Crossbows and imperial craft organisation: The bronze triggers of China's Terracotta Army

Abstract

The Terracotta Army that protected the tomb of the Chinese emperor Qin Shihuang offers an evocative image of the power and organisation of the Qin armies who unified China through conquest in the third century BC. It also provides evidence for the craft production and administrative control that underpinned the Qin state. Bronze trigger mechanisms are all that remain of crossbows that once equipped certain kinds of warrior in the Terracotta Army. A metrical and spatial analysis of these triggers reveals that they were produced in batches and that these separate batches were thereafter possibly stored in an arsenal, but eventually were transported to the mausoleum to equip groups of terracotta crossbowmen in individual sectors of Pit 1. The trigger evidence for large-scale and highly organised production parallels that also documented for the manufacture of the bronze-tipped arrows and proposed for the terracotta figures themselves.

Full text

Crossbows and imperial craft
organisation: the bronze triggers of
China’s Terracotta Army
Xiuzhen Janice Li1,2, Andrew Bevan2, Marcos Martin´
on-Torres2,Thilo
Rehren2,3, Wei Cao1,YinXia
1& Kun Zhao1
Xi’an
Beijing
0 km 2000
N
The Terracotta Army that protected the
tomb of the Chinese emperor Qin Shihuang
offers an evocative image of the power and
organisation of the Qin armies who unified
China through conquest in the third century
BC. It also provides evidence for the craft
production and administrative control that
underpinned the Qin state. Bronze trigger
mechanisms are all that remain of crossbows
that once equipped certain kinds of warrior in
the Terracotta Army. A metrical and spatial
analysis of these triggers reveals that they were
produced in batches and that these separate
batches were thereafter possibly stored in an
arsenal, but eventually were transported to
the mausoleum to equip groups of terracotta crossbowmen in individual sectors of Pit 1. The
trigger evidence for large-scale and highly organised production parallels that also documented for
the manufacture of the bronze-tipped arrows and proposed for the terracotta figures themselves.
Keywords: China, Xi’an, third century BC, Qin period, Terracotta Army, standardisation,
logistics, technology, spatial analysis
Introduction
Since their rediscovery by Chinese archaeologists in the 1970s, the tomb complex containing
the Qin terracotta warriors of Xi’an has become one of the world’s most impressive
archaeological sites. An army many thousand strong, the ceramic warriors were originally
equipped with fully functional bronze weapons and were stationed at the eastern end of the
mausoleum of China’s first emperor, Qin Shihuangdi (Figure 1) (259–210 BC) (SIAATQ
1988; Yuan 1990, 2002; Wang 1994; Ledderose 2000; Portal 2007; Duan 2011). To date,
1Emperor Qin Shihuang’s Mausoleum Site Museum, Lintong, Xi’an, 710600, China
2UCL Institute of Archaeology, 31–34 Gordon Square, London WC1H 0PY, UK
3UCL Qatar, a partner of Hamad bin Khalifa University, PO Box 25256, Doha, Qatar
C
Antiquity Publications Ltd.
ANTIQUITY 88 (2014): 126–140 http://antiquity.ac.uk/ant/088/ant0880126.htm
126

Research
Xiuzhen Janice Li et al.
Figure 1. a) Plan of the mausoleum complex, with the imperial tomb towards the centre and the Terracotta Army pits
(including Pit 1) to the east; b) view looking south over warriors in Pit 1 (image courtesy of Xia Juxian).
C
Antiquity Publications Ltd.
127

