Content uploaded by Xiuzhen Janice Li
Author content
All content in this area was uploaded by Xiuzhen Janice Li on Jan 17, 2018
Content may be subject to copyright.
Inscriptions, filing, grinding and polishing marks on the bronze weapons from the
Qin Terracotta Army in China
Xiuzhen Janice Li
a
,
b
,
*
, Marcos Martinón-Torres
b
, Nigel D. Meeks
c
, Yin Xia
a
, Kun Zhao
a
a
Museum of Emperor Qin Shihuang’s Terracotta Army, Xi’an 710600, China
b
UCL Institute of Archaeology, 31-34 Gordon Square, London WC1H 0PY, UK
c
British Museum, Great Russell Street, London WC1B 3DG, UK
article info
Article history:
Received 4 April 2010
Received in revised form
7 September 2010
Accepted 11 September 2010
Keywords:
Inscriptions
Polishing marks
Bronze weapons
Scanning electron microscopy (SEM)
Rotary mechanical tools
Qin period
First Emperor
China
abstract
This paper is concerned with the inscriptions and finishing marks present on the surfaces of the thou-
sands of bronze weapons recovered together with the Terracotta Army at the mausoleum complex of Qin
Shihuang, the First Emperor of China (259e210BC). After utilising the textual information from the
inscriptions to reconstruct aspects of labour organisation and political control during the production of
the weapons, the work concentrates on documenting and explaining the techniques employed to
produce the inscriptions, to file casting imperfections, and to obtain the smooth, shiny and sharp finish
still noticeable on many of the weapons. Silicone rubber impressions of surface features of swords,
lances, crossbow triggers and arrows were examined under the scanning electron microscope (SEM), and
these observations were supplemented with examination of the artefacts under the stereomicroscope
and the SEM. The evidence indicates the use of a variety of chisels for making the inscriptions, and of files
for removing excess metal from surfaces. In addition, the grinding and polishing marks demonstrate the
large-scale, systematic use of rotary wheels to achieve an ideal final polish. These findings are con-
textualised in the broader history of Chinese metallurgy, with special attention to the emergence of iron
tools and of rotary mechanical devices.
Ó2010 Elsevier Ltd. All rights reserved.
1. Introduction
Qin Shihuang (259e210BC) is one of the most famous and
controversial figures in Chinese history. Widely known as ‘the
First Emperor’, he forcibly unified China under his power, prose-
cuted intellectuals and opponents, abolished feudalism, and
standardised philosophy, script and law. The Herculean mauso-
leum complex he commissioned for himself, an analogue of his
life, has remained as the material representation of his world,
providing an almost inexhaustible source of information about
the powerful political and symbolic system orchestrated around
his personality (Rawson, 2007). Among numerous other installa-
tions and constructions within the mausoleum complex, the
Terracotta Army has acquired an almost iconic status in modern
popular culture, as a manifestation of the exorbitant power, lavish
wealth and artistic achievement of the Qin Empire. Stationed in
three pits to the east of his tomb, the terracotta warriors are
supposed to have been placed there to host and protect the
Emperor in his afterlife. So far, approximately 200 0 individually
crafted warriors have been recovered during archaeological
excavations, although it is estimated that their number may reach
up to 800 0 (Institute and Museum, 1988; Yuan,1990; Portal, 2007;
Guo and Lindesay, 1998).
While previous research has investigated the technology and
artistic achievement behind the production of the warriors them-
selves (Yuan, 1990; Ledderose, 2000), comparatively little work has
addressed the bronze weapons recovered in the army pit. The
excavated assemblage includes over 40,000 arrowheads as well as
hundreds of crossbow triggers, swords, lances, spears, halberds,
hooks, honur weapons (Su) and the ferrules that were fixed at the
end of wooden hafts. As a part of a collaborative agreement between
the Emperor Qin Shihuang’s Terracotta Army Museum and the UCL
Institute of Archaeology, the research focushas now shifted to these
bronze weapons. A major aim of this project is to investigate
patterns of standardisation and labour organisation within this
single, very large and intentional assemblage as well as to consider
the nature of bronze production during the Qin period (475e206BC,
i.e. including the Qin Kingdom during the Warring States era e
475e221BC eand the Qin Dynasty after the unification e
221e206BC). The work is involving an exhaustive typological
analysis combined with extensive measurements and archaeo-
metric studies, integrated with a spatial analysis of the data.
*Corresponding author. UCL Institute of Archaeology, 31-34 Gordon Square,
WC1H 0PY, UK
E-mail address: Xiuzhen.Li@ucl.ac.uk (X.J. Li).
Contents lists available at ScienceDirect
Journal of Archaeological Science
journal homepage: http://www.elsevier.com/locate/jas
0305-4403/$ esee front matter Ó2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jas.2010.09.012
Journal of Archaeological Science 38 (2011) 492e501
This paper concentrates on a relatively inconspicuous yet
important dimension of the weapons’study: inscriptions, filing,
grinding and polishing marks identified on many of their surfaces.
