information associated Toth,

numbers

debitage

about the relationships

than was generated

replication

of this site could yield even

between the cores/retouched

by the generic

with

experiments

(Bunn

pieces and et al., 1980;

1982).

46 cores and retouched pieces from the excavated site were replicated in this study. Only lava artifacts (and not ignimbrite, chert, or quartz artifacts) were considered. The methodology

was as follows:

(1) All cores and retouched form and patterning (2)

Based

on this

cobble/pebble (3) A similar flaking

lava cobble/pebble

excavated

In practice,

analysis,

analysed

to reconstruct

the approximate form”)

their original

sequencing

size and

shape

of the

was estimated;

or flake to that estimated

by a blow-by-blow

to the prehistoric

(4) The population population

technological

or flake used as a blank (“original

and reduced

identical

pieces were carefully

of flake removal;

in stage 2 was selected

of flake removals

for

that was almost

specimens;

of flakes generated

by the experiments

from the archaeological

it was more difftcult

was compared

to the flake

occurrence.

than anticipated

to produce

acceptable,

near-exact

facsimiles to the excavated cores and retouched pieces, because so much of the end product of the prehistoric Oldowan cores appeared to be due to nuances in the raw material as well as chance during flaking. An average of approximately four trials was required for a given core/retouched piece to achieve a reasonable facsimile, and up to fifteen trials. Only the flakes from the acceptable core replicas were considered in this analysis. Figure 6 and Table 4 shows the comparison of the expected and actual (prehistoric) flake samples. Interestingly, although the total number of flakes is similar in both cases, there

776

N.

TOTH

FxJj

50

SITE

REPL.

Figure 6. Comparison of flakr types from site Fx,lj 50 and the rcplicatrd sample. Kotr the highrr percentages of types IV-VI from the archaeological sitrs. For flakr types 1-1’1. greatrr or rqual to two cm.

Table 4.

Comparison of the flakes produced during the replication of FxJj 50 lava cores and the actual sample of lava flakes from site FxJj 50. Only flakes greater or equal to two centimeters in maximum dimension are considered

Predicted Actual

I

II

III

I\!

v

\‘I

Total

35 (0.12) 19 (0.07)

125 (9.44) 62 (0.24)

41 (0.14) 35 (0.14)

13 (W1J51 18 (OTJi)

52 (0.18) 77 (0.30)

17 (0.06) 44 (0.17)

283 255

are some interesting technological differences hot observed in the generic experiments. In the experiments, more flakes with cortical platforms (representing the unifacial flaking of cobbles) with

were

produced;

non-cortical

stages

of cobble

conjoining

sets

represented

reduction from

within

It should representative

this

were may site

present, have

suggests

the excavated

areas

which

been

more

that

proportionately

suggest

that

preferentially

only

(Schick,

hand,

would parts

1986; Toth,

frequency distributions (expected frequencies). results and the blind

validity two-sample

tests

were

carried

out

more

the

later,

represented

of entire

flakes bifacial

there.

The

reductions

are

in being from Koobi

more Fora.

core

1982, 1985).

be pointed out, however, that this site is exceptional of all of the stages of flaking than any other site sampled

Assessing experimental Kolmogorov-Smirnov

rejected

at site FxJj 50, on the other

platforms

to compare

the

cumulative

of flake types at the sites with those derived from experiments No significant difference was found between the experimental test at the 0.05 level of significance (the null hypothesis could be

only at the 0.20 level).

Among

the seven

sites sampled,

five exhibited

significant

differences at the 0.05 level or better: FxJj 1, FxJj 3, FxJj 50, FxJj 18GL, and FxJj 63. Only at FxJj 33 (with an extremely small site sample size of 6 flakes) and FxJj 10 were no significant differences found between the site and the cumulative frequency distributions predicted discussed

from experiments below.

(Table

5). Explanations

for the observed

discrepancies

will bc

BEHAVIORAL

INFERENCES

777

TOOLS

Results of Kolmogorov-Smimov two-sample tests for cumulative frequency differences: site flake type samples compared to the experimental results (expected), and comparison of frequency distributions in the blind test (results versus predictions). (The null hypothesis of no significant difference was tested at the O-05level of significance)

Table 5.

Site FxJj FxJj FxJj FxJj FxJj FxJj FxJj

Pvalue

1 3 10 50 18GL 33 63

Blind test

Special considerations Behavioral

FROM EARLY

vs. geological

n(Obs/exp)

Reject H,, at 0.05 level?

0.20
23151 13/21 69196 3531331 513/1692 61692 199/357

Yes Yes No Yes Yes NO Yes

-Co.20

1791239

No

of the predictive

model

effects on assemblage composition

With the data that I had recorded from my generic replicative experiments, it was possible to make computer simulations of what a number of different behavioral or geological agencies would do to Oldowan flake populations. These included: (1) If all the stages of flaking were represented at a prehistoric site, or whether certain stage were preferentially represented (simulated by excluding (a) the first half, and (b) the first three-quarters of the flake population; (2) If larger, more useful flakes tended to be transported away from prehistoric sites (simulated by excluding flakes greater than 4 cm in maximum dimension from the flake population); (3) If water action selectively winnowed the smaller size fractions of flakes away (simulated by excluding flake (a) less than 2 cm and (b) less than 4 cm from the flake population). The effects of agencies (1) and (3) are shown in Figure 7 and Table 6. Water action selectively winnowing away the smaller, lighter flakes, would tend to preferentially leave behind larger cortical flakes (especially types II and V). And, if preferentially later stages of flaking were represented at prehistoric sites, non-cortical flakes (type VI) would tend to represent a higher proportion of the flake population. Selectively removing larger, more easily utilized flakes for future use elsewhere (agency 2) would also tend to leave more smaller, non-cortical (type VI) flakes (the inverse pattern of what would be produced by water action). Technological

considerations:

East African Acheulean assemblages

In East and Central Africa, the vast majority of Acheulean handaxes, picks, and cleavers were made on large flakes struck from massive lava cores (Clark, 1969, 1970, 1972, 1975, 1980; Isaac, 1977, 1982). For the thousands of these large bifaces which have been recovered from most of these sedimentary basins, a remarkably small number of the boulder cores are known (Isaac, 1977, 1982).

