Journal of Archaeological Science 94 (2018) 12–20

Contents lists available at ScienceDirect

Journal of Archaeological Science

journal homepage: www.elsevier.com/locate/jas

New osteological criteria for the identification of domestic horses, donkeys T

and their hybrids in archaeological contexts
Pauline Hanota,∗, Corentin Bochatonb
a
Laboratoire "Archéozoologie et Archéobotanique: Sociétés, Pratiques et Environnements" UMR 7209 – CNRS, MNHN - Muséum National d’Histoire Naturelle - Sorbonne
Universités, 55 rue Buffon, CP 56, 75005 Paris, France
b
Max Planck Institute for the Science of Human History, Department of Archaeology, Kahlaische Straße 10, D-07745, Jena, Germany

A R T I C LE I N FO A B S T R A C T

Keywords: The identification of domestic equid remains is a recurrent issue and an intense subject of discussion in
Morphological variability zooarchaeological studies. Indeed, despite historical sources describing the key role of equids in numerous past
Equus societies, their accurate identification on archaeological sites is still problematic, and only few methods have
Osteology been developed in order to distinguish the bones of horses, donkeys and their hybrids. Moreover, some of the
Taxonomic identification
extant published visual macroscopic criteria are considered as possibly unreliable, partly because of the absence
Zooarchaeology
of preliminary test on a large sample of modern specimens. In this work, we try to solve these issues by testing a
set of macroscopic visual criteria, collected in the literature or newly described, on a comparative sample of 107
modern skeletons of domestic equids. We quantified the reliability of these criteria and found evidence of 26
osteological characters allowing for the identification of between 90% and 100% of the horses and donkeys of
our comparative sample. A method to identify the complete or sub-complete skeletons of hybrids is also pro-
posed using combinations of characters observed on several bones. Finally, the defined osteological criteria are
observed on a set of archaeological skeletons, coming from antique to modern sites, in order to demonstrate the
applicability of our approach to archaeological remains. The use of our methodology on zooarchaeological
samples could allow for a better assessment of the presence of donkeys and hybrids in archaeological sites, and
thus, could help improve the knowledge of their respective importance and use by human past societies.

1. Introduction respective uses by humans (Clutton-Brock, 1992). For instance in the

The reliable identification of closely related taxa is a recurrent issue
in zooarchaeology and this question is critical for taxa of specific socio-
economical interest, such as equids (Clutton-Brock, 1992). The latter
are represented by two domestic species which may have been si-
multaneously present in European archaeological sites since the Iron
Age (Bökönyi, 1991): the horse (Equus caballus Linnaeus, 1758), and the
donkey (Equus asinus Linnaeus, 1758). The identification of these taxa is
made even more complex by the fact that they can crossbreed, which
implies the potential occurrence of their hybrids in archaeological de-
posits: mules (Equus asinus x Equus caