Document Outline

1993

). In gen-

eral, the former have generally used sieves longer and traditionally

pay more attention to chipped stone remains during recovery,

analysis, and publication.

797

0305–4403/99/070797+12 $30.00/0

1999 Academic Press

third possible reason for the paucity of archaeological

metal finds is that early metals were chemically un-

stable and decomposed relatively rapidly under most

conditions. Considering any or all of these reasons,

little direct metallic evidence exists to show the range

of all types of early metal tools (

) and to determine

exactly when the change-over from a stone- to a

metal-oriented technology took place. Did this tran-

sition take place slowly or rapidly? Was the spread of

metallurgy a relatively uniform process? These are vital

questions which must be answered before one can

address the question of causal priority in the adoption

of metallurgy.

This paper will present the results of new research

into the origins and spread of metallurgy from a new

perspective—the analysis of cut-marks on the bone

remains of animals slaughtered and butchered by metal

and stone implements. It will be shown that cut-marks

on bones made by chipped stone tools during the

butchering of animals can be distinguished from those

made by metal tools. By examining the di

fferences in

cut-marks, it should also be possible (1) to expand our

understanding of the types of butchering tools in each

of the prehistoric periods and (2) to more accurately

calculate the relative importance of stone versus metal

in the subsistence technology. Thus, cut-marks can be

used, in the absence of metal tools, to study the

introduction and spread of metallurgy both within and

between regions (and potentially even within complex

societies) through time.

This investigation was accomplished in two steps:

first, through the analysis of modern experimental cut-

marks made by the author with metal and stone tools

and, second, by the comparison of the results of the

cut-mark experiments with cut-marks on bones from

prehistoric sites spanning the introduction of metal

tools in the central Balkans. The central Balkans of

southeastern Europe were chosen to supply the com-

parative archaeological material because this is one of

the Old World regions which experienced the auton-

omous development of metallurgy (

Jovanovic´, 1980

Renfrew, 1969

). The zooarchaeological remains with

cut-marks used in this study come from two prehistoric

sites: Petnica and Ljuljaci (

both located in central Serbia. Their data will be used

to demonstrate the utility of the method.

The slicing cut-marks examined in this study are the

residual remains of slaughtering, butchering and skin-

ning activities. A cut-mark is functionally equivalent to

a slice on the bone created by the drawing of a knife (or

dagger) blade across the surface of the bone. It is this

type of cut-mark that is being studied here. A slicing

cut-mark is not to be confused with a chop-mark,

which is created by the impact of a knife, sword, or

axe-like blade. It is also not to be confused with the

slicing-like activity of a saw. The marks produced by

chopping and sawing are easily distinguished from

those of slicing cut-marks (

Blumenschine, Marean & Capaldo, 1996

Previous Research on Later Prehistoric Metal

Versus Stone Tool Cut-marks

There has been a great deal of research over the last 20

years in distinguishing chipped stone tool cut-marks on

bones from other kinds of marks on bones (teeth,

trampling, vascular grooves, roots, preparator-marks,

etc.—e.g.,

Olsen & Shipman, 1988

Greenfield, Jezˇ & Starovic´, n.d.

Potts & Shipman, 1981

). In later

prehistoric/early historic faunal assemblages, slicing

cut-marks are not easily confused with other kinds of

marks commonly studies (e.g., tooth and preparator-

marks). There has been little attention directed at

distinguishing cut-marks made by prehistoric or

historic stone from metal butchering implements.

Walker & Long (1977)

conducted a series of exper-

iments that initially established the relationship

between the edge characteristics of a series of stone and

metal cutting tools and the marks they produce when

applied to bone. Their experiments were the first to

indicate that clearly recognizable morphological di

ffer-

ences existed between the cut-marks of metal and stone

knives. The results of their research are supported by

this study.

The most extensive replication study of metal versus

stone-cut tool marks was conducted by

Olsen (1988)

a seminal, but relatively unnoticed study. She was the

first to examine the relative abundance of metal versus

stone tool slicing cut-marks on bone, to do so through

the experimental replication of cut-marks on bone by

a variety of metal and stone tools, and was the first

to utilize a scanning electron microscope (SEM) to

investigate stone and metal cut-marks in a later

prehistoric context. She developed a series of morpho-

logical criteria for distinguishing stone from metal

tools and types of metal tools using a SEM. Olsen was

mainly concerned with the analysis of bone and antler

artefacts from the British Bronze and Iron Ages, and

was attempting to understand the production tech-

niques for such tools. The results of her study are

corroborated and enhanced by the data presented here.