778

N.

%

OLDOWAN

100

ALL

52cm.

54cm.

I II Ill IV v VI

50

0 1

TOTH

did &Lbd

ALL

% 100

50

0 i

% 100

50

0 1

There are palaeogeographic reasons to explain this discrepancy: first ofail, lava boulders large enough to produce suitable flakes tend to he found near volcanic outcrops, either as primary sources (at the lava flows) or at secondary

sources (such as alluvial fans or in the

upper reaches of streams) which tend to be erosional environments. of suitable boulders tend to be in volcanic highlands, whereas

Most of these sources most of the hifacial

implements

basin.

are found along stream courses within the sedimentary

BEHAVIORAL

INFERENCES

FROM EARLY

779

TOOLS

Flake types for “Oldowan” forms by stage of reduction and increasing size fraction, to simulate curation of cores/retouched pieces and water action. These computer simulations are based on the sum of the experiments from the different reduction modes

Table 6.

Flake sizes

stage of All

reduction All Type I Type I I Type III Type IV Type V Type VI Last half

32

49 (0.03) 146 (0.09) 154 (0.09) 125 (0.07) 567 (0.34) 63.5 (0.38)

It seems were

likely

either

exfoliation,

many

of the large

on the surface

and fragmentation.

also been erosional basin margin as susceptible times. Thus,

that

exposed

Others,

environments,

Acheulean

for long

cores

periods

19 (0.04) 96 (O-2 I) 21 (0.05) 34 (0.08) 209 (0.46) 71 (O-16)

5 (O+Jl) 49 (0.10) 20 (0.04) 24 (0.05) 232 (0.45) 186 (0.36)

l(O.01) 31 (0.17) 7 (0.04) 6 (0.03) 96 (0.53) 42 (0.23)

0 (0.00) 17 (0.07) I1 (0.04) 7 (0.03) 110 (0.43) 112 (0.44)

10 (O-14) 6 (0.08) 0 (0.00) 37 (0.51) 20 (0.27)

that produced

of time

in the upper

(0.04) (0.12) (0.07) (0.08) (0.40) (0.28)

46 140 84 95 463 325

Type I Type II Type III Type IV Type v Type VI Last quarter Type I Type II Type III Type IV Type V Type VI

reaches

34 cm

cm

to have

flakes for bifaces caused

to stream

and even if these cores were buried

0 (0.00)

weathering,

systems, in sediments

may have near the

(the interface of sediments with the surrounding bedrock), they would not be to erosion and exposure (and subsequent archaeological discovery) in recent

most Acheulean

sites yield bifaces,

smaller

cores,

and debitage,

with an absence

of

larger cores. It is likely that since the quarrying of the large flakes and the rough-out stages of biface manufacture were within proximity ofvolcanic highlands, the entire technological system

would

not necessarily

be represented

in the sedimentary

basin.

Functional considerations The stages

approach

outlined

at archaeological

in this article occurrences;

deals

primarily

it concerns

with

the lithic

the nature

of lithic

reduction

technological

system

of artifact

production, and does not, in this first instance, consider tool function. But clearly, the overall pattern of stone artifact distributions found in prehistoric sediments would have been strongly influenced by the activities for which stone tools were used, the types of artifacts chosen for activities, and the locations of those activities on the palaeolandscape. Determining the functions of stone artifacts has been a problem that has plagued prehistorians for over a century. Attempts to identify functions of early stone artifacts have included enthnographic analogs, feasibility experiments, contextual associations of artifacts and other features, modification to animal bones or other organic materials,

N. TOTH

780

edge-damage and microwear polish patterns, and recently, attempting to identify adhering organic matter (Loy, 1983). The problem becomes even more difficult for simple technologies templates”

such as the Oldowan,

since the identification

of “end products”

or “mental

is difficult.

Nonetheless,

some progress

has been made

to this end. Perhaps

evidence has come in the form of bone modification 1982; Bunn & Kroll,

1986; Potts, 1982, 1983, 1984a and b; Potts and Shipman,

suggest that skinning, sharp-edged

the most dramatic

patterns at Oldowan sites (Bunn,

dismembering,

stone artifacts,

and meat processing

and bone-breaking

1981) which

were being carried

with hammerstones.

198 1,

Microwear

out with analysis of

a small number of Oldowan artifacts from Koobi Fora (Keeley & Toth, 1981) has also suggested a range of activities by 1.5 million years, including animal butchery. woodworking, and cutting of soft vegetable matter. For early stone age sites, it seems likely that the location artifacts

was primarily

(1) Discard

of lithic artifactual

material

(2) Discard of lithic artifactual material possible dulling/wear) of the tool;

In more recent

of stone

as waste, at the location

materials

flaking;

that had been transported/curated

for caching purposes,

technological

ofstone

at the location of tool use, after utilization

(3) Discard of the lithic artifactual material use, but for some reason never employed; (4) Discard oflithic 1983, 1984a and b).

of abandonment

due to one or more possibilities:

systems,

with intent oflater

the ultimate

for later

use (Potts,

discard of artifacts

(and

1982,

could have also

been due to a range of causes that could be related to more complex cultural and technological systems: ritual grave goods or votive caches, areas in which worn hafted tools were replaced

al., 1979), behavioral

by fresh tools (which might not be near the location of actual use (Cahen et

in special patterns,

waste

middens,

or at locations

however, are probably

of commerce

not major considerations

and trade.