Scratches and marks visible on some of the artefacts suggested that
grinding or polishing took place after casting. Furthermore, some of
the weapons display incised inscriptions, which vary from small
symbols to relatively long sentences referring to the workshop or
workers involved in their production (Fig.1). One specific objective
of this study was to investigate the techniques employed to engrave
and polish bronze during the Qin period, including the shapes and
materials of the tools, and their modes of utilisation. Particular
attention was paid to the potential identification of mechanical
devices, including rotary tools. Furthermore, combining these
technical features with the information recorded in the inscrip-
tions, we sought to reconstruct aspects of workshops’organisation
and the distribution of labour during the construction of the
mausoleum.
2. Materials and methods
The bulk of this study has concentrated on Pit 1, where the
majority of the weapons have been recovered. Most of these
weapons were excavated during 1970s, although their recovery has
continued during subsequent excavations. Except for a few items
displayed on exhibitions, the majority of these are now housed at
the Conservation Department of the Museum of Emperor Qin Shi-
huang’s Terracotta Army. An initial typological analysis, including
detailed measurements and spatial analyses of the distribution of
the weapons in the pit, was carried out. Swords, lances, arrows and
triggers were taken out of their storage, and macroscopic features
of their inscriptions, grinding and polishing marks were noted.
A good number of these weapons were examined and photo-
graphed under a stereomicroscope in order to identify areas of
interest. Subsequently, a subset of these, so far including pieces of
twenty triggers, two swords, two lances and six arrowheads, were
selected for more detailed analysis under the scanning electron
microscope (SEM) at the Wolfson Archaeological Science Labora-
tories of the UCL Institute of Archaeology.
In order to view the characteristics of the inscribed features,
grinding or polishing marks on the surface of the weapons under the
SEM, over fifty detailed impressions of appropriate areas were made
with silicone moulds. The selected area was cleaned with ethanol
solution using a small piece of cloth. The moulding material was
automatically dispensed from a gun fitted with a mixing tip together
with a finely-pointed tip (Sax et al., 2004). Because the silicone dries
quickly, a line of samples was arranged so that the moulding
materials were applied as quickly as possible to the surface of the
impressions. Delay would cause the silicone to dry and adhereto the
gun fitting. The advantages of this method have been demonstrated
by Sax et al. (1995, 1998, 2000, 2004) in their research on jade and
other materials: firstly, it avoids the need either to apply a conduc-
tive gold coating to the artefacts (notnecessary now if using variable
pressure SEM) or to subject them to the high vacuum of the SEM;
secondly, it allows the examination of deeply carved parts of an
object that may be difficult to view directly, as moulds of carved
lines are seen as positive features that show details very clearly.
Furthermore, moulds can be made of areas of large objects that
cannot fit into an SEM. A major advantage of this method is that it
allowed the study of a relatively large number of samples at the
laboratory in London without having to remove the actual artefacts
from China.
The silicone impressions were mounted on 51 mm diameter
aluminium stubs, prior to being coated with gold for SEM obser-
vation. This was carried out in Hitachi S-3400N SEM, at high
vacuum, using a large area 5 segments solid state backscattered
electron detector for imaging. This imaging mode gave the best
contrast and visual display of the depth of engraved features
compared with conventional secondary electron imaging. The
moulds were oriented and tilted in the SEM to give the best angles
of observation, using detector segments A, B, C, D and E to create
shadows and allow the photomicrographic records to represent
accurately the geometric shapes and topographic features. It is
important to note that the moulds record the carved features in
reverse: features which appear as protrusions in the SEM images of
the moulds represent depressions in the actual artefacts. This
facilitates the observation of the depth and detail of engraved
features.
Only a small number of actual artefact fragments, namely
a sword and an arrowhead tang, were analysed directly under the
SEM. The instrument employed in this case was a Philips XL30.
3. Terminology
In order to facilitate the characterisation of the marks identified
in the weapons and the tools involved in their manufacture, it is
appropriate to clarify the terminology to be employed. Here we use
the terms as defined by Untracht (1969) and employed in modern
metal crafts.
3.1. Filing, grinding and polishing
Filing, grinding and polishing are different techniques used in
metal production and design, but sometimes they have overlapping
definitions and functions. All of these methods are used to remove
a certain amount of the metal surface to enhance its appearance
and function, but the tools used for these different procedures are
all slightly different. Filing is the act of using a file to shape or
smooth an object. A file is a long, narrow tool with sharp ridges or
points on its surfaces, typically used in an abrading motion. Designs
can be made on thick metal objects by first removing large amounts
of the surface using heavy files, and then refining the shape using
small files, such as a needle file. The file needs to be harder than the
material being processed.
Nowadays, most grinding is done on grinding wheels mounted
on a frame attached to a grinding or buffing motor. These remove
a relatively small amount of metal from the surface (although
sometimes a coarse grinding wheel can remove a lot quickly form
the surface). Three natural abrasives are used in wheels: corundum
(aluminium oxide), emery (aluminium oxide plus abrasive iron
oxide) and diamond, the hardest known substance. Hand grinding
is still used in modern times as well, moving an abrasive paper back
and forth over the metal surface. This is known to be a long and
Fig. 1. The inscriptions on a halberd.
X.J. Li et al. / Journal of Archaeological Science 38 (2011) 492e501 493
tedious process, but in certain cases it may allow better control of
the operation (Untracht, 1969).