fferences Between Stone and Metal Tools

There are some fundamental di

fferences between stone

and metal tools that are relevant to the analysis at

hand. First, experiments with steel knives have shown

them to be superior to stone flake tools in a number of

ways. They are stronger, have greater longevity, retain

their cutting edge longer, are generally sharper, can be

more frequently and extensively sharpened, and re-

quire less energy to cut through greater amounts of

tissue with fewer strokes (

Walker, 1978

Second, as a result of the heavy investment in raw

material procurement and manufacture, metal tools

are kept and used for long periods of time and not

quickly discarded. In contrast, chipped stone tools

have a shorter functional life (

Brose, 1975

). Stone tools

798 H. J. Greenfield

have the advantage that their raw material is often

much more readily available and their production

requires less energy and specialized manufacturing

technology. This implies that they can be more easily

produced and were probably more frequently

discarded.

Third, the relative e

fficiency of stone and metal tools

seems to vary by function. For example,

Steensberg’s

(1943)

experiments on flint, bronze, and iron sickles

indicate virtually no di

fference in efficiency between

sickles of bronze and flint.† In contrast,

Mathieu &

Meyer’s (1997)

experiments with stone, bronze, and

steel axes show that bronze is as e

fficient as steel for

felling trees, and that both types of metal axes are more

fficient than stone axes.

The Experiment: Methodology

A series of experiments comparing metal and stone

tool cut-marks was conducted by the author. The

resultant marks were examined under various levels of

power using a SEM. The SEM o

ffers high resolution

images, with a great depth of field and a wide range of

magnifications. Most or all of the surface of the object

can be brought into focus at once with the SEM

(

Olsen, 1988: 341

). This contrasts with the use of

photomicrographs from an optical microscope where

the curvature of the bone and the depth of many of the

cut-marks inhibit high quality photomicrographs. A

variety of shapes of steel knives and chipped stone

tools were chosen to try to account for the source of

variability in the analysis. Each blade was drawn

across a soft wooden board (pine) in the same direc-

tion, and with the same angle and hand-held pressure.

A soft wood was chosen as the medium, rather than

bone, because it is softer and more likely to accurately

record details of the imprint of the blade during the

cutting process. The problem with conducting the

exercise on bone is that di

fferent parts of each bone

have varying degrees of hardness and angle (

Lyman,

1994: 238–252

In order to analyse the cut-marks in a SEM, small

moulds of the cut-marks were made.‡ A variety of

magnifications were used for viewing the specimens.

The morphology of the cut-mark was obscured if

the magnification was too high. In general, lower

magnifications (30–100

power) were sufficient for

observing the diagnostic criteria. Higher power obser-

vations served to confirm what was already visible at

the lower levels. The magnification used was, to some

extent, dependent on the size of the object under

observation and the range in sample size was a func-

tion of the cut-mark itself. The larger the cut-mark

(width, not length), the lower the power that could be

used.

The angle of observation was also important for the

accurate identification of slicing cut-marks. When

viewed from directly overhead (90

angle), cut-marks

lose their shape and depth. In general, an angle of

75–90

was preferred because it enhanced rather than

obscured the morphological characteristics of slicing

cut-marks. The best perspectives were generally from

the side of the specimen where the edge of the mould

was cut and the profile could be brought into view with

the ridge behind it. This allowed the profile to be

accurately drawn. However, the shape of the ridge and

any evidence for ancillary striations were also crucial,

and the SEM often had to be moved to a di

fferent

position for their viewing.

Results of the Experiment

Steel knife-marks
Twelve di

fferent metal steel knives were used during

the experiment (

Table 1

). These knives were chosen to

reflect a variety of blade shapes, some of which were

similar to metal blade shapes from prehistoric assem-

blages. In general, the metal knife-marks can be

grouped into two categories: flat-edged and serrated-

edged blades. Two significant di

fferences exist between

the modern sample (used in this study) and prehistoric

assemblages (not used in this study). First, the blades

tend to be narrower in the modern assemblage.

Second, serrated-edged metal blades are absent from

prehistoric assemblages in the central Balkans.

Serrated-blades were included in the study to deter-

mine if they would have a di

fferent morphology than

smooth blades. The results from each type were quite

fferent and are described below.

Serrated-edged blades

Knives with serrated-edged blades could be divided

into two types: those with high and widely spaced

serration (such as steak and bread cutting knives) and

knives with a low and tightly spaced serration (which

are very saw-like in function). The characteristics of

the high and widely spaced serrated knives (

Figure

) include a wide and shallow cut-mark, with poor

definition of the edges and bottom of the groove.

The edges slope very gradually and unevenly, while the

apex seems to have a wave that weaves across the

surface.