These

in the early stone age.

Considerations of “landscape archaeology” or the “scatter between the patches ” As has been previously mentioned, it is likely that most of the technological

elements

stone

within

artifact

Plio-Pleistocene

manufacture sediments

by Koobi

Fora

hominids

have been preserved

and that various stages offlakingofcobbles

be easy to discern within given archaeological

deposits.

of the

and pebbles should

If we could somehow observe the

total distribution of artifacts over the paleolandscape at any given stratigraphic horizon, and see the spatial distribution of the various stages of core reduction, we could develop a much

better

extraction

sense of a number

and artifact

of hominid

behavior

patterns

involving

raw material

transport.

A decade ago, Isaac & Harris ( 1978) b e g an a study at Koobi Fora dubbed “The Scatter Between the Patches” to document the distribution and nature of lithic materials on erosional outcrops, and this work has more recently been taken up by Nicola Stern at Koobi Fora. The mapping information on the geographic

of such artifact distributions should provide valuable distribution of artifacts during the Plio-Pleistocene in the

Koobi Fora area. Obviously the area sampled by this approach is limited to the erosional faces sedimentary outcrops, and it is thus by no means a complete sample: vast quantities

of of

BEHAVIORAL

Plio-Pleistcene recent

sediments

streams

sediments

(and associated

and erosion

scarps,

FROM EARLY

artifacts

781

TOOLS

and fossils) have been eroded away by

while perhaps

an equally

large volume

of ancient

lies buried under many meters of more recent deposits. Also, since the materials

studied represent the contextual detailed

INFERENCES

ex sitzl artifacts association

technological

“landscape

(as is the case with most fossil hominids

of these

surface

artifacts

is rarely

in East Africa),

proof-perfect.

However,

analysis

of the materials found in such surveys (sometimes termed could provide a valuable window by or “off-site” archaeology),

archaeology”

which to view the overall stone technological

system of early hominids.

&ration versus expedience Lewis Binford

(1979) has proposed a bipartite

system for classifying

hominid technological

behavior based on the amount of foresight and planning that was involved in the organization of activities. An “expedient” technology would be one that was an immediate response

to a given need; Binford

expedient

technology.

prepared

in advance

conceivably,

gives the example

A curated

technology,

of a predicted

caching)

of prepared

need,

materials

of gathering

in Binford’s

firewood

terminology,

and normally

involves

as a typical

is one that

more

transport

is (or

for later use.

Although Binford has recently stated (1987) that these terms were devised for and should be restricted to anatomically and cognitively modern humans, and not for earlier hominid forms, it is nonetheless continuum

Koobi Fora sites (Toth, “curation” modern

very interesting

early stone age organizational

1982, 1985), the organizational

would appear chimpanzees

Kortlandt,

1986;

to attempt

to have been habitually

(Boesch

McBeath

&

to determine

where on such a

skills lay. As has been suggested in studies of the

Boesch,

& McGrew,

skills involved in tool transport

more complex

than is observed

and

among

1981, 1984; Hannah & McGrew, 1982; McGrew, 1974, 1977; Nishida,

1987; 1973,

Nishida & Hiraiwa, 1982; Sugiyama & Koman, 1979). Oldowan and early Acheulean transport distances were often overall several kilometers, normally outside the range observed

in modern

hominids

did, however, appear to be on a much more modest scale than is observed among

chimps.

modern hunter-gatherers DeVore,

1976; Yellen,

As other

researchers

Transport

(Gould,

distances

of lithic materials

1968, 1969, 1980; Hayden,

for Plio-Pleistocene

1977, 1979; Lee, 1976; Lee &

1977). have

suggested,

differences

in planning

and transport

continuum

model rather than a bipartite

it would

behaviors

seem

to be more

within technological

model. For example,

realistic

to view

systems in terms of a

the organizational

skills for

gathering firewood would be profoundly influenced by the abundance and distribution of suitable fuel on the landscape. In an area where dead wood was abundant and relatively evenly distributed, and distributed

have been required There are organizational (1) Distance

little planning

would be required.

in localized patches, to efficiently

several variables skills as exhibited of transport

In areas where firewood was scarce

a great deal more planning

and organization

which are potentially quantifiable by lithic technology. These include:

from the original

with

regard

to

source of raw material;

(2) An estimated minimum time between artifact manufacture and use; (3) Analysis of the distribution of various elements of the lithic technology landscape.

may be

exploit that resource.

on a

There are, of course, a number of serious methodological problems to be encountered in attempting to compare ethnographic or ethological data with prehistoric patterns. Not the least of these is the fact that, whereas modern actualistic

studies normally record transport

of materials

over short intervals of time (a few years at most), the patterns

exhibited

prehistoric

record

in areas whcrc

deposition

could rcprescnt

very long periods

of time (especially

is slow) and there is a grcatcr chance of‘lithic materials

re-used. Thus, transport the cumulative

distances

bring retransportcd

evident in thr early archaeological

total of scvcral separate

in thr and

rrcord may represent

rvcnts.

In more recent prehistoric periods, as well, the distancrs oftransport could bc more from trade or reciprocity than individual trips, so that distances from the geological sources ma) be a function

of much wider spheres of influence

rather than individual

fi)raging events.

Conclusions It is hoped that this exposition palaeolithic

occurrences

informative materials

when lithic tcchnolo,gies and primary

most or all stages archaeological provide

will show the value of using a similar approach

throughout or secondary

of reduction

deposits.

important

Such

insight

the

world.

Admittcdll.,

arc relatively

this

at other

approach

is most

simple, when the original forms of raM

sources of acquisition

can bc ascertained,

and when

arc represented within the overall catchmcnt area of’ predictive modeling of lithic technological systems can

into

the

nature

and

distribution

of Palaeolithic

artifact

occurrences, large or small. and help us undrrstand them in terms of hominid behavior patterns. By using experimentation to gain a brttcr understanding of the cntirc lithic reduction

system,

archaeological

it is possible

occurrences

the distribution

to identify

and. iflarge-scale

and preferential

various

stages of lithic reduction

sedimentary

location ofstagcs

oflithic

at specific

exposures permit, to investigate reduction

in a palaeogcographic

context. Some

of

the

mentally-derived

major

behavioral

infercmccs

model to the lithic technologies

(1) Early hominids source. The transport

transported observed

materials

from

Fora arc:

applying

up to sc\.cral kilomctrrs

both at Koobi

outside the range of modern chimpanzees,

dcrivrd at Koobi

Fora and at Olduvai

who rarcl)

transport

this

cxperi-

fi-om their geological Gorge appears

materials

to bc

more than a f&

hundred meters, though it is much less than distances normally obscr\.cd among hunter-gatherer populations. This suggests more complex organizational skills requiring more planning

and foresight

(2) Early hominids

collectrd

than that obser\,ed among extant apes. and transported

focal points on the palaeo-landscapes; into the thousands. hammers observed

lithic (and prcsumabl!,

these concentrations

faunal) remains at

of lithic materials

produced at some modern or recent hunter-gathrrcr sites. (3) Rarely art= all stages of lithic reduction rcprcscnted at cxcavatcd usually,

ran number

These concentrations are very diffrrrnt than the small clusters ofstonc amona some modern chimpanzcrs, but more similar to thr quantities

later stages of corr reduction

are prrfercntiall)

reprcscntcd.

prrhistoric

as observed

sites: by the

high frequency of non-cortical flakes. This pattrrn is quite different. fi)r example. than the Lower Palaeolithic site of Terra Amata in Francr (\‘illa, 1983), where earlier stages of beach cobble reduction were very well representrd: (4) The majorit_v of stone artifacts distributed on palace-landscapes

at Koohi

Fora,

BEHAVIORAL

exhibited

along

palaeolithic

erosional

“sites”,

reduction

appear

reduction

experiments.

in small

There are a number of alternative

(a) Earlier

clusters

at the Koobi

at Koobi

flaked

in dense

concentrations

or as isolated

Fora

or typical

specimens

sites that are expected

reasons why the earlier stages offlaking Fora sites. These

stages of flaking were conducted

such as in stream channels of partially

783

TOOLS

(G.

Isaac,

1985). With the exception ofFxJj 50 (which I suspect was within source), it is curious that fewer flakes and later stages of core

to be represented

be fully represented

FROM EARLY

are not found

outcrops,

but rather

personal communication, sight of a raw materials

INFERENCES

from lithic

do not tend to

include:

away from the dense concentration

sites,

or areas outside the alluvial valleys. This would imply transport

cores

to

the

site,

with

non-cortical

flakes

being

preferentially

represented; (b) Earlier stages offlaking were selectively removed from the site by hominids. more readily utilized flakes were removed, they would tend to be cortical flakes; (c) Lightly or moderately

flakes cores (such as “choppers”)

finished tools, whereas more extensively were

the

principal

non-cortical

sources

were carried into site areas as

flaked cores such as “polyhedrons”

of useable

flakes,

and

Iflarger,

producing

and “discoids”

higher

proportions

of

flakes;

(d) Very heavily flaked cores were preferentially cores were not considered been predicted the site;

in the flake simulations

in such a situation).

(e) Hydrological and re-concentrated

carried away from the site, so that these (more non-cortical

This would imply transport

flakes would have

of selected cores awa?, from

forces either preferentially winnowed away lighter non-cortical flakes them downstream (although the presence of heavy cores at these sites

does not suggest such reconcentrations). It should be noted that with the exception transport

and curation

of lithic materials,

With the possible exception

of(e),

each of these hypothesis

I think it likely that a combination

(as well as others that were not considered) sites found at Koobi Fora. To adequately the technological

of(e),

either to or from the excavated

project some sites.

of these hypotheses

could have led to the types of Plio-Pleistocene

solve this puzzle, it will be necessary

to document

nature of stone artifacts awayfrom as well as at these dense concentrations

that archeologists

tend to devote most of their attention

to. It is predicted

that the earlier,

cortical flakes from core reduction will be well-represented in channel gravel systems or at “off-site” low density scatters. Current research by Stern at Koobi Fora should shed light on this problem. As Harris (1978)

has pointed out, most Upper Member

Koobi

Fora sites appear to be

located within close proximity to a stream bed, usually less than 150 meters away. Since these stream beds would have been the principal source of raw material acquisition, it seems, at first glance, probable probable

transport activity

(1) Although

unusual that dramatic of partially

flaked

areas. Two factors

evidence of later stages of core reduction

material

would

should be considered

many of these prehistoric

be preferentially

discarded

and at

in this regard:

sites may have been within sight of a stream

system, it is quite possible that exposed, accessable gravel bars along these streams were quite localized, and could conceivably be some distance from the sites themselves; in that

784

N.

case proximity

to raw material

would

(Schick,

suggest

(2) That

systems

occupation,

deposited

archaeological

sites

relationships

be much

finally

stream

exactly

research

some possibly

aggrading

before

between

are not necessarily Future

could

further

artifact

occurrences should

they moved and burying the sites:

and the cultural

formation

tool-makers, allow

by both

and

and curation.

be in part

the movements and

palaeoanthropologists

organizational

excavation

valuable

Comparing

sources

but as meandering

much closer to these this would suggest that ofthe

archaeological

sites

contemporaneous.

will hopefully

yield

evidence

away from stream

topography,

sediments,

flooding

systems

distance

on higher

directed

towards

relationships bewteen spatial distributions ofstone artifacts at Koobi Fora and other early Stone Age sites, focusing detected

than the prehistoric

communication):

some of these sites may have been a greater

at the time ofhominid stream

sources

personal

TOTH

skills which

and into

of lithic

raw material these

to document profound

the

survey.

insights contrasting

have

surface

more

land-us?

major

The

scrutiny

patterns

patterns

or artifacts

patterns

of the

in

that

of early

by hominid

through

changes

evolutionary

careful

and their technological nature on both high and low density

space

will be hominid

transport

and time should

hominid

behavior

and

implications.

Acknowledgements The Koobi Fora Research Project was National Museums of Kenya, and the

directed b) Richard Leake),. Director of thr late Glynn Isaac, then at the University of

California, Berkeley. The archaeological programe at Koobi Fora was directed Isaac and Jack Harris, University of Wisconsin at Milwaukee. Their invitation participate research

in the archaeological

The

and

and

their

continuous

support

made

this

possible.

Additional Schick,

program

by Glynn for me to

advice

and assistance

and Stanley

Ambrose.

Koobi

Research

Fora

was given

Project

was

by J. Desmond

funded

h>, grants

Clark, from

Barbara

Isaac,

the National

Kath) Science

Foundation.

References Behrensmeyer, A. K. i 1975). The taphonomy and palacoecoluq~ oI‘Plio-Plristoc~nr wrtrhratr assemhlaqrs cast of Lake Rudolf. Kenva. Bull. Mu.c. (:omfi. Zoo.. Haroard (~niwrcztr 146, 473-578. Behrensmeyer. A. K. (1978). The habitat of Plio-Pleistoccnc~ hominids in East Africa: taphonomic and microstratigraphic evidence. In (C. ,Jolly, Ed.) ~S~hr!sHominidr o$.!fiira, pp. 16.5~90. London: Duckworth. Binford, L. R. (1979). Organization and formation processes: lookinq at curated technologies.,]. nnthropol. Res. 35, 225-273. Binford, L. R. (1987). Discussant at Societ>- for American Arrharulogy meetings. Toronto. Boesch, C. & Borsch. H. (1981). Sex differences in the use of natural hammers hv wild chimpanzees: a preliminary report. J. hum. Eml. 10, 585-593. Boesch, C. & Boesch, H. (1984). Mental map in wild chimpanzees: and analysis of hammer transports for nut cracking. Ptimates 25, 16&l 70. Brain, C. K. (19670). Bone weathering and the problem of how pseudo-tool. .S Afr. ,/. .STZ.63, <)7-W Brain, C. K. (19676). Hottentot food remains and their hearuq on the interpretation ofIbsail honr assemhlaqes.

Scienti’c Papers

of the Kamih Desert Research Station 32.

Brain, C. K. (1981). The Hunterr 01 the Hunted? .4n Introdurtion to .4frican Chre 7hphonom_y. (:hicago: Iinivcrsitv r~l Chicago Press. Plio-Plristocenr chronology fi)r thf Brown, F. H., McDougall. I.. Davies. T. Bi Maier, R. (1985). A II integrated Turkana basin. In (E. D&on. Ed.) Ancestorr: The Hard EGdence. pp. 82-90. New York: Alan R. Liss.

BEHAVIORAL

INFERENCES

FROM

EARLY

TOOLS

785

Bunn, H. T. (1981). Archaeological evidence for meat-eating by Plio-Pleistocene hominids from Koobi Fora and Olduvai Gorge. Nature 291, 574-577. Bunn, H. T. (1982). Meat-Eating and Human Evolution: Studies on the Diet and Subsistence Patterns of Plio-Pleistocene Hominids in East Africa. Ph.D. Dissertation, University of California, Berkeley. Bunn, H. T. (1983). Evidence on the diet and subsistence patterns of Plio-Pleistocene hominids at Koobi Fora, Kenya, And Olduvai Gorge, Tanzania. In (J. Clutton-Brock & C. Grigson, Eds) Animals and Archaeology: Hunters and Their Prey, pp. 21-30. Oxford: British Archaeological Reports (BAR). Bunn, H. T., Harris, J. W. K., Isaac, G. L., Kafulu, Z., Kroll, E., Schick, K., Toth, N. & Behrensmeyer, A. K. (1980). FxJj50: An early Pleistocene site in northern Kenya. Wld Archaeol. 12, 109-136. Bunn, H. T. & Kroll, E. M. (1986). Systematic butchery by Plio-Pleistocene hominids at Olduvai Gorge, Tanzania. Current Anthropology 27(5), 43 l-452. Cahen, D., Keeley, L. H. & Van Noten, F. L. (1979). Stone tools, toolkits, and human behavior in prehistory. Current Anthropology

20, 661-683.