Polishing involves very little removal of metal, and its main
purpose is to bring the condition of the metal to its final surface
appearance. The abrasive polishing material can be applied by hand
with a cloth or linen, or by the use of various types of buffing
wheels mounted on a buffing motor. The technique chosen and the
materials used depend on many factors, among which is the orig-
inal condition of the surface of the piece in relation to the desired
surface, and whether or not the object is plain or decorated. The
selection of the method of polishing, whether by hand or by
machine, typically depends on these factors. If a machine is used,
then the speed of the motor, the size and the material of the buffing
wheel, altogether are variable and affect the final outcome
(Untracht, 1969).
3.2. Carving, engraving, and chiselling
Carving is a process where specific parts from hard materials are
removed to create a desired pattern or shape. Engraving is a process
of incising lines employing a thin, ‘V’shaped pointed tool called
a graver or burin. This can be done completely by hand, partly with
the aid of mechanical devices, or entirely by machine (Untracht,
1969). Chiselling uses sharp, wedge-like tools to cut into and chip
away the metal. Each method produces a different result, and will
be selected by the craftsperson depending on the patterns they
want to make and the hardness of the metal they are working on.
4. Inscriptions on the weapons
4.1. Chronology and production organisation
Not all the weapons of the terracotta warriors display inscrip-
tions. Lances and halberds bear long sentence inscriptions, while
the swords, triggers, hooks, and ferrules were only partially marked
with numbers, a note of the Gong or Sigong (the name of the
workshop) and/or other symbols. To date, no inscriptions have been
found on the surface of the arrows.
During the partial excavation in Pit 1, 16 lances and 4 halberds
were unearthed, and one more lance has been discovered in the
continuing excavation of the pit in recent years. The long inscrip-
tions on these weapons indicate the regnal year when they were
produced, the name of the person in charge of production, the
official or workshop, and the name of the specific worker, thus
providing basic information on both the chronology and the orga-
nisation of production during their manufacture (Institute and
Museum, 1988; Yuan, 1990). Based on this information, it is
possible to establish that the inscriptions on the lances and
halberds date from 244 to 228BC, that is, before the unification of
the Qin Empire, which took place in 221BC (in the twenty-sixth
year of the reign of Yingzheng, King of Qin). Interestingly, the times
for their production do not overlap, as the diagram demonstrates
(Fig. 2).
The remaining information on the inscriptions denotes a form of
accountability and hence of quality control. The inscriptions on the
halberds confirm that there was a supervisor called Lu Buwei (吕不
韦), officials (Sigong,
寺工
), craftsmen (Cheng,
丞
) and workers
(Gong,
工
) involved in bronze production (Institute and Museum,
1988)(Fig. 3). The supervisor, Lu Buwei, was the Prime Minister
of the Qin kingdom before the unification. According to the written
documents of the Qin period, official Sigong were in charge of the
production of bronze weapons and ritual bronze vessels in the
governmental workshop. The craftspeople involved were very
skillful workers, responsible for obtaining raw materials, training
the workers in the manufacturing process and technology, moni-
toring the quality of the bronze weapons, and reporting to the
officials. The workers were the actual producers of the weapons,
and were usually slaves, convicts or soldiers (Yuan, 1984). Unfor-
tunately, there is no clue on the inscriptions as tothe location of the
workshops.
The inscriptions on the lances and halberds also offer some hints
of the political influence in this mass production of the bronze
weapons. One of the obvious differences between themaking of the
halberds and the lances is the supervisor. The Prime Minister, Lu
Buwei, appears as the supervisor of the halberds production (Fig. 3),
but not of the lances (Fig. 4), as he presided over the Qin State when
the First Emperor was young, but was condemned for his connec-
tion with a rebellion and dismissed in 237BC. All the halberds were
produced before that date, and the lances were produced after it.
The fact that the Prime Minister was inscribed as supervisor on the
halberds indicates the strong degree of political control over the
production of weapons. However, the new Prime Minister, Li Si,
does not appear as supervisor in the later lances, suggesting
changes in the production organisation connected with the polit-
ical changes during the Qin era (Yuan, 1984).
Compared to the long inscriptions on the lances and halberds,
the triggers’inscriptions are relatively simple. Approximately 150
out of the 229 triggers examined have inscriptions, including
numbers, symbols and Gong. The function of these inscriptions has
been interpreted as relating to quality control (Yuan, 1984;
Ledderose, 2000). However, the high diversity of symbols noted
(with almost no repetitions between triggers), together with the
fact that some triggers bear the same number or symbol in two or
all three mechanical parts, indicates that they may be related to
their assembly. Among these inscribed triggers, the Gong character
(
工
) has been found on 20 pieces.
Fig. 2. Chronology of the bronze weapon production, based on the regnal years noted in the weapons’inscriptions and their relationship with major events during the Qin period.
The Prime Ministers are noted at the bottom.
X.J. Li et al. / Journal of Archaeological Science 38 (2011) 492e501494
Ongoing work combining the inscriptions with typological,
compositional and spatial analysis will show more insight about
these bronze weapons, which will be discussed in a separate paper.