†Whether iron sickles were more e

fficient than bronze or flint sickles

remains to be determined from experimental studies.

‡The SEM chamber accepts relatively small-sized samples (2–3 cm).

Small silicone rubber moulds of each of the experimental cut-marks

were made using Dow Corning Silastic 9161 molding compound and

Cutter Perfourm Light Vinyl Polysiloxane Impression Material (type

I, low viscosity) dental impression compounds. These are extremely

sensitive media for replicating microscopic morphology (

Rose,

1983

). The shape of the mould is the reverse of the original

specimen—it is everted rather than inverted. After curing, the mould

was peeled o

ff, attached to an aluminum stub with an epoxy

adhesive, and sputter-coated with gold palladium. Gold palladium

(often mistaken as silver because of its greyish colouration) yields a

better image in the SEM because its grain size is much smaller than

any other metal (Sergio Mejia, University of Manitoba, Faculty of

Geology, Computer Imaging Laboratory—pers. comm., November

1, 1996).

The Origins of Metallurgy 799

Table 1. Summary of results of experimental tests of stone and metal blades on a soft wooden board

Raw

material

Sample

Type of instrument

Edge

Angle of V

Comments on knife

Quality of mould

Petnica Inventory #

Steel

Scalpel/razor for paper cutting

Flat-sided

Even V-shape

Did not take

groove too narrow

Steel

Medical scalpel

Flat-sided

Even V-shape

Not very sharp-used

Did not take,

groove too narrow

Steel

Medical scalpel

Flat-sided

Even V-shape

Not very sharp-used; broken tip

Did not take,

groove too narrow

Steel

Eating (table) knife

Flat-sided

Uneven V-shape

Good

Steel

Eating (table) knife

Shallow, tightly

spaced serration

Good

Steel

Serrated steak knife

Deep and widely

spaced serration

Good

Steel

Bread cutting knife

Deep and widely

spaced serration

Bread cutting side

Good

Steel

Bread cutting knife

Small, tightly spaced

serration

Bone cutting side

Good

Steel

Kitchen knife with wooden handle

Flat-sided

Uneven V-shape

Good

Steel

Kitchen knife with plastic handle

Flat-sided

Uneven V-shape

Good

Steel

Pocket (folding) knife

Flat-sided

Even V-shape

Large

Good

Steel

Pocket (folding) knife

Flat-sided

Even V-shape

Small

Good

Stone

Backed short blade

Retouched on one

side

Uneven on one side

and smooth on other

Good

6762

Stone

Triple backed short blade

Without retouch

Uneven on one side

and smooth on other

Good

6056

Stone

Curved single backed short blade

Without retouch

Uneven on one side

and smooth on other

Good

5099

Stone

Triple backed short blade

Without retouch

Uneven on one side

and smooth on other

Good

5613, 5013 or 58

Stone

Scraper

Without retouch

Uneven on one side

and smooth on other

Good

6118

Stone

Short blade

Without retouch

Uneven on one side

and smooth on other

Good

4976

Stone

Long blade

Without retouch

Uneven on one side

and smooth on other

Good

8389

Stone

Long blade

Without retouch

Uneven on one side

and smooth on other

Good

5635

Stone

Curved short blade

Without retouch

Uneven on one side

and smooth on other

Good

5166

Stone

Large scraper

Without retouch

Uneven on one side

and smooth on other

Good

Stone

Small scraper

Without retouch

Uneven on one side

and smooth on other

Good

Stone

Long blade fragment

Without retouch

Uneven on one side

and smooth on other

Good

800

H.
J.

Greenfield

The low and tightly spaced serrated knives (

Figure 2

) have

a di

fferent pattern. The blade is flat on one side and scalloped

on the other. It makes a broad and relatively shallow

groove, with sides that gradually slope downwards until

half the depth is reached and then slope at a steeper angle.

The slope is much more gradual on the left side of the

groove than on the right side. This pattern is found on all

serrated knives, but is accentuated on the tightly serrated

knife edges. This pattern would be di

fficult to distinguish

from that of some of the stone tool cut-marks because

both sides of the blade do not have the same shape.

Flat-edged blades

This type includes modern scalpels, razors, typical

carbon steel kitchen knives, and most pocket knives.

The cutting edges are sharpened on both sides in order

to maintain their sharpness. Both sides steeply angle at

the same degree toward the cutting edge to form a

V-shape profile (

Figure 3

). The bottom of the cut-mark

by metal blades is often slightly flattened. Only in

razor-edged blades is the bottom of the blade a sharp

V-shape.