Crossbows and imperial craft organisation
more than 40 000 arrowheads and several hundred other bronze weapons have been found
associated with the warriors in three different pits within the tomb complex. Only parts of
these weapons have survived to the present day, but this remarkable assemblage includes
crossbow triggers, sword blades, lance tips, spearheads, dagger-axe blades and halberds, as
well as a few ceremonial weapons (Qin & Zhang 1983; Yuan 1984; Liu 1986; SIAATQ
1988; Huang 1990; MEQSTA 2008). The arrangement of the weapons in the pits reflects
prevailing Qin battle formations and military strategies, and these have been widely discussed
in the existing literature, both in Chinese and English (Yuan & Qin 1975; Wang 1980,
1983, 1990; Yuan 1990, 2002; Yates 2007). Far less has been said about what the bronze
weapons might tell us about craft standardisation, workshop organisation or the logistics of
this monumental funerary project (Yuan et al. 1981). One particular challenge our research
faces is that no production remains have been found, and hence our reconstruction of
production organisation has to be based on the reverse engineering of finished items in their
depositional contexts. The present study focuses on the most extensively explored of the
Terracotta Army pits (Pit 1) and on one particularly intriguing component: the multi-part
triggers that were key devices behind China’s early development of crossbow technology.
Detailed typological and metrical characterisation of these artefacts combined with spatial
analysis allows us to reconstruct possible workshop practices, storage behaviours and the
logistical feat that placed these artefacts in their final positions beside the warriors.
If historical records from later periods are to be believed, the construction of the First
Emperor’s mausoleum was commissioned as soon as he became ruler of the Qin state in
246 BC. Construction took about 40 years and involved 700 000 labourers, who were
conscripted to the Qin capital. It is likely that many of the items in the pit were produced
specifically for the mausoleum: not only the terracotta warriors themselves, but the ceramic
acrobats, officials, musicians and horse keepers, the bronze chariots and birds, as well as the
stone armour (SIAMEQSTA 1998, 2000, 2006, 2007; Yuan 2002).
For the bronze weapons, several options should be considered: a) they could have been
commissioned specifically for the Terracotta Army; b) they could have been made in one
or more weapon-making facilities following ordinary practice for the supply of real armies;
or c) they could have been removed from circulation in order to be deposited in the tomb,
perhaps after a period of use. In any of these cases, the weapons may or may not have spent
some time in storage prior to being transported to the tomb complex and being placed
alongside the warriors.
The style of the triggers and other weapons is different from contemporary examples
found in other parts of China, but comparable to those at other Qin sites (Yang 1980:
206–17). In addition, the ‘regnal years’ inscribed on a small handful of the weapons predate
the ascent of Qin Shihuangdi to the imperial throne (Yuan 1984; SIAATQ 1988: 251; Li
et al. 2011). However, most of the weapons, including the triggers, are not dated, and they
could conceivably have been made later (see below). Our study of filing and other tool
marks on the trigger surfaces showed no obvious wear marks (Li et al. 2011, 2012). The
inscriptions on some of the triggers included the character gong (), probably related to
sigong () or ‘government workshop’ (Yuan 1984; Huang 1990; Jiang & Liu 2011: 248).
Many others are uninscribed. We must also bear in mind the weapons’ post-depositional
life, given the evidence that the warrior pits were partially damaged by later human activities
C
Antiquity Publications Ltd.
128

Research
Xiuzhen Janice Li et al.
Figure 2. Plan of Pit 1 showing the spatial distribution of: left) warriors; and right) weapon types (including crossbow
triggers).
such as burials, looting and agriculture. The challenge in the present study is to extract
information about the structure, tempo and geography of some of these slowly unfolding
activities, with a particular emphasis on craft standardisation, workshop organisation and
the logistics of tomb construction.
Bronze crossbow triggers and the Qin Terracotta Army
Crossbows were complex technical artefacts, closely linked with the military advances of
the emerging Qin Empire. The introduction of trigger-fired crossbows during the Warring
States period (476–221 BC) revolutionised military warfare, as they required less skill and
strength than a composite bow. Crossbows would have enabled soldiers to fire heavier arrows
(i.e. bolts or quarrels) more accurately, with greater force and penetrating power, and over a
longer distance (Yuan 1990: 256; Yates 2007: 42). In the pits of the Terracotta Army, the
wooden and bamboo parts of the crossbows have perished after 2000 years, but fortunately
the bronze triggers have largely survived in good condition (SIAATQ 1988: 280–96). The
vast majority of these come from the extensive excavations in the easternmost area of Pit 1
(SIAATQ 1988: 10) (Figure 2). Some 216 triggers, considered in this paper, were discovered
here, most lying with numerous arrows next to their crossbowmen in the front corridor or
along the two side corridors. During the Qin period, armies consisted of chariots, cavalry,
C
Antiquity Publications Ltd.
129

Crossbows and imperial craft organisation
infantry and crossbowmen. As the above arrangement of crossbow triggers implies, the latter
were normally placed in front or on the flanks of the force to enable them to fire the arrows
over a long distance before the enemy could come close (Yuan 1990: 260; Yates 2007: 43).
Figure 3. Diagram illustrating operation of the bronze
trigger mechanism of a Qin crossbow (image courtesy of
Zhao Zhen).
In addition, a few specialised crossbowmen
were placed in the middle corridors,
especially behind chariots.
As complex devices made of several
components, the bronze crossbow triggers
should be particularly informative about
the processes of standardisation and mass
production employed in the making of Qin
bronze weapons (Figure 3). A trigger was
formed of three parts, held together by two
bolts. They were designed and assembled
in such a way that the string could be held
in place and released effectively without
the need for a spring. We shall refer to
thesepartsasA,B,C,DandE.PartA,
the trigger proper, is a handle. Part B, the
tumbler, involves two functional elements:
one end is used to catch the string of the
crossbow, while the other extends upward
as a ‘regulator’ or sighting pin. Part C, the
rocking lever, was used to link parts A and B so that they could be manipulated effectively.
Parts D and E are the bolts or transverse screws used to strengthen the crossbow stock
and trigger. The trigger was made by casting each part separately in a mould, then filing
the resulting parts down to remove unwanted rough edges, before finally assembling them
as a trigger and inserting them into the wooden body of the crossbow (Williams 2008).
Standardised morphology and dimensions were essential to allow the crossbow to work
properly, and we should expect this concern to be reflected in the way workshops were
organised.
Typology and measurements
One cased trigger (ID 5525), possibly dated to the Han Dynasty (206 BC–AD 220), was
excluded from the study; its presence in Pit 1 remains to be explained (Yang 1980: 218). The
rest of the triggers showed only minor differences in their shapes and styles, as was obvious
from the outset, and from previous discussion (SIAATQ 1988: 293). Our first objective
was therefore to combine macroscopic identification of these differences (where they were
obvious) with precise measurements of three selected dimensions across each trigger part
(to discover less obvious variability) and thereby classify the overall trigger assemblage
into a series of sub-groups (Figure 4). Various multivariate methods were used to explore
these typological and metrical observations, including hierarchical cluster analysis and
principal components analysis (PCA). These different approaches all suggested very similar
C
Antiquity Publications Ltd.
130