The following sections will focus on the technical examination.
4.2. Tool marks emanufacturing traits
Fig. 5, obtained under the stereomicroscope, shows clear tool
marks on the inscription of a lance. The three vertical strokes show
the overlapping marks forming them, each of them with the overall
shape of an elongated triangle. It can be interpreted that the line
was chiselled step by step, with a finely-pointed tool ending on
a wedge shaped like the letter ‘V’. The way in which the marks
overlap indicates that the chisels were hammered into the bronze
sliding at an angle, starting from the bottom of the cone shape. In
other words, the base of the triangle was the starting point of the
cut, and the sharp end of the mark was the finishing point. These
observations are supported by our limited experimental work in
the laboratory. The horizontal strokes are relatively smooth and
thinner, and chisel marks are not so obvious. However, both lines
are still rough along the edges, slightly curved in overall design, and
sharp at one end and round at the other end eall features typical of
a hand chiselling technique but using, perhaps, a slightly different
chisel.
Turning to the triggers, Fig. 6 shows three SEM micrographs of
inscriptions on triggers. As already noted, the SEM micrographs,
Fig. 4. The organisation structure during production of the bronze lances, based on
information from the inscriptions. The numbers at the bottom denote regnal years
(15th year being 232BC). Note the absence of both Cheng and supervisor.
Fig. 5. Inscription on the No. 860 lance under stereomicroscope, showing the chisel-
ling technique. This area shown is a part of character “Sigong”(寺工).
Fig. 3. The organisation structure during production of the bronze halberds, based on information from the inscriptions. The numbers at the bottom denote regnal years (3rd year
being 244BC).
X.J. Li et al. / Journal of Archaeological Science 38 (2011) 492e501 495
taken from silicone impressions, show the inscription in reverse, so
that the positive features in the micrographs represent the negative
grooved features on the weapons. These images show relatively
clear tool marks, which are also interpreted as chiselling. The long
lines are formed by individual overlapping marks, resulting in
ragged edges. Like the chisels employed on the halberd, the tool
here would have been pointed, ending in a ‘V’shape.
However, the bottom part of each mark appears more rounded
and, particularly, the back side of each impression (the shorteredge
of the triangular shape) is different too eappearing straight or
rounded on the trigger, while it forms an angle in the halberd.
These small differences suggest the use of slightly different tools
and, most probably, the involvement of more than one person in
this chiselling activity for the variety of weapons.
Some of the inscriptions and the surface finishing marks (dis-
cussed in the next section) overlap, and from this a sequence of the
processes can be interpreted. Some grinding marks were found
beneath the inscriptions and, in other cases, inscriptions break
through the grinding traces. This suggests that the filing, grinding
and polishing were applied to the weapons at the first stage after
casting, and the inscriptions were added thereafter.
4.3. Discussion
Based on the above observations, it is possible to establish the
nature of the inscribing technique on the weapons, namely chis-
elling. Although carving and engraving are terms loosely used to
describe techniques for inscribing bronze, and indeed they are used
frequently when describing these weapons, the correct term should
be chiselling.
The chisels would have to be made of a material both harder and
tougher than the bronze weapons, to provide the sharp marks
without becoming blunt after a few strokes. While stones can be
harder than bronze, they are relatively brittle and thus not suitable
for chiselling. It seems more likely that either high tin bronze, cast
iron or quenched steel were used. During the Warring States period
(475-221BC), iron tools and implements were already employed,
and indeed some iron weapons, such as one iron spear and three
iron arrows, have been unearthed fromthe pits of the Qin terracotta
warriors (Institute and Museum, 1988: 249), as well as some iron
implements discovered from the Emperor Qin Shihuang’s mauso-
leum complex itself (Yuan, 2002). Although the chisels and other
ferrous tools found in the mausoleum complex are too big to have
been used on these weapons, their presence illustrate that the
technology and tools were familiar to them. Unfortunately, due to
lacking analytical data it is not possible to determine whether they
are bloomery or wrought iron, or their carbon content. Evidence of
quenched steel and cast iron has been discovered in many
contemporary archaeological sites in central China (Wagner, 2008:
98e128; Han and Ke, 2007), and some even show the earlier iron
production activities, which can be traced back to the 8th century
BC, for example, the iron farming implements and weapons
discovered from the Jin state (one of the seven states during the
Warring State era, which is located in today’s Shanxi province) (Han
and Duan, 2009: 99). A cast iron or a quenched steel chisel would
have been hard and tough enough to engrave the bronze, and the
evidence available within the Qin mausoleum complex suggests
this should have been available to Qin craftspeople.
The presence of symbols or texts on the surface of Chinese
bronzes is widely documented in Shang (1600e1100BC) and Zhou
(1100e221BC) bronzes. However, the predominant methods of
application in the earlier periods concerned the design of the
moulds, either by adding thin coils or strips of clay that would
appear as depressions in the artefacts cast, or by carving the
inscriptions directly on the clay mould ehence achieving inscrip-
tions in relief (Ma, 1986; Li, 1980). Technically different from the
cast inscriptions, the use of chisels directly onto the surface of the
bronzes, as documented in the First Emperor’sweapons, becomes
common only during the Warring States period (Dong, 2006). Such
a technical change may be related to both sociological and tech-
nological developments. In particular, it is tempting to think that
the chiselling technique is associated to the availability of cast iron
Fig. 6. Micrographs of inscriptions on 976B (a, b) and 974A (c) showing the chiselling
techniques. Grinding marks are also noticeable on the surfaces (silicone impressions).