Stone blade-marks
Twelve di

fferent sharp-edged chipped stone tool

types were initially selected for the analysis from the

prehistoric assemblage at Petnica (

). The stone tools can be typologically

divided into three groups. These are common lithic

types found on Neolithic and post-Neolithic sites in

the central Balkans.† There were six short blades

†The artefacts from the Petnica assemblage were representative of

the prevalent chipped stone types that morphologically could have

been used for butchering activities. No alternative slicing tool types

were found in the assemblage. It is unlikely that part of the

assemblage is not represented owing to the extensive nature of the

excavations and that the site is a sedentary settlement (

Greenfield,

1986a

). The names of the tools are based upon the formal typology

used by local prehistorians. The local lithic typology system is based

upon formal morphology rather than use-wear. For example, the

fference between short-and long-blades is probably an artificial

fference as a result of breakage during use. The tools selected for

this analysis appear to still be functional for butchering activities

since there is no evidence of damage to the slicing edge and they still

possessed sharp edges. They may have been used originally for a

variety of activities but this cannot be determined without extensive

edge-wear analysis.

Figure 1. SEM photograph of the groove from modern metal knife

8, 25

magnification, 80.

Figure 2. SEM photograph of the groove from modern metal knife

7, 206

magnification, 75.

Figure 3. SEM photograph of the groove from modern metal knife

9, 200

magnification, 80.

The Origins of Metallurgy 801

(lamella—Serbian) (Stone 1–4, 6, & 9), three long

blades (nozˇ) (Stone 7, 8, & 12), and three scrapers

(strugac) (Stone 5, 10, & 11). The blades included single

(Stone 1 & 3) and triple backed or platformed blades

(Stone 2 & 4), and curved blades (Stone 3 & 9). The

scrapers included a large (Stone 10) and a small

example (Stone 11). One blade (Stone 1) had retouch

on its cutting edge,† and this was the side used in the

experiment. All of the other stone samples lacked any

obvious evidence of retouch. It was anticipated that

fferent types of chipped stone tools would yield

characteristic cut-marks.

Tools from Petnica were used in the study since most

of the faunal remains with cut-marks are derived from

that site. Using tools from the same site as the faunal

remains arguably minimizes the morphological vari-

ability in cut-mark shape and some of the di

fficulties in

associating cut-marks with particular types of stone

tools. The tools were in extremely good shape, did not

have any evidence of di

fferential wear or patina, and

were still sharp. The same procedure for making cuts

on wood, making the mould, and observing it under an

SEM was carried out with the chipped stone tools.

Each tool was hand-held and sliced across the wooden

board—the same as for the metal tools, but on the

opposite side of the board.‡

Long blade

The cross-section of long blades (

Figure 4

) was steeply

sided on one side, and more gradually sloping in a

series of parallel ridges on the other. The apex was

relatively narrow, but not razor sharp or flat.

Short blade

The pattern of the short blades (

Figure 5

) is similar to

that of the long blades. No distinguishing diagnostic

criteria could be identified that would allow them to be

fferentiated from long blades. The cut-mark of one of

the blades (Stone 2—a triple-backed blade) resembled

that of a scraper at the terminating end of the cut-

mark. This illustrates the danger of relying only

upon di

fferent ends of the cut-marks for the analysis.

Each identification should be based upon the

same (initiating) end of the cut-mark to ensure

comparability.

At lower magnifications (50

), one of the short

blade cut-marks is similar to that made by a metal tool.

It has sharply angled sides that rise steeply from the

base of the mould. At higher magnifications (100

and

above), it does not resemble metal knives. It has typical

characteristics of stone tool cut-marks. Here, the two

ridges along the apex are visible and the left ridge is

lower than the right. The left side descends more

gradually than the right side, which is steeply sloping.

Scraper

The cut-mark of scrapers (

Figure 6

) resembles that of

the scallop-edged metal knives. It is very shallow, with

slowly sloping edges, and the appearance of a wave-

like pattern along one side. The other side tends to be

smoother. The bottom of the groove tends to be

relatively horizontal, with only a slight slope to the side

†It could not be determined whether the retouch was the result of

sharpening or caused by use.

‡This experiment is only the first step in a more extensive study. The

experiment will be replicated in the future with modern fresh stone

blades, with di

fferent types of metals, and on bone and wood. A

preliminary comparison of the cut-marks made on wood and bone

with fresh tools indicates no major di

fference between them.

Figure 4. SEM photograph of the groove from Petnica stone tool

12, 139

magnification, 80.

Figure 5. SEM photograph of the groove from Petnica stone tool 1,

magnification, 75.