Clark, J. D. (1969). Kalambo Falls Prehistoric Site, Vol. 1. Cambridge: Cambridge University Press. Clark, J. D. (1970). The Prehistory of Aff.ica. New York: Praeger. Clark, J. D. (1972). Palaeolithic butchery practices. In (P. J. Ucko & R. Tringham, Eds) Man, Settlement, and Urbnnism, pp. 149-l 56. London: Duckworth. Clark, J.D (1974). The Kalambo Prehistoric Site, Vol. 2. Cambridge: Cambridge University Press. Clark, J: D. (1975). A comparison of the late Acheulean industries of Africaand the Middle East. In (K. W. Butzer & G. L. Isaac, Eds) After the Austrulofiithecines. 605-660. The Haeue: Mouton. Clark, J. D. (1980). Raw materra and African lithic technology. Man andYEnvironment 4, 44-55. Curtis, G. H., Drake, K. L., Cerling, T. E. & Hampel, J. H. (1975). Age of KBS Tuff in Koobi Fora Formation East Rudolf, Kenya. Nature 258, 395-398. Dart, R. A. (1953). The predatory transition from ape to man. Znt. Anthropol. Linguistic Reu. 1, 201-218. Dart, R. A. (1957). The osteodontokeratic culture of Australopithecus Prometheus. Transvaal Museum Memoirs IO. Davis, D. D. (1978). Lithic assemblage variability in relationship to early hominid subsistence strategies at Olduvai Gorge. In (D. D. Davis, Ed.) Lithics and Subsistence: The Analysis ofStone Tool else in Prehistoric Economies, pp. 35-86. Nashville: Vanderbilt University Publications in Anthropology, No. 20. Gleadow, A. J. W. (1980). Fission-track age of the KBS Tuffand associated hominid remains in northern Kenya. Nature 284, 225-230. Gould, R. A. (1968). Living archaeology: the Ngatatjara of Western Australia. Southwestern J. Archneol. 24, 101-122.

R. A. (1969). Yintam: Foragers ofthe Australian Desert. New York: Scribner. R. A. (1980). Living Archaeology. Cambridge: Cambridge University Press. Hannah, A. C. & MeGrew, W. C. (1987). Chimpanzees using stones to crack open oil palm nuts in Liberia. Primates 28(l), 31-46. Harris, J. W. K. (1978). The Karari Industry: Its Place in East African Prehistory. Ph.D. Dissertation, University of California, Berkeley. Harris, J. W. K. (1983). Cultural beginnings: Plio-Pleistocene archaeological occurrences from the Afar, Ethiopia. A&can Archaeological Review 1, 3-31. Harris, J. W. K. & Bishop, W. W. (1976). Sites and assemblages from the early Pleistocene beds of Karari and Chesowanja. Union Intemationales des Sciences F’rehistoriques et Protohistoriques, Congrez IXe, Colloque V, Nice, pp. 7%117. Harris, J. W. K. & Isaac, G. L. (1976). The Karari Industry: early Pleistocene archaeological evidence from the terrain east of Lake Turkana, Kenya. Nature 263, 738-740. Hay, R. L. (1971). Geological background of Beds I and II: stratigraphic summary. In (M. Leakey, author) Olduoai Groge: Excavations in Beds I &3 II, 1!%&1%3, pp. 918. Cambridge: Cambridge University Press. Hay, R. L. (1976). Geology of Olduoai Gorge. Berkeley: University of California Press. Hayden, B. (1977). Stone tool functions of the Western Desert. In (R. S. V. Wright, Ed.) Stone Tools a~ Cultural Markers, pp. 178188. Canberra: Australian Institute of Aboriginal Studies. Hayden, B. (1979). Palaeolithic Reflections: Lithic Technology and Ethnographic Excavation Among the Australian Aborigines. Canberra: Australian Institute of Aboriginal Studies. Howell, F. C. (1978). Hominidae. In (V. J. Maglio & H. B. S. Cooke, Eds) Evolution of African Mammals, pp. 154-248. Cambridge, Mass.: Harvard University Press. Howell, F. C. & Clark, J. D. (1963). Acheulean hunter-gatherers ofSub-Saharan Africa. In (F. C. Howell & F. Bourliere, Eds) African Ecology and Human Evolution, pp. 45S533. Chicago: University of Chicago Press. Hurford, A. J. & Gladow, A. J. W. (1976). Fission-track dating of the pumice from the KBS Tuff, East Rudolf, Kenya. Nature 263, 73&740. Isaac, G. L. (1969). Studies of Early Culture in East Africa. World Archaeof. 1, 31-57. Isaac, G. L. (19760). The activities of early hominids: a review of archaeological evidence from the time span two and a half to one million years ago. In (G. L. Isaac & E. R. McCown, Eds) Human Origin: Louis Lenkey and the East Ajicun Evidence, pp. 483-518. Menlo Park: W. A. Benjamin. Gould, Gould,

786

N.

TOTH

Isaac, G.