X.J. Li et al. / Journal of Archaeological Science 38 (2011) 492e501496
and steel tools. Further archaeometallurgical and experimental
research should focus on the relationship between the casting and
chiselling of inscriptions, and the development and use of cast iron
and steel tools.
5. Filing, grinding and polishing marks
Filing, grinding and polishing marks were discovered on a large
number of the weapons unearthed from the tomb complex. In some
cases, these appear rather coarse and easily discernable with the
naked eye. In other cases, the marks are much finer, and only
perceptible thanks to the generally exceptional state of preserva-
tion of most of these artefacts. Crossbow triggers, lances, swords,
and arrows were examined for signs of these.
5.1. Filing marks
Filing and grinding marks are discovered on the composite
bronze crossbow triggers, and mainly found on the overlapping
areas between them. Each trigger comprises three mechanical parts
joined by two bolts, and each part has its own shape and function
(Fig. 7a). The five parts were all cast separately, assembled together
and fixed into the crossbow. The introduction of the trigger
mechanisms during the Warring States period revolutionised
military warfare, as it ensured the crossbow required less skill and
strength to use than a composite bow and was said to be able to fire
heavier arrows over a longer distance (Yuan, 1990; Yang,1980). The
majority of the grinding lines generally surround the holes of the
three mechanical parts, i.e. they are most prominent in the over-
lapping areas between the three parts, where they would have to
move by sliding over each other. These traces are probably the
results of filing off excess metal from the surfaces to make them fit
together properly during assembly, to ensure an effective func-
tioning of the crossbow. However, similar traces have also been
found on some of the bolt heads, where they would serve no
obvious technical purpose (Fig. 7b).
Visually, the orientation of the marks on the triggers may appear
parallel. However, under the SEM, it is possible to see that these
marks are typically multi-directional, and that they cross each
other (Fig. 8). These features are diagnostic of filing by hand, which
involves back and forth movements of the file on the surface of the
object. Based on measurements of the individual grooves, the
diameter of the ridges or points on the file would have been
approximately 100e200
m
m.
Although we lack direct evidence for filing tools employed
during the Qin period, the relative depth and the sub-parallel
arrangement of the filing marks suggest that hand-held files made
of a hard material, such as stone, steel or cast iron could have been
employed. Experimenting with a hand-held polishing stone in
a back-and-forth motion on a brass surface, we produced sub-
parallel filing marks that, albeit thinner, resembled the arrange-
ment of the archaeological ones (Fig. 9). However, considering the
better definition of the marks in the archaeological examples, we
are inclined to think that the filing would have been carried out
with metallic tools. Only future experimentation may be able to
clarify this point conclusively.
Similar marks from filing or a relatively coarse grinding are also
noticeable on the tangs of arrowheads. They typically appear
broadly parallel to each other and oriented diagonally. Although
they appear predominantly on the casting seams, seemingly trying
to remove imperfections, in some cases the marks embrace the
entire circumference of the tangs. Under the SEM, the marks appear
quite similar to those identified on triggers, hence suggesting
a similar filing tool (Fig. 10). However, the fact that the filing marks
sometimes turn around the curvature of the tangs suggests that
either the files were curved or, alternatively, the tangs were
mounted on a lathe to facilitate the filing.
Afinal explanation for the massive effort invested in filing the
surfaces of several tens of thousands of arrow tangs has not been
reached. It is important to note that the arrows would not have
moving mechanical parts, hence the need of such filed or ground
surfaces would seem less necessary. Furthermore, based on better
preserved examples, it is known that these tangs were fitted inside
a longer bamboo stick and would therefore not be visible ethus, it
is unlikely that the filing responded to aesthetic concerns. The most
Fig. 7. Trigger mechanism (a) and detail of grinding marks and inscriptions (b).
X.J. Li et al. / Journal of Archaeological Science 38 (2011) 492e501 497
plausible explanation for the filing on the tangs may be that their
makers wanted to achieve perfectly symmetrical arrows whose
trajectory in the air would not be unbalanced by such small
imperfections.
5.2. Grinding and polishing marks
Much finer grinding or polishing marks are found all over the
surfaces of swords and lances, as well as on the arrowheads proper.
In these cases, it is assumed that the aim of the polish was to obtain
a smooth and shiny surface, as well as a sharp blade. On the swords
and lances, the marks are generally longitudinal on the spine
section, and transversal on the blade section (Fig. 11). On the arrow
surfaces, these marks are transversal on all sides, and they often
continue from one plane to another (Fig. 12). In some cases, it is
possible to discern two stages of fine grinding/polishing on the
arrowhead surfaces, one finer than the other, carried out in orien-
tations perpendicular to each other. Sometimes these finer marks
extend onto the surface of the arrow tangs, removing the evidence
of the coarser filing described in the previous section.