802 H. J. Greenfield

where it rapidly descends. One scraper (

Figure 7

)

exhibits a very di

fferent pattern. It is also low and

broad, but rises quickly on the left side and descends

more slowly to the right, in a series of parallel ridges.

This example exhibits a pattern common to scrapers. It

can be expected that because of the variability in edge

morphology of scrapers that there will be substantial

degree of variability in scraper cut-mark patterns.

The cut-marks of the long and short blades, as a

whole, can be distinguished from those of scrapers. The

former are more sharply defined, with higher sides,

narrower cross-sections, and well-defined ancillary

ridging, while the latter lack these characteristics. In

contrast, long blades were not distinguishable from

short blades.

Distinguishing Stone from Metal Cut-marks

Based upon the above experiment and previous studies,

it is possible to identify a readily observable set of

diagnostic criteria for distinguishing stone from metal

slicing cut-marks (

Figure 8

). This study confirmed

some of what has already been observed by others

Figure 6. SEM photograph of the groove from Petnica stone tool

10, 50

magnification, 75.

Figure 7. SEM photograph of the groove from Petnica stone tool 5,

magnification, 75.

Figure 8. Profile of characteristic metal and stone tool cut-marks.

(a) Profile of sharp metal blades in

Figure 3

; (b) profile of dulled

metal blades; (c) profile of metal blades in

Figures 1

; (d) profile

of stone blades in

Figures 4

, &

; (e) profile of stone blade in

Figure 5

The Origins of Metallurgy 803

(e.g.,

Blumenschine, Marean & Capaldo, 1996

), but

allowed the first comprehensive identification of stone

and metal tool cut-mark features.

Metal knife-marks are deep and steeply sided, cul-

minating in an apex that has a sharp point or a

horizontal platform. They will have a smooth-sided,

and uniform or slightly o

ff-angle V-shaped profile,

depending on the angle of the cut. The cut can be deep

and narrow or deep and wide depending upon the

nature of the blade. Iron and steel metal knives often

create a flat-bottomed

/_/-shaped profile when they

have dulled or were not sharpened properly. In con-

trast, high scalloped cutting edges yield cut-marks that

are very uncharacteristic of metal knife-marks. They

are broad and poorly defined, and somewhat similar to

a saw (described in

). These criteria can be

summarized as follows:
(a) metal knives produce either a narrow V-shaped

groove with a distinct apex at the bottom or a

broader

/_/ shaped groove with a flat bottom;

(b) metal knives make more uniform patterns on the

bone, often removing material in the groove more

ffectively. They leave either no striations or

striations of a more uniform depth and spacing

than when stone tools are used;

more even slicing cut (except for scalloped-edge

knives and saw-like blades).

Chipped stone tools produce a shallower, less even cut,

and tend to exhibit considerably more variability in

shape (

Walker & Long, 1977:608

). The cut appears

dirty (full of debris), with the apex weaving back and

forth. Because of the sinuosity of their cutting edges,

chipped stone tools tend to produce wide and irregular

grooves (

Walker & Long, 1977:608

Figure 4

). These

grooves appear as a series of ancillary parallel

striations, lateral to the apex of the cut, and are of

uneven length and thickness. The lateral striations

appear as ridging along one side of the apex of the

ridge in SEM photos of moulds. The striations reflect

the uneven (and often retouched) dorsal surface of the

stone blade. The smooth side reflects the smooth

bulbous ventral surface of the blade. The cut-marks

are always uneven in cross-section, with one side

rising relatively steeply to the apex, then descending

gradually or in a series of ancillary ridges.

These results can be summarized as follows: metal

tools have steep and smooth V-shaped profiles, while

stone tools have two distinctly di

fferent sides—a

smooth and a rough side. The smooth side rises steeply

and smoothly; the rough side rises more gradually,

with multiple striae from the various facets left over

from production.

A Case Study: Petnica and Ljuljaci

The Balkans of southeast Europe is one of the

independent centres for the development of a metallur-

gical technology (

Jovanovic´, 1980

Renfrew, 1969

From the Balkans, this technology spread to the rest of

Europe. In the central Balkans (the location of the case

study), it is frequently assumed that the transition from

a stone- to a metal-oriented subsistence technology

occurred by 3300

, at the advent of the Eneolithic.

However, this transition is not so simple. In Neolithic

(6100–3300

†) deposits, stone implements and

waste are very common, while metal objects are

extremely rare and presumed to be limited to ritual or

social functions. During the subsequent Eneolithic

(3300–2500

) and Bronze Ages (2500–1000 ),

metal objects became more common and functional,

while stone implements became relatively scarce. The

declining frequencies of stone tools during the Copper

and Bronze Age have been the major indicator of the

importance of metallurgy since significant quantities of

metal tools do not appear in the archaeological record

until the Late Bronze Age (1300–1000

).