I,. (197661. Plio-Plristoccnr arrifact assrmhlagcs Iix~m I‘:abt Rudolf. Krnla. In (Y. (:opprns. F. C:. in the Lake RudolfBasin: .Stratierabhr. Howell, G. I,. Isaac & R. E. F. Leakev. Eds) Earliest .Ilon and ICronments II , Palaeoecolo,~~, and Evolution, pp. 552~3i4, Chicago: University of CZhirago Press. Isaac, G. L. (1977). O/or,pailie: .irrhaeo/ogical .Studzes q/ u .\fidd/r Pl&forene Lakr Hn.rv~ (Chicago: I’niversity of Chicago Press. Isaac, G. I,. (197&z). ‘I‘he food sharing hchavior of pr~~twhunun hominids. .Sr/m/. ;Im. 238, 90&108. Isaac. G. 1~. ( 1978h). Food-shariq and human r\vlution: .~rrha~~olo,gical cvidcncr from thr Plio-Pleistocene 01 East .4frica. J. Anthropo/. I&J. 34, 3 I 1~325. Isaac, G. L. (1980). Casting the net wide: a review ~~l’arcl~,tr~~h~gical cviderrcc liar carI) hominid land-use and rxolo,+cal relations. In (I,. K. Konig-sson. Ed.) C’urwnf.4rqw~~n/~ on Earir.Mnn, pp. 226 231. Oxford: Pugamon Press. Isaac. G. I.. ( 1981a). Archaeological trsts of altwnatiw m<~drlc 01 carrly hominid bchavlor: excavation and rxperiments. Philosoph. Trans. K. .SOL.London B292, 17i-188. Isaac. G. L. (1981b). Stone Ag-e visiting cards: approachrs to thr stud\ ofrarly Iand-use patterns. In (I. Hodder, G..L. Isaac & N. Hammond, Eds) Patterns in the Past. pp. 37- 103. (:amhrid,gc: Camhridqc Univrrsity PTPS\. Isaac. G. I,. (1982). ‘l‘he earliest archarological tracrs. In [J. I). (Clark. Ed.) The Cambri&r Hzrto~>‘qfL~fCca. L’o[. 1. pp. 157-247. (:ambrid,qe: Cambridge Univrrsit) Prrss. Isaac. G. I,. (1983). Bones in contention: competing explanations tiw the .juxtaposition of rarly Pleistocrnc artifacts and fauna1 remains. In (J. Glutton-Brock & C:. Grigson. Eds) .InimolrandArchaeolo~~~: C’ol. I: Huntercand Their Py, pp. 3-19. Oxford: British Archaeological Rrports 1B.4R).InternationaI Serirs. Isaac. G. L. (1984). The archaeology ofhuman origins: studies ofthc Loww Plristocrnr in East Africa 197 1-1981. In (F. Wendorf & A. Close, Eds) Adwu-es in CVorld ArchaPoiqp 3, l-87. Isaac. G. I,. & Harris..]. 1%‘.K. (1975). Scattrrbuwcen thr patches. Paperdclivrrrd to Kr(,rhcr,\nthropological Sociq Annual Mrrtings, Bcrkrlry. Isaac. G. L. & Harris.,J. W. K. (1978). Archaeology. In (11. L.t~kv\ ti R. E. F Ixakrv. &is) &ohl Ipora Koearth Project. Oxford: Clarcndon Press. Isaac. G. I,., Harris,,J. W. K. 9r Grader. 1). (1976). Archacoluqc,d r,\idcncr Ii-om thr Koobi Fora Formation. In iY. Copprns, F. C. Howell, G. I,. Isaac & R. E. F. Lcakry. Edsi ICor(iert .lIann and Enoimnmentr in the Lake &do!/ Baszn: Strati,~ra,b~v. Paleoecolqv. and Evolution, pp. 533-551. (Zhicaqr: University of Chicago Press. Isaac. G. I,., Leakey. R. E. F. & B~hr~nsmcycr, A. K. i 197I). .4rchacological traces oEcarlv hominid activities. east of Lake Rudolf; Kenya. Scierrcr 173, 1129-l 134. Kaufulu, 2. (1983). The Geological Context of Some Early :1rchacological Sites in Kenya. .\lalaw. and Tanzania: Microstratigraphy, Site Formation, and Interpretation. Ph.D. dissrrtation, University ofCalifornia. Keelry, L. H. & Toth. N. (1981). Microwcar polishes on parlv stonr tools from Koohi Fora, Kenya. Yaturr 293, 466465. Klein. R. J, & Cruz-Uriha, K. ( 1984 ). The .4na~wv oj~.~rrimu/ Bonr\ [rum .Archaeologlral .Site,. (Chicago: University ol Chicago Press. Kortlandt, A. (1986). The use of stone tools by Lvild-living chimlunzws and earliest huminids. ,/, hum. ELJO/. 15, 77-132. Kroll, E. M. (n.d.) ‘l‘hc Anthropologuzal Meaning of Spatial (1onligurations at Plio-Plristocenc Sites in East Africa. Ph. I). Dissertation. University of(:alifornia. Bcrkrlpy. Kroll. E. M. & Isaac, G. I,. (1984). Conligurations ofartifacts and bunt-s at early Plcistocrrrr sites m East Africa. In (H. J. Hietala. Ed. I Intrasite .S,bpatial .4na!wir in .4rchaeol,gqv. pp l-87. Cambridgr: (Iamhridgr L[niwrsit) Press. Leakey, M. D. (1970). Early artrfacts from the Koohi Fora area .Ihfure 226, 2X-230. Leakey, M. (1971). Olduoai Gorge: f&ar~atzons in Beds I &II. /96&/W. Cambridge: Camhridgc Univrrsity Press. Leakey, M. (1975). Cultural patterns in thr Olduvai sequence. In (K. W. Butzer & G. I.. Isaac. Eds) .4jier the ilustralopithecines, pp. 477-493. Thr Hague: Mouton.