Research
Xiuzhen Janice Li et al.
sub-groups, and so for clarity we only present the PCA plots below. The respective members
of the sub-groups are identified by coloured data points in the PCA graphs; even if these
groups are not always statistically clear-cut, their archaeological significance is supported by
Figure 4. Parts A–E of a crossbow trigger indicating the
position of the measurements taken.
the correlations among specific sub-groups
of the various parts assembled together,
discussed later in this paper. The following
sections discuss variability in parts A, B and
C in turn, before considering how these
parts were assembled.
The metrical data for part A allow three
main groups (Ag1–3) to be identified fairly
easily both on a PCA plot (Figure 5a)
and by eye in terms of the degree to
which the part tapers along its length
(Figure 5b).
Part B can be macroscopically divided
into two types depending on whether or
not they have a clearly defined notch
(Figure 6b). The notch protrudes on the
side that holds the string. Its function
remains unexplained, but may be to do with
different strategies for holding or releasing
the string efficiently. Figure 6a plots the first
two principal components of the metrical
data, showing a further rough division into
two groups, Bg1 and Bg2. Since these
metrical groups do not match the macroscopic distinction between notched and un-notched
parts, there is further subdivision into Bg1n, Bg1u, Bg2n and Bg2u, where n and u stand
for notched and un-notched respectively.
Typology indicates that part C can be divided into curved and bevelled versions
(Figure 7b). Thereafter, the first two principal components suggest three possible groups,
one with a subset containing the bevelled examples (Figure 7a). This gives a total of four
metrical groups: Cg1, Cg2, Cg3c and Cg3b, where c and b stand for the curved and bevelled
versions respectively.
Understanding trigger-part groupings
Our overall interpretation of these results is that each trigger-part sub-group was made:
(a) in the same casting mould, and/or (b) in nearly identical casting moulds based on
the same parent model (e.g. the same prototype), and/or (c) from very similar moulds
with a more complicated but linked workshop history. Even if a single parent model was
used as a standard for producing several moulds for the trigger parts, the shrinkage of
mould materials and the subsequent pouring of bronze and filing of the trigger parts by
C
Antiquity Publications Ltd.
131

Crossbows and imperial craft organisation
individual craftspeople would affect the degree of standardisation in the final product. To
assess this within-group variability, a coefficient of variation (CV) was calculated for each set
of individual measurements (Table 1). The results for trigger parts A, B and C all fall within
Figure 5. a) Plot of the first two principal components of
all parts A from Pit 1 revealing three groups: Ag1, Ag2 and
Ag3; b) typological variation of trigger part A.
the 0.6–6.3 per cent range, which
indicates relatively high standardisation
when compared to what we might expect
cross-culturally and theoretically (i.e. with
reference to the Weber fraction; see Eerkens
& Bettinger 2001; Eerkens & Lipo 2005).
It should be noted that the above analyses
identify the minimum number of sub-
groups in the assemblage; it is quite
possible that more moulds existed but
cannot be discerned metrically, given their
similarities. For example, the lengths for
group Ag2 range from 76 to 83mm, and
it is quite likely that more than one mould
is represented here. Furthermore, although
we have assumed discrete groups for the
purposes of this study, the triggers may
exhibit a continuum of variation as a result
of gradual drift in their parent models and
moulds.
It is difficult to know what the
typical turnover was in models and
moulds. The trigger groups from the same
models and moulds should, however, be
linked with single production events (at
intervals that cannot be established) or
manufacturing units in the workshop.
From the information presently available,
changes in model and mould could also
relate to different phases of production.
The production dates inscribed on lances
and halberds from the pits make it clear
that they were made between 244 BC and 228 BC, after the future Emperor Qin
Shihuang became ruler of the Qin state in 246 BC, but before he unified China
in 221 BC (SIAATQ 1988: 265; Yuan 1990, 2002; Li et al. 2011). Emperor Qin
Shihuang died in 210 BC and the bulk of the extensive construction of his tomb
complex has been argued to have been carried out within 10 years of the unification
(Yuan 2002: 12; Duan 2011: 56–58). It is not clear whether the triggers were produced
at the same time as the halberds and lances or somewhat later, but their production
may likewise have extended over several years, given the numbers and variability
involved.
C
Antiquity Publications Ltd.
132