Generally, it is assumed that there was an increase in

the use of metal tools for slaughtering and butchering

of animals through time. If this hypothesis is valid,

there should be an increasing frequency of metal tool

cut-marks on animal bones over time. To test this

hypothesis, a prehistoric sample that cross-cuts the

Neolithic–Bronze Age divide was sought. As a result,

data from two sites in the central Balkans are presented

here—Petnica and Ljuljaci. The sequences from the

two sites encompass the Neolithic–Bronze Age divide.

The prehistoric site at Petnica is located near the

town of Valjevo (in central Serbia, Yugoslavia), in a

valley in the Serbian foothills about 90 km SW of

Belgrade. The faunal assemblage was excavated by

Z{eljko Jezˇ from 1980–1986. It is a small (c. 3 ha in

area) open-air site, at the base of a steep escarpment,

with a view all the way down the stream valley to the

Kolubara river. It represents the remains of a small

agricultural village, without any evidence for special

function or high status. It has a well-preserved and

continuous occupational sequence from the Middle

Neolithic (Vincˇa B culture), Late Neolithic (Vincˇa C–D

cultures), and Eneolithic (Baden-Kostolac culture;

3300–2500

), followed by a break until the Late

Bronze Age and Early Iron Age (Halstatt A–B culture;

1300–800

), with another break until the Roman

period (

Greenfield, Jezˇ & Starovic´,

n.d.

Starovic´, (1993)

. Unfortunately, Petnica is miss-

ing a crucial phase of the regional culture history—the

Early and Middle Bronze. This gap is filled by the data

from Ljuljaci.

Ljuljaci (Milica Brdo) is located near the town of

Kragujevac (in central Serbia, Yugoslavia), among the

foothills of mount Rudnik. It lies on a small raised

plateau, is di

fficult to access, and commands a good

view of the surrounding countryside. The site has

been excavated on and o

ff since the 1930s. The data

presented here derive from 1976–1979 excavations

†All dates are based upon calibrated radiocarbon dates (

;

804 H. J. Greenfield

conducted by Dragoslav Srejovic´ (University of

Belgrade)

and

Milenko

Bogdanovic´

(National

Museum, Kragujevac). Three phases of occupation by

the Vatin culture were identified at the site: Ljuljaci

I—Early Bronze Age; Ljuljaci II–III—Middle Bronze

Age. The site was a small fortified village, which was

probably the residence of relatively high status individ-

uals. Ljuljaci is argued to be a high status settlement

based on a number of anomalies when compared to

other contemporary sites in the area—e.g., the presence

of metal artefacts, substantial structures, a large

quantity of fine wares, and a faunal assemblage

containing an unusually large number of wild animals

(boars) and domestic horses (

Bogdanovic´, 1986

Methodology
Two methods were employed in the following analysis

in order to determine the temporal distribution of

stone versus metal cut-marks: (1) observation of the

original bone cut-marks at low power with a reflecting

light microscope (data summarized here), and (2) ob-

servations using silicone moulds made of some of the

cut-marks, which have been examined with a SEM.

The procedure described below follows that sug-

gested by

Olsen (1988: 341

). First, the bones were

examined for macroscopic traces of tool cut-marks.

Tooth-marks on bones (dogs, pigs, rodents, etc.—

Greenfield, 1988

Lyman, 1994

) are easily distinguish-

able from cut-marks and must be removed from the

sample beforehand. Since the prehistoric sample of

bones from Petnica and Ljuljaci examined in this study

had a substantial fraction of canid gnawed bones

(

), such bones were identified and

removed from the sample prior to this study. All of the

bones were initially examined for cut-marks that were

generally visible to the naked eye during the initial

analysis of the zooarchaeological assemblage from the

site (

Champion et al., 1984: 163

). Bones with identifiable

cut-marks were set aside for further analysis. Second,

samples were selected for study through the SEM.

Almost one-quarter of the Petnica cut-mark assem-

blage was examined in the SEM (23·2%; N=45 of 194)

to check the accuracy of observations made with a

low-power optical microscope. They were chosen to be

representative of each period and cut-mark type.

Third, since most pieces of bone are too large to be

placed into and studied directly in the SEM chamber,

small silicone rubber moulds of the cut-marks were

made of the same material as the experimental moulds

(above).