Leakey, R. E. F. & M. G. Leakey (Eds) (1978). Koobi F ora Rermrch Prqect, Volume 1, The Fossil Hominzds and an Introduction to Their C’ontext, 19&%197-/. Oxford: Clarendon Press. Ler. R. B. (1976). The !Kurq San: Men, Women. and W’ork zn a Fora,pq .Sucie~. (Zamhridge: (Cambridge Universit, Press. Lee, R. B., & DeVore. I. (Eds) (1976). Kalahari Hunter-(Atherers: .Studies of the !Kuq San and theu Xiighbon. Cambridge, Mass.: Harvard University Press. Loy. ‘I‘. H. (1983). Prehistoric blood r&dues: detection on tool surlhccs and idrntification ofspecics of origin. Sczence 220, 126Sl270. McBeath. N. & McGrew. W. ( 1982). ‘l‘ools used hy wild chimpanztv to obtain termites at Mt. Assirik. Scnrgal: The influence of habitat. J. hum. Ewl. 11, 6572. McDougall, I. (1981). 40Ar/39Ar spectra from the KBS ‘l’ulf: Koohi Fora Formation. ‘V&we 294, 120-124. McDougall, I. R., Maier. R., Sutherland-Hawkes, P. & Gleadow. A. (1980). K-Ar age estimates for the KBS TufX East Turkana, Kenya. Nature 284, 230-234. McGrew, W. C. (1974). Tool use by wild chimpanzees in feeding upon wild driver ants.,/. hum. Ewl. 3,X-508.

BEHAVIORAL

INFERENCES

787

FROM EARLY TOOLS

McGrew, W. C. (1977). Socialization and object manipulation ofwild chimapnezees. In (S. Chevalier-Skolnikoff & F. E. Poirer, Eds) Primate Biosocial Development, pp. 261-281. New York: Garden Publishers. Nishida, T. (1973). The ant-gathering behavior by the use of tools among wild chimpanzees of the Mahali Mountains. J. hum. Euol, 2, 357-370. Nishida, T. & Hiraiwa, M. (1982). Natural history ofa tool-using behavior by wild chimpanzees in feeding upon wood-boring ants. J. hum. Euol. 11, 73-99. Potts, R. (1982) Lower Pleistocene Site Formation and Hominid Activities at Olduvai Gorge, Tanzania. Ph.D. Dissertation, Harvard University. Potts, R. (1983). Foraging for fauna1 resources by early hominids at Olduvai Gorge, Tanaznia. In (C. Clutton-Brock & C. Grigson, Eds) Animals andArchaeology: I: Hunters and TheirPrey, pp. 51-62. Oxford: British Archaeological Reports (BAR) International Series. Potts, R. (1984a). Home bases and early hominids. Am. Scient. 72, 338347. Potts, R. (19846). Hominid hunters? Problems of identifying the earliest hunter/gatherers. In (R. Foley, Ed.) Hominid Evolution and Community Ecology, pp. 129-166. London: Academic Press. Potts, R. & Shipman, P. (1981). Cutmarks made from stone tools on bones from Olduvai Gorge, Tanzania. Nature 291,577-580.

Schick, K. (1986). Stone Age sites in the making: experiments in the formation and transformation of archaeological occurrences. Oxford: Bristol Archaeological Reports (BAR), International Series, No. 314. Sept, J. (1984). Plants and early hominids in East Africa: a study ofvegetation in situations comparable to early archaeological site locations. Ph.D. dissertation, University of California, Berkeley. Shipman, P. (1983). Early hominid lifestyle: hunting and gathering or foraging and scavenging? In (J. Clutton-Brock & C. Grigson, Eds) Animals and Archaeologv: I. Hunters and Their Prey, pp. 31-49. Oxford: British Archaeological Reports (BAR), Shipman, P. (1986). Scavenging or hunting in early hominids: theoretical framework and tests. Am. Anthrop. Stiles, D. N., Hay, R. L. & O’Neal. J. R. The MNK chert factory site, Olduvai Gorge, Tanzania. Wld Archaeol. 5, 285308.

Struhsaker,

T. T. & Hunkeler,

P. (1971). Evidence of tool-using

by chimpanzees

in the Ivory Coast. Fol. Primatol.

15,212-219.

Sugiyama,

Y. & Koman,

J, (1979). Tool-using

and making

in wild chimpanzees

at Bossou, Gninea.

Primates 20,

513-524.

Tavoso,

A. (1972). Les industries

Monaco 18, 113-144.

de la moyenne

terrasse du Tarn a Tecou (Tarn).

Bull. Mus. d’Anthropo1. Prehist.

1 Toth, N. (1982). The Stone Technologies of Early Hominids from Koobi Fora, Kenya. Ph.D. dissertation, University of California, Berkeley. Toth, N. (19850). Archaeological evidence for preferential right-handedness in the Lower and Middle Pleistocene, and its possible implications. J. hum. Euol. 14, 607-614. Toth, N. (19856). The Oldowan reassessed: a close look at early stone artifacts. J. Archaaol. Sci. 12, 101-120. Toth, N. (1987). The emergence of flaked stone technologies. Scient. Am., April. Toth, N. & Schick, K. (1986). The first million years: the archaeology ofproto-human culture. In (M. Schiffer, Ed.) Advances in Archaeological Method and Theory. pp. l-96. London: Academic Press. Toth, N. and Woods, M. (in prep.). The use of mollusc shell as a cutting implement, and modification patterns on bone surfaces. Villa, P. (1983). Terra Amata and the Middle Pleistocene Archaeological Record ofSouthern France. Berkeley: University of California Press. Vincens, A. (1979). Analyse palynologique du site archeologique FxJj 50, Fromation de Koobi Fora, Est Turkana (Kenya). Bull. Sot. Geol. France 7, t. XXI, (3), 343-347. Vrba, E. (1980). The significance ofbovid remains as indicators ofenvironment and predation patterns. In (A. K. Behrensmeyer & A. P. Hill, Eds) Fossils in the Making: Vertebrate Taphonomy and Palaeoecolo~, pp. 247-271. Chicago: University of Chicago Press. White, T. E. (1952). Observations on the butchering techniques of some aboriginal peoples, 1. Am. An+. 17(4), 337-338. Yellen, J. (1977). Archaeological Approaches to the Present. New York: Academic Press.