Research
Xiuzhen Janice Li et al.
Trigger assembly and labour organisation
How were parts A, B and C assembled into a working trigger? If production was organised
in a flow-line model (i.e. a large production unit with strong division of labour around a
Figure 6. a) Plot of the first two principal components of all
parts B from Pit 1.The green dots show the notched and the
blue dots the un-notched examples; b) typological variation
of trigger part B.
single assembly line; Groover 2010), we
might expect each part to be cast in large
batches by separate units of craftspeople.
The parts would be assembled at a later
stage of the flow line, perhaps by another set
of workers. In this case, trigger parts cast in
different batches would be mixed together
in the final trigger assembly. Conversely,
in a batch or cellular production model
(Bagley 1995; Li 2006; Groover 2010) a
single set of craftspeople would both cast
and assemble all the required parts, possibly
including the non-metallic components of
the crossbow. Thus a single production unit
could cast a relatively small number of parts
A, B and C, and assemble them before
moving on to produce the next batch. Such
a production model would result in more
homogenous and concentrated patterns in
assembled trigger part groups.
Our search for associations between
trigger parts A, B and C belonging to
the different sub-groups identified above
revealed a surprisingly limited range of
combinations (Table 1). Out of 48 possible
combinations (3 ×4×4 sub-groups of
each part), only eight appear consistently
repeated (assembly groups 1–8), plus a few
mixed-batch groups each represented by
only one or two triggers (assembly groups
9–14).
The limited degree of part mixing
outside those eight combinations is strongly indicative of a batch or cellular production
model, whereby a production unit or cell would cast and assemble a small number of
all the required parts before proceeding to manufacture and assemble the next batch.
It has been proposed that the trigger parts were cast in so-called ‘stack moulds’, i.e.
multi-layered moulds containing several impressions of the same part (Williams 2008).
If so, it is possible that, after casting a ‘stack’ of each of the required parts, weapon
makers would proceed immediately to assemble them. The morphological differences
between the sub-groups of trigger parts would relate to the different production units or
C
Antiquity Publications Ltd.
133

Crossbows and imperial craft organisation
cells, each relatively autonomous albeit working under shared standards and supervisory
structures.
Figure 7. a) Plot of the first two principal components of
measurements from trigger part C, revealing three groups; b)
typological variation of trigger part C.
It is difficult to know whether all the
various parts were totally interchangeable
regardless of sub-group, e.g. whether a
specific Ag3 sub-type of part A could
be fitted to any sub-type of part B
without compromising performance. The
meticulous filing of excess metal on the
various parts (Li et al. 2011), together
with the symbols often inscribed in several
parts of the same assembled trigger (Li
& Gao 2010), perhaps to guide assembly,
indicate that this was probably not the
case. Working in small, carefully controlled
batches and assembling the weapons in
thesameproductionunitbeforeproducing
the next batch would minimise the risk of
mismatches.
Although their diversity and complexity
warrants further attention, an initial
exploration of the marks inscribed on
70 per cent of the triggers (n =
160) corroborates these morphological
groupings. Occasional traces of ink suggest
that more inscriptions may originally have
been present. Engraved marks include
a wide range of individual symbols,
characters and numbers which are probably
related to assembly (and possibly quality control), as they are often repeated on more than one
component of the same trigger (Li et al. 2011). In some cases there is a broad correspondence
between the assembly groups and the types of inscriptions: for example, assembly groups
1 and 2 show no inscriptions; three stem-branch characters ( )arefoundonlyon
parts B of assembly group 4; the character gong () appears predominantly in group 6 (15
cases), with only two exceptions appearing in assembly group 3 (which is very similar in
its component parts). As noted earlier, the character gong is of special interest as it may
denote the involvement of a government workshop or sigong, but we lack direct evidence
to clarify the organisational or chronological relationships between this and other potential
workshops producing the other triggers.
Previous research suggests that similar workshop cells were also responsible for the
production of the terracotta warriors themselves, which we can assume beyond reasonable
doubt were produced specifically for the mausoleum (Yuan 1990: 352–65; Ledderose 2000:
50–73; Yuan & Liu 2009: 6–24). Names of at least 92 different potters were found
C
Antiquity Publications Ltd.
134