All of the bones in the Petnica assemblage were

examined for cut-marks. Over 300 temporally-

provenienced animal bones with slicing cut-marks were

originally identified from the various strata. A substan-

tial proportion of this sample, however, was not

included in the final analysis because of evidence of

erosion on the bone surface which damaged the fine

characteristics necessary to discriminate between stone

and metal tools. In the end, only 194 bones with

cut-marks were used in this analysis.

Far fewer (N=26) bones were identified as having

cut-marks from Ljuljaci, but the overall sample size is

also much lower. Only 13, however, were well-enough

preserved to permit identification of the type of instru-

ment used to make the cut-marks. Owing to the small

sample size of identifiable remains from this site, the

data from all three horizons at Ljuljaci were lumped

together for the purposes of this analysis. No valid

temporal trends were perceptible from the data when

separated by horizon.

Results and discussion
The results of the optical microscope are summarized

Table 2.

Stone tool cut-marks appear in each of

the periods. Their percentage declines over time (the

unusual metal frequencies in the later phases will be

discussed later). Metal cut-marks have a very di

fferent

distribution. In general, the data demonstrate that the

incidence of metal cutting implements is minimal prior

to the Bronze Age. In the Middle Neolithic levels, they

are found in such small numbers (5·8%; N=1) that they

Table 2. Summary of results of optical microscope cut-mark analysis on prehistoric faunal remains from Petnica

(1980–1986 excavations) and Ljuljaci

Stratum (culture)

Date

(cal.)

Stone

Metal

Middle Neolithic (Vincˇa B)

4500–4200



94·12

5·88

Late Neolithic (Vincˇa C)

4200–3800



90·91

9·09

Late Neolithic (Vincˇa D)

3800–3300



83·72

16·28

Eneolithic (Baden-Kostolac)

3300–2500



86·36

13·64

Early-Middle Bronze Age (Vatin)

2500–1500



15·38

84·62

Late Bronze-Early Iron Age (Halstatt A–B)

1000–800



58·54

41·46

Roman*

 100–300

91·67

8·33

Total

150

* Roman pits, filled with animal bones, intrusive into Vincˇa C horizon—94% Vincˇa ceramics.

The Origins of Metallurgy 805

can probably be attributed to the occasional mis-

identification owing to the use of an optical micro-

scope. Metal tools begin to appear in some quantity

during the Late Neolithic (Vincˇa D culture) at the site

(16%). This is the period of earliest metallurgy in the

Balkans. Large copper veins were mined in nearby

eastern Serbia and metal axes and other implements

appeared in sites throughout the region (

Jovanovic´,

1980

During the Eneolithic, the frequencies of metal tools

remain low (13%) attesting to their continued but low

representation. The quality of metal tools for slicing is

probably minimal ultimately resulting in their low

numbers. The presence of substantial frequencies

of metal cut-marks during the Late Neolithic and

Eneolithic (13–16%) is quite surprising since early

copper tools would probably not have been very

fficient for cutting (

Greenfield, forthcoming

). The

cut-mark analysis indicates that early metal tools are

being used for cutting despite their supposed in-

fficiency. This could imply that copper is somehow

being hardened. Even pure copper, when cold-worked,

can be hardened to the level of tin–bronze before

cold-working (Brinnel value of 100—

Shephard, 1980:

165

). This has implications for the assumption that

early copper tools were not utilitarian.

The numbers of metal cut-marks dramatically

increases during the E-MBA at Ljuljaci. There is a

substantial increase in the proportion of metal cut-

marks (84%) that coincides with the appearance

of high tin–bronze tools (

). This would

indicate that bronze tools are e

ffective for butchering

from early on in the Bronze Age, contrary to the belief

that metal tools would only become e

ffective butcher-

ing implements when high tin–bronzes are developed

(e.g.,

). The dramatic

increase in metal cut-marks between the previous

Neolithic periods and this period may be a result of the

fferent types of sites being studies (see later).

The number of cut-marks from Petnica during the

Late Bronze and transition to the Early Iron Age is

dramatically lower than at E-MBA Ljuljaci (41%). It is

interesting that even though high tin–bronze knives are

typical of this period and are e

ffective cutting tools,

stone tools remain important at Petnica. Part of the

reason that the proportion of cut-marks in the two sites

do not follow the same temporal pattern may be their

relative position within the regional settlement system.

Ljuljaci is a regional centre, with dramatic evidence for

high status residences. Petnica is a small undistin-

guished farming settlement. Therefore, it is not surpris-

ing that access to high tin–bronze bronze metal cutting

implements was greater and earlier at Ljuljaci than at

Petnica.