When examined under the SEM, the grinding marks on all three
types of artefacts are remarkable in two aspects: on the one hand,
their extreme fineness and high density; on the other hand, their
perfectly parallel disposition (Fig. 13). Such characteristics leave
little doubt that they originate from a mechanical grinding device,
most likely a rotary tool. The comparative shallow archaeological
marks suggest that only a very small amount of material was
Fig. 9. Micrograph of filing marks experimentally produced on a brass surface using
a hand-held stone. Compare to Fig. 8, but note the different scales.
Fig. 10. Micrograph of filing marks on the tang of an arrowhead.
Fig. 11. Grinding marks on a sword, as seen under the stereomicroscope. Note the
different orientation of the fine grinding and polishing marks on the various planes of
the blade.
Fig. 8. (a) and (b): Micrographs of grinding marks on trigger 912C (silicone
impression).
X.J. Li et al. / Journal of Archaeological Science 38 (2011) 492e501498
removed from the weapons’surfaces during this finishing stage.
These technical interpretations are supported by our experimental
replications created by grinding a brass surface using a small hard
stone rotary wheel (Fig. 14).
The fine grinding and polishing marks on the surface of the
lances, swords and arrows seem not to have been made by loose
abrasive materials. A relatively hard stone was most probably the
material used for this finishing process, since natural abrasive
materials were easy to acquire and hard enough for grinding the
bronze weapons. It is also possible that relatively soft organic
materials, such as linen, could have been employed for the last and
finest stage, which would furnish the weapons with their charac-
teristic sheen. The tools employed for this purpose could be
comparable to the bristle brushes or buffing wheels used in modern
crafts. The softness and flexibility of these materials would better
explain the continuity of polishing marks overlapping different
planes of the same weapons, and their very presence in small,
hardly accessible, intricate parts of the arrowheads. When we used
a bristle brush on a brass surface, however, we obtained a pattern of
marks that does not resemble exactly those in the weapons
(Fig. 15). Further research will have to concentrate on the potential
differences between grinding and polishing marks, and the possible
materials and tools used during these processes.
Another interesting finding of this work is the noteworthy
similarity among the grinding marks detected on the different
Fig. 12. Detail of the grinding marks on an arrowhead, as seen under the stereomi-
croscope. Note the presence of grinding lines that overlap two different planes.
Fig. 13. SEM photomicrograph of the surface of a sword, showing dense, shallow and
perfectly parallel grinding marks. The bright crusts and grains covering some of the
marks are corrosion products.
Fig. 14. SEM photomicrograph of a brass surface experimentally ground using a rotary
wheel made of stone.
Fig. 15. SEM photomicrograph of a brass surface experimentally polished using
a rotary bristle brush.
X.J. Li et al. / Journal of Archaeological Science 38 (2011) 492e501 499
artefacts. This similarity applies to the thickness, density and
relative shallowness of the marks (Fig. 16). Although the exact dates
for the manufacture of the swords and arrowheads are unknown,
the inscriptions on the lances indicated a narrow chronological
range of 232e228BC (see above). The technical similarity of the
grinding marks on all three artefact categories may be taken as an
indication that all the weapons could have been ground in the same
period and possibly at the same workshop, or at least employing
very similar techniques. Future comparative examination of
grinding marks on the halberds, dated to a different moment of Qin
Shihuang’s reign, may confirm whether or not the same grinding
technique was employed over several years, or whether indeed
arrowheads, lances and swords may be deemed as contemporary.
5.3. Discussion
Anecdotal visual evidence on Shang and Zhou bronze vessels
indicates that they were subject to filing and grinding finishing
techniques. However, while their casting technology is abundantly
discussed in the literature (e.g. Liu, 2009; Han and Ke, 2007; Ma,
1986; Barnard, 1961), these finishing marks are rarely studied,
which makes a detailed history of this technology difficult to
reconstruct. In particular, until further analytical and experimental
data are available, it may not be possible to ascertain whether iron
or steel files could have been employed on these bronze weapons.
But these marks are almost certainly not made by loose abrasive
materials.
Of particular importance, however, is the positive identification
of the use of rotary mechanical tools on a systematic and industrial
scale, for the grinding of tens of thousands of bronze weapons.
Rotary wheels were used in jade carving long before they were
used on bronzes (Dong, 2006:104;Sax et al., 2007: 25). Rotary
abrasive methods were probably used in jade carving by the end of
Neolithic period in the Hongshan culture, which dates from about
5000BC (Dong, 2006: 104). Evidence of rotary incising wheels has
also been found on a Shang dynasty ceremonial jade blade now in
the British Museum (Sax et al., 2007: 25). However, during the
subsequent Zhou, Qin and Han dynasties, hand-held tools
continued to be preferred to make designs that are often intricate
and minute. It would appear that it was not until the Tang dynasty
(AD 618e907) that rotary incising wheels become commonplace in
jade carving (Sax et al., 2007: 25).
As regards bronzes, it appears that rotary wheels were used on
metals as early as the Shang and Zhou dynasties (Dong, 2006:
44e45). Bronze masks unearthed from Guanghan Sanxingdui (广汉
三星堆), and dating from the Shang dynasty, were cut with square
holes on the forehead or on the cheek side (Chen, 2000:48e49).
One of the big masks from pit 2, measuring about 65 cm in height
and 138 cm in width, with a 10 cm long rectangular hole on the
forehead, is thought to have been cut using a rotary wheel, because
the four corners of the hole display intrusive crossing lines typical
of wheel cutting (Dong, 2006:44e45). This would constitute the
earliest example of the use of rotary mechanical means discovered
so far on Chinese bronzes.
The rotary incising wheel is supposed to be slightly different
from the rotary grinding wheels, even though they were both
mechanical devices. The wheel shape and the results intended to be
achieved would be different, but the rotary incising wheels and
grinding wheels both involved the same mechanical principles.
Rotary grinding wheels appeared to have also been used on
bronze weapons from the Warring States era based on the obser-
vations of the surface traces. Dong (1999) has argued that the
Yuewang Goujian (越王勾践) sword, reputed to have been owned
by the king of the Yue state, had its blades probably trimmed and
ground using rotary wheels, because the lines are all parallel. The
handle of the sword has a circular feature that seems to have been
ground with a lathe, which was probably not balanced (Dong, 1999:
83). Furthermore, two gears from the following Han dynasty
(202BCe212AD) show a similar mark, suggestive of mechanical
Fig. 16. Details of the silicone casts of the grinding marks on lances, swords and
arrowheads, taken at the same magnification to facilitate comparison. From top to
bottom, (a) lance 860, (b) sword 854 and (c) arrow 2118 (the bright crusts on the latter
are impressions of corrosion products). Compare to coarser filing marks on a trigger in
Fig. 7(b), taken at the same magnification.
X.J. Li et al. / Journal of Archaeological Science 38 (2011) 492e501500
means of manufacture (Dong, 1999: 84). The weapons from the
Terracotta Army constitute the earliest evidence of the use of such
rotary means in mass production.
6. Conclusion
Examination of the inscriptions and other tool marks on the
surfaces of bronze weapons of the Terracotta Army has increased
our understanding of the organisation of production during the
creation of this unique assemblage, while contributing technical
insights into the history of the bronze making craft and its adap-
tation to Qin Shihuang’s commission.
As highlighted in previous work, the textual information
recorded on the inscriptions of halberds and lances allows us to
deduce a chronology for their fabrication, indicating that they were
produced in two different stages, both predating the unification of
the Qin (Qin Kingdom) and Qin Shihuang’s ascent to the throne as
an emperor (Qin Empire). In addition, the various names of
supervisors, workshops, officers and workers, as recorded in the
inscriptions, allow inferences as to the craft organisation and the
relationship between weapon production and the changing polit-
ical situations. Future work, comparing this and other assemblages,
will facilitate an understanding of the singularity of the industry
that worked directly for the mausoleum.
The tool marks in the inscriptions reveal the use of a variety of
hand-held chisels to individually mark the weapons. Besides noting
aspects related to quality control, it is suggested that the individ-
ualised symbols on the crossbow triggers may have been added to
guide the assembly of joining parts, given the repetition of the
symbols on several parts of the same triggers. In this sense, the fact
that the inscriptions took place after the filing and grinding of the
surfaces indicates that triggers were only inscribed once their
makers, or their supervisors, were satisfied with their quality and
the possibility of an effective assembly of the various parts.
Excess material such as casting seams or surface imperfections
were removed from both triggers and arrow tangs employing files
that were used for manually and painstakingly abrading their
surfaces. Although the material employed for chisels and files
remains uncertain, we suggest that at least some of these tools
could have been made of cast iron or steel, since their extensive use,
as documented here, broadly coincides with the development of
ferrous metallurgy in China.
Finally, the marks left by the grinding that ensured that the
surfaces of all the bladed weapons remained lustrous, and their
blades sharp, demonstrate the use of rotary wheels on an unprec-
edented scale.
Altogether, the sophistication and sheer amount of specialised
labour invested in just the finishing touches of these bronze
weapons further highlights that no resources or skills were spared
in the production of the First Emperor’s Mausoleum. Without
a doubt, further studies of this unique archaeological site will
continue to yield surprises.
Acknowledgements
We are very grateful to the Museum of the Emperor Qin Shi-
huang’s Terracotta Army and the UCL Institute of Archaeology, as
we all benefit from the cooperative aspects of this project, as well as
to the late Professor Peter Ucko, who made this collaboration
possible. Many thanks are due to the Museum Director, Wu Yongqi,
for valuable suggestions regarding the making of these tool marks,
and to Thilo Rehren for his practical advice and constructive
comments. By courtesy of the British Museum, working experience
was gained by Xiuzhen Janice Li using the SEM at the Department of
Conservation and Scientific Research. Thanks are due to Roseleen
Bains for generously sharing her gun, silicone and experience on
mould casting. We would also like to thank Chang Qiuyue, Yan
Hongxia and other colleagues in the Conservation Department of
the Terracotta Army Museum for their support during the exami-
nation of the weapons and the casting of the silicone moulds, as
well as to many other friends who support this project.
References
Barnard, N., 1961. Bronze casting and bronze alloys in ancient China. Monumenta
Serica Monograph 14. Australian National University, Canberra.