Research
Xiuzhen Janice Li et al.
Table 1. Averages (in mm) and coefficients of variation (CV, in %) for the several
dimensions in each sub-group of trigger parts A, B and C.
A1 A2 A3
Part A group n =Mean CV Mean CV Mean CV
Ag1 37 78.5 2.3 15.6 3.5 15.8 2.7
Ag2 167 79.9 1.8 20.9 2.5 18.9 4.4
Ag3 12 81.9 0.6 22.5 1.0 20.2 0.9
B1 B2 B3
Part B group n =Mean CV Mean CV Mean CV
Bg1n 9 82.8 2.7 50.9 3.1 16.0 5.0
Bg1u 43 82.8 2.1 50.3 3.4 16.0 6.3
Bg2n 19 82.2 1.4 53.4 1.8 19.3 3.5
Bg2u 145 82.4 2.3 51.2 5.0 19.1 4.9
C1 C2 C3
Part C group n =Mean CV Mean CV Mean CV
Cg1 132 58.7 2.1 29.2 4.9 25.8 3.0
Cg2 36 63.2 3.5 31.2 3.9 25.3 3.8
Cg3b 18 56.5 1.4 25.5 6.0 21.7 3.4
Cg3c 30 60.0 1.6 27.5 3.6 22.1 5.0
Table 2. Trigger assembly groups and their
constituent parts.
Assembly group Assembly n =
1 Ag1-Bg1n-Cg2 7
2 Ag1-Bg1u-Cg2 23
3 Ag2-Bg1u-Cg1 4
4 Ag2-Bg1u-Cg3c 13
5 Ag2-Bg2n-Cg3b 18
6 Ag2-Bg2u-Cg1 126
7 Ag3-Bg2u-Cg3c 11
8 Ag2-Bg2u-Cg3c 6
9 Ag1-Bg1n-Cg3c 1
10 Ag1-Bg1u-Cg1 1
11 Ag1-Bg1n-Cg1 1
12 Ag1-Bg2u-Cg2 2
13 Ag2-Bg2u-Cg2 2
14 Ag1-Bg2n-Cg2 1
marked on the back of some 1000 terracotta warriors, and each of these individuals
probably supervised several apprentices. It is possible to identify different styles and
characteristics in the figures that were created by different artisans. Many of the more
robust-looking figures, for example, bore the names of artisans from the Qin capital
Xianyang, while others were slimmer in build and had different marks (Yuan & Liu 2009:
22).
C
Antiquity Publications Ltd.
135

Crossbows and imperial craft organisation
Figure 8. a) Spatial distribution of assembly group 2 triggers (combining parts Ag1-Bg1u-Cg2), shown as triangles, within
the five easternmost trenches of Pit 1 in relation to all other triggers (grey dots); b) pairwise correlation of this group (red line)
compared with a 95% critical envelope produced by random Monte Carlo simulations.
Cellular production is also suggested by study of the arrowheads from the Terracotta
Army (Martin´
on-Torres et al. 2012). The extremely high degree of standardisation made
it impossible to identify typological or metrical groups, but chemical analyses showed that
the arrows in each of the 100-arrow bundles form a specific chemical group, most likely an
individual batch deriving from a single crucible load (for chemical batches see also Freestone
et al. 2009a & b, 2010). The low incidence of batch mixing between different bundles
indicates that each unit would produce all of the components and assemble them as a
finished bundle of arrows before moving on to produce the next.
The cellular production of weapons in batches leads us to ask whether the same cells
would have had the versatility to produce the different categories of weapons present in
the assemblage (for example, both triggers and arrows), and indeed whether they also
manufactured the non-metallic components of these weapons (e.g. the crossbow frames or
the arrow shafts). Logistical arguments and the available evidence (Martin´
on-Torres et al.
2012) suggest that both hypotheses are likely to be true. The spatial analyses presented
below add weight to that possibility.
C
Antiquity Publications Ltd.
136