The proportions of cut-marks at Petnica during the

Roman period are di

fficult to determine since the

Roman deposits were pits that cut into and were mixed

with material from the underlying layers (i.e., Late

Neolithic strata). Since most of the ceramics in the

Roman pits were from the Late Neolithic (90%), it is

not surprising that the high stone percentage of cut-

marks on bones in the Roman pits reflects a more

Neolithic frequency pattern. In other words, these data

should be ignored.

In conclusion, the hypotheses that there should be

an increasing frequency of metal tool cut-marks on

animal bones over time has not been falsified. As a

result, it can be concluded that it is supported by the

data, although di

fferential access by status is a

complicating factor.

Conclusion

While most research concerned with the origins of

metallurgy has relied upon the metal artefacts, this

approach is confounded by a major problem: the

number of early metal tools from the earliest pre-

historic periods (Neolithic, Eneolithic, and Bronze

Ages) is small, and almost certainly, does not reflect the

full range of metal tools then available.

The research presented here provides the means to

investigate the origins and spread of metallurgy in the

absence of metal artefacts. This was accomplished first

through the analysis of modern experimental cut-

marks made with metal and stone tools and second by

the comparison of the results of the cut-mark exper-

iments with cut-marks on bones from the prehistoric

sequence of the central Balkans. The zooarchaeological

remains with cut-marks from the prehistoric site at

Petnica and Ljuljaci (

both located in central Serbia, were presented to

demonstrate the utility of the method.

Experimental replication of cut-marks using chipped

stone tools and steel knives yielded consistent di

ffer-

ences in morphology which allowed their cut-marks to

be distinguished under high magnifications. Metal

knives produced cuts with either a sharp V- or a broad

/_/-shaped profile, and which lacked any parallel ancil-

lary striations. In contrast, stone knives produced cuts

with more irregularly shaped profiles, with a deep

groove at the bottom of a steeply angled side, and then

a gradual rising of the slope with one or more parallel

ancillary striations.

When the knowledge gained from the experimental

results was applied to the faunal remains of the two

central Balkans sites, little evidence for metal tool use

for butchering was found during the Late Neolithic

and Eneolithic periods. Metal tool cut-marks appeared

in substantial numbers during the Bronze Age, and

continued into the Early Iron Age. The presence of

much higher metal cut-mark frequencies at Early–

Middle Bronze Age site of Ljuljaci, in comparison to

the Late Bronze Age site of Petnica, was interpreted

as evidence of di

fferential availability of high quality

metal cutting implements between settlements. It was

somewhat surprising to observe the relatively high

806 H. J. Greenfield

frequency of metal cut-marks so early in the Bronze

Age at Ljuljaci. This pattern somewhat contradicts the

belief that bronze tools would not have been e

ffective

butchering tools until the end of the Bronze Age. The

opposite seems to have been the case. Stone tools,

however, continued in popularity for butchering

throughout the Bronze Age, especially at low level sites

in a regional settlement hierarchy (Petnica).

The patterns observed for the central Balkans

parallel those documented in the Levant (

Rosen, 1984:

504

) and Britain (

). Functional chipped

stone tool types gradually disappear between the end

of the Chalcolithic and Iron Age. The first stage in the

adoption of metallurgy did not involve the wholesale

replacement of flint tools (as is commonly assumed).

Rosen (1984: 504

) has suggested that until a clear

improvement in e

fficiency emerges, the economy would

perpetuate the use of the traditional material. Thus,

one would not expect the replacement of flint sickles

until iron became readily available and cheap enough

to supplant them in the Levant. Similarly, one would

not expect the replacement of bronze axes with iron or

steel axes until some factor besides relative e

fficiency

intervened (Meyer & Mathieu, in press).

The development and adoption of metallurgy by the

cultures of southeast Europe had a ramifying influence

upon the prehistoric cultures of the rest of Europe.

Contemporary with the adoption of metallurgy there

appear changes in the archaeological record of south-

east Europe which may signal shifts in economic,

social, and political systems (e.g., hereditary elites

begin to dominate the landscape, controlling the

production and distribution of goods). The introduc-

tion of metal also encouraged one of the greatest

post-Glacial ecological changes—a dramatic increase

in the tempo of the cutting down of forests and

spread of tilled land and pastures (

140;

: chapter 1;

By being able to map out the introduction and

spread of metallurgy, it will become possible to begin

to understand the dynamic relationship between

metallurgy and the origins of complex societies. In the

Near East, it would seem that early complex societies

did not arise to control the functional metal trade.

Rosen (1984

1993

) has amply demonstrated that the

spread and acceptance of a functional metallurgy was a

long-term process, more or less completed in the Near

East only by the end of the Bronze Age. The same can

be said to be true for the Balkans (above) and for

England (