Chen, Dean, 20 00. San Xing Dui. Sichuan People Press, Chengdu. 陈德安：I三星
堆J，四川人民出版社，20 00版。
Dong, Yawei, 1999. An investigation into the origin of mechanical means based on
ancient Chinese bronze objects. Journal of Echou University, 1999 Supplement.
董亚巍：从古代青铜文物看中国机械加工的渊源.I鄂州大学学报J1999年增刊。
Dong, Yawei, 2006. Mould Casting Bronzes. Beijing Art and Science Press. 董亚
巍：I范铸青铜J，北京艺术与科学电子出版社，2006版。
Guo, Baofa, Lindesay, W.,1998. The Terra-cotta Army of the First Emperor of China.
The Genius of China. The Guidebook Company, Hong Kong.
Han, Rubin, Duan, Hongmei, 2009. An early iron-using centre in the ancient Jin state
region (8thd3rd century BC). In: Mei, J., Rehren, Th. (Eds.), Metallurgy and
CivilisationdEurasia and Beyond. Archetype Publications, London, pp. 99e106.
Han, Rubin, Ke, Jun, 2007. Mining and Metallurgy: a History of Chinese Science and
Technology. Science Press, Beiing. 韩汝玢柯俊：I中国科学技术史٠矿冶卷J.科学
出版社，20 07版。
Institute and Museum, 1988. Excavation Report on Pit No. of the Terra-cotta
Warriors and Horses (1874e1984). Culture Relics Press, Beijing. 陕西省考古研究
所始皇陵秦俑考古发掘队：I秦始皇陵兵马俑坑一号坑发掘报告（1974-1984）J,
文物出版社，1985版。
Ledderose, L., 2000. Ten Thousand Things: Module and Mass Production in Chinese
Art. Princeton University Press, Princeton.
Xueqin, Li, 1980. The Wonder of Chinese Bronzes. Foreign Languages Press, Beijing.
Liu, Yu, 2009. Origins and evolution of the casting technology of Anyang bronze
ritual vessels: an exploratory survey. In: Mei, J., Rehren, Th. (Eds.), Metallurgy
and CivilisationdEurasia and Beyond. Archetype Publications, London, pp.
55e61.
Chengyuan, Ma, 1986. Ancient Chinese Bronzes. Oxford University Press, Oxford.
Portal, J., 2007. The First Emperor eChina's Terracotta Army. The British Museum
Press, London.
Rawson, J., 2007. The first Emperor's tomb-the Afterlife Universe. In: Ja ne, P. (Ed.),
The First Emperor eChina's Terracotta Army. The Britis h Museum Press,
London, pp. 114e151.
Sax, M., Meeks, N.D., 1995. Methods of engraving Mesopotamian quartz cylinder
seals. Archaeometry 37, 25e36.
Sax, M., McNabb, J., Meeks, N.D., 1998. Methods of engraving Mesopotamian quartz
cylinder seals: experimental confirmation. Archaeometry 40, 1e21.
Sax, M., Meeks, N.D., Collon, D., 2000. The introduction of the lapidary wheel in
Mesopotamia. Antiquity 284, 380e387.
Sax, M., Meeks,N.D., Michaelson,C., Middleton, A.P.,2004. The identificationof carving
techniques on Chinese jade. Journal of Archaeological Science 31, 1413e1428.
Sax, M., Meeks, N.D., Ambers, J., Michaelson, C., 2007. The introduction of rotary
incising wheels for working jade in China. In: Douglas, J.G., Jett, P., Winter, J.
(Eds.), Scientific Research on the Sculptural Arts of AsiadProceedings of the
Third Forbes Symposium at the Freer Gallery of Art. Archetype Publication,
London.
Untracht, O., 1969. Metal Techniques for Craftsmen-a Basic Manual on the Methods
of Forming and Decorating Metals. Robert Hale & Company, London.
Wagner, D.B., 2008. Science and Civilisation in China. Vol. 5: Chemistry and
Chemical Technology. Part 11: Ferrous Metallurgy Cambridge University Press,
Cambridge.
Yang, Hong, 1980. Research on the Ancient Chinese Weapons. Cultural Relics Press,
Beijing. 杨泓:I中国古代兵器论丛J.文物出版社，1980版。
Yuan, Zhongyi, 1984. A summary of the inscriptions on the Qin bronze weapons.
Archaeology and Cultural Relics 5 袁仲一：秦中央督造的兵器刻辞综述.I考古与文
物J，1984年5期。
Yuan, Zhongyi, 1990. Research on the Terra-cotta Warriors and Horses and
Mausoleum of Emperor Qin Shihuang. Cultural Relics Press, Beijing. 袁仲一:I秦
始皇陵兵马俑研究J.文物出版社，1990版。
Yuan,Zhong Yi, 2002. A Researchinto the ArchaeologicalDiscoveriesfrom the Emperor
Qin Shihuang'sTomb Complex.Shaanxi People Press, Xi’an.袁仲一:I秦始皇陵考古
发现与研究J.陕西人民出版社，2002版。
X.J. Li et al. / Journal of Archaeological Science 38 (2011) 492e501 501