Research
Xiuzhen Janice Li et al.
Spatial analysis and logistical activities
We have established that well-defined batches of triggers were produced by different
groups of craftspeople and/or at different times. After manufacture and assembly, each
Figure 9. Spatial distribution of trigger assembly groups
indicating both small-scale clusters of similar triggers (one
example circled with a dotted line) and larger activity areas
(circled in solid lines). The terracotta warriors are shown as
light grey squares and chariots as grey-outlined boxes.
batch or group of triggers might have been
stored in a specific place in a warehouse
or dedicated arsenal. Alternatively, they
couldhavebeendelivereddirectlytothe
Terracotta Army without any significant
interim storage. The spatial distribution
of triggers in Pit 1 is affected first and
foremost by the position of crossbowmen
in the battle order of the army. Above and
beyond this pattern, however, the spatial
distribution of a given trigger assembly
group can inform about the organisational
processes which governed the placement
of the crossbows in the pit. For example,
if this distribution proved to be regular
(i.e. uniformly spaced) across those limited
zones where triggers of any kind were
found, then this might reflect a form of
intentional arrangement by the workers. In
contrast, if a given weapon assembly group
proved to be spatially clustered in the pit,
we might consider a number of possible
causal factors, but the transportation and
placement of a discrete production batch in
onegointothepitisalikelyexplanation.
In contrast, a random spatial distribution
for a trigger assembly group can be treated
as the statistical null hypothesis, indicating
that no specific constraints dictated the particular location of each trigger in the pit. More
precisely, it might suggest the mixing up of different batches of weapons in storage before
they ended up in the pit.
The spatial patterning was explored by a pair correlation function (PCF; Illian et al. 2008:
218–23; for archaeological examples, see Orton 2004; Bevan et al. 2013) which measures
patterns of clustering or segregation over multiple distances and evaluates the significance of
these patterns via Monte Carlo simulation (see Robert & Casella 2004). Figure 8 shows the
application of this approach to the distribution of assembly group 2 (Ag1-Bg1u-Cg2), in the
fully excavated section of Pit 1 (SIAATQ 1988: 10). The analysis reveals that assembly group
2 is strongly clustered at almost all separation distances up to about 7m and occasionally
so thereafter. This suggests both very localised clustering of a few very similar triggers, and
some wider grouping as well.
C
Antiquity Publications Ltd.
137

Crossbows and imperial craft organisation
Consideration of other trigger assembly groups using the same methods produces similar
results. The localised clusters (Figure 9) probably result from some combination of workshop
production and interim storage and transportation methods. Similar triggers may well have
been cast using the same mould (or conceivably even in the same batch) by one group
of craftspeople, attached to crossbow frames, and stored together in a Qin arsenal before
being transported to the pit together. At a larger scale, the overall distribution of the various
assembly groups across the pit also allows the identification of what may be broad ‘activity
areas’ (Figure 9), revealing how the equipping of the warriors in the pit was organised both
spatially and chronologically.
Interestingly, the easternmost activity areas show much more trigger group diversity than
those in the northern and southernmost corridors. The latter two areas contain all of the
‘gong’ inscriptions: this may imply that the artisans from that workshop had a greater output
capacity, and we may recall that the ‘gong’ inscriptions suggest production in a central
government workshop. For the other activity areas, several cells may have had to work in
unison to produce all the triggers required in a short time so that the construction of the
army could proceed, resulting in a lower level of overall standardisation. Alternatively, these
areas may reflect slightly different chronological phases in the laying out of the pit.
Conclusion
The model of imperial craft organisation we have inferred from the trigger data is consistent
with that obtained from the analysis of the arrows and suggested for the terracotta warriors.
Having several versatile cells, possibly functioning in parallel, to produce weapons and equip
various sectors of the pit would require a multiplication of tools and skills, as well as a strong
supervisory structure to ensure comparable standards—especially if a central government
‘gong’ worked in parallel with other production cells. This model, however, would offer
greater logistical adaptability and accommodate potential breakdowns or changes in any
masterplan. Overall, it is fair to say that the organisation of production in Pit 1 offers a view
in microcosm of wider patterns of close political and economic control in the Qin Empire.
Acknowledgements
Xiuzhen Li was supported by Sun Hung Kai Properties Kwok Foundation for her PhD project and by Rio Tinto
plc for post-doctoral research. Thanks are due to the National Bureau of Cultural Heritage, China, and Shaanxi
Provincial Bureau of Cultural Relics for their approval and encouragement. We are indebted to the late Peter
Ucko and to Wu Yongqi, who made this co-operation possible. Thanks also go to Chang Qiuyue, Yan Hongxia
and other colleagues in the Conservation Department of the Emperor Qin Shihuang’s Mausoleum Site Museum
for their assistance. We are very grateful to the British Academy for supporting the on-going co-operative research
between the Emperor Qin Shihuang’s Mausoleum Site Museum and UCL Institute of Archaeology as a British
Academy Research Project.
References
BAGLEY, R.W. 1995. What the bronzes from Hunyuan
tell us about the foundry at Houma. Orientations
26(1): 46–54.
BEVAN, A., E. CREMA,X.LI&A.PALMISANO. In press.
Intensities, interactions and uncertainties: some
new approaches to archaeological distributions, in
A. Bevan & M. Lake (ed.) Computational
approaches to archaeological spaces. Walnut Creek
(CA): Left Coast.
C
Antiquity Publications Ltd.
138

Research
Xiuzhen Janice Li et al.
DUAN, Q. 2011. Qin shihuangdi lingyuan kaogu yanjiu
[Research on the tomb complex of Emperor Qin
Shihuang]. Beijing: Beijing University Press.
EERKENS,J.W.&R.L.BETTINGER. 2001. Techniques
for assessing standardisation in artefact assemblages:
can we scale material variability? American Antiquity
66: 493–504.
EERKENS,J.W.&C.P.LIPO. 2005. Cultural
transmission, copy errors, and the generation of
variation in material culture and the archaeological
record. Journal of Anthropological Archaeology 24:
316–34.
FREESTONE,I.,J.PRICE &C.CARTWRIGHT. 2009a.
The batch: its recognition and significance, in K.
Janssens, P. Degryse, P. Cosyns, J. Caen & L. Van’t
dack (ed.) Annales of the 17th Congress of the
International Association for the History of Glass,
2006, Antwerp: 130–35. Antwerp: University Press
Antwerp.
FREESTONE,I.,Z.YEGINGIL &R.ARIK. 2009b.
Scientific analysis of glazed tile from the Seljuq
palace of Kubad-Abad, Lake Beysehir, Turkey, in B.
McCarthy,E.S.Chase&L.Allis(ed.)Scientific
research on historic Asian artefacts: proceedings of the
Fourth Forbes Symposium at the Freer Gallery of Art:
3–8. London & Washington, D.C.:
Archetype, in association with the Freer Gallery of
Art.
FREESTONE,I.,J.KUNICKI-GOLDFINGER,H.
GILDERALE-SCOTT &T.AYERS. 2010.
Multi-disciplinary investigation of the windows of
John Thornton, focusing on the Great East
Window of York Minster, in M.B. Shepard, L.
Pilosi & S. Strobl (ed.) The art of collaboration:
stained glass conservation in the twenty-first century:
151–58. Turnhout: Harvey Millar.
GROOVER, M.P. 2010. Fundamentals of modern
manufacturing: materials, processes and systems.
Upper Saddle River (NJ): Prentice Hall.
HUANG, S. 1990. Qinyongkeng Chutu Bingqi
Mingwen Yu Xiangguan Zhidu Fafu [The
inscriptions on the bronze weapons for the Qin
Terracotta Army and related systems]. We n b o
1990(5): 25–35.
ILLIAN, J., A. PENTTINEN,H.STOYAN &D.STOYAN.
2008. Statistical analysis and modelling of spatial
point patterns. New York: Wiley Interscience.
JIANG,W.&Z.LIU. 2011. Qinyongkeng Chutu
Keming Jinian Bingqi Yanjiu [A research on the
inscriptions and dating of the weapons found in the
pits of the terracotta warriors]. Journal of Emperor
Qin Shihuang’s Mausoleum Site Museum 2011:
239–50.
LEDDEROSE, L. 2000. Ten thousand things—module and
mass production in Chinese art. Princeton (NJ):
Princeton University Press.
LI, Y. 2006. The Anyang bronze foundries:
archaeological remains, casting technology, and
production organisation. Unpublished PhD
dissertation, Harvard University.
LI,X.&J.GAO. 2010. Cong Qintong Nuji Mingwen
Kan Qin Bingqi Shengchan Biaozhunhua De
Juxianxing [A discussion about the limitation of the
Qin bronze weapons’ standardisation from the
inscriptions]. Wen b o 2010(2): 43–46.
LI,X.,M.MARTIN ´
ON-TORRES,N.D.MEEKS,Y.XIA &
K. ZHAO. 2011. Inscriptions, filing, grinding and
polishing marks on the bronze weapons from the
Qin Terracotta Army in China. Journal of
Archaeological Science 38: 492–501.
LI,X.,M.MARTIN ´
ON-TORRES,N.D.MEEKS &Y.XIA.
2012. Scanning electron microscopy imaging of
tool marks on Qin bronze weapons using silicone
rubber impressions, in N. Meeks, C. Cartwright, A.
Meek & A. Mongiatti (ed.) Historical technology,
materials and conservation: SEM and microanalysis:
62–68. London: Archetype & British Museum.
LIU, Z. 1986. Qinyongkeng Gongnu Shitan [A
preliminary study on the crossbows from the pits of
the terracotta warriors]. Wenb o 1986(4): 21–27.
MARTIN´
ON-TORRES, M., X. LI,A.BEVAN,Y.XIA,K.
ZHAO &TH.REHREN. 2012. Forty thousand arms
for a single emperor: from chemical data to labor
organization in the production of bronze arrows for
the Terracotta Army. Journal of Archaeological
Method and Theory October 2012.
http://dx.doi.org/10.1007/s10816–012–9158-z
MEQSTA Museum of Emperor Qin Shihuang’s
Terracotta Army. 2008. Qinshihuang ling erhao
bingmayong keng fajue baogao [An excavation
report on Pit 2 of the terracotta warriors and horses
of Emperor Qin Shihuang]. Beijing: Science Press.
ORTON, C. 2004. Point pattern analysis revisited.
Archaeologia e Calcolatori 15: 299–315.
PORTAL, J. (ed.). 2007. The first emperor: China’s
Terracotta Army. London: British Museum.
QIN,B.&Z.ZHANG. 1983. Chang Pi Chu Tan [A
preliminary research on the Qin lances], in Shaanxi
Kaogu Yanjiusuo Shaanxi Sheng Kaogu Xuehui
Diyici Niahui Lunwenji [Proceedings of the 1st
annual conference of Shaanxi Archaeological
Committee]: 209–14. Xi’an: Sanqin.
ROBERT,C.P.&G.CASELLA. 2004. Monte Carlo
statistical methods.NewYork:Springer.
SIAATQ Shaanxi Institute of Archaeology and
Archaeological Team of Qinshihuangling. 1988.
Qinshihuang Ling Bingmayong Keng – Yihao Keng
Fajue Baogao 1974–1984 [An excavation report on
Pit 1 of the terracotta warriors and horses for the
Emperor Qin Shihuang 1974–1984]. Beijing:
Cultural Relics Press.
C
Antiquity Publications Ltd.
139

Crossbows and imperial craft organisation
SIAMEQSTA Shaanxi Institute of Archaeology and
Museum of Emperor Qin Shihuang’s Terracotta
Army. 1998. Qinshihuangdi Ling Tongchema Fajue
Baogao [An excavation report of the bronze chariots
within the Emperor Qin Shihuang’s tomb complex].
Beijing: Cultural Relics Press.
— 2000. Qinshihuangdi Lingyuan Kaogu Baogao 1999
[An archaeological report of the Emperor Qin
Shihuang’s tomb complex 1999]. Beijing: Science
Press.
— 2006. Qinshihuangdi Lingyuan Kaogu Baogao 2000
[An archaeological report of the Emperor Qin
Shihuang’s tomb complex 2000]. Beijing: Cultural
Relics Press.
— 2007. Qinshihuangdi Lingyuan Kaogu Baogao
2001–2003 [An archaeological report of the Emperor
Qin Shihuang’s tomb complex 2001–2003]. Beijing:
Cultural Relics Press.
WANG, X. 1980. Qinyong keng chutu qingtong bingqi
de keji chengjiu [Research into the scientific
achievement based on the Qin bronze weapons
from the pits of the Qin Terracotta Army].
Kaoguyuwenwu 1980(3): 99–107.
– 1983. Qinyong bingqi zhou lun [A preliminary
discussion about the weapons for the Qin terracotta
warriors]. Kaoguyuwenwu 1983(4): 59–80.
– 1990. Yi fu qindai de chen bing tu [A military battle
formation of the Qin dynasty]. Wenb o 1990(5):
21–35.
– 1994. Qinyong Zhuanti Yanjiu [A monographic study
on the Qin terracotta warriors]. Xi’an: Sanqin.
WILLIAMS, D. 2008. Mass-produced pre-Han Chinese
bronze crossbow triggers: unparalleled
manufacturing technology in the ancient world.
Arms and Armour 5: 142–53.
YANG, H. 1980. Zhongguo gudai bingqi luncong
[Research on ancient Chinese weapons]. Beijing:
Cultural Relics Press.
YATE S, R. 2007. The rise of Qin and the military
conquest of the Warring States, in J. Portal (ed.)
The first emperor: China’s Terracotta Army: 30–57.
London: British Museum.
YUAN, Z. 1984. Qin zhongyang duzhao de bingqi keci
zongsu [Interpreting the inscriptions about the Qin
governmental supervision on weapons’ production].
Kaoguyuwenwu 1984(5): 21–26.
– 1990. Qinshihuang ling bingmayong yanjiu [Research
on the terracotta warriors and horses from the Emperor
Qin Shihuang’s tomb complex]. Beijing: Cultural
Relics Press.
– 2002. Qinshihuang ling kaogu faxian yu yanjiu
[Archaeological discoveries and research on the
Emperor Qin Shihuang’s mausoleum]. Xi’an: Shaanxi
People Press.
YUAN,Z.&Y.LIU. 2009. Qin taowen xinbian [New
edition of the Qin characters found on ceramics].
Beijing: Cultural Relics Press.
YUAN,Z.&M.QIN. 1975. Qinyong Keng Bingmayong
Junzhen Neirong Ji Bingqi Shitan [An investigation
on the battle formation of the Qin Terracotta Army
and weapons discovered]. Wen w u 1975(11): 15–19.
YUAN,W.,Z.YUAN,P.ZHAO,Z.ZUO,X.WANG &Y.
WEI. 1981. Qinyongkeng qingtong bingqi gongyi
biaozunhua shixi [Research into the standardisation
of Qin bronze weapons found in the pits of the
terracotta warriors]. Shaanxi Standardisation
1981(2): 5–9.
Received: 6 November 2012; Accepted: 30 May 2013; Revised: 8 July 2013
C
Antiquity Publications Ltd.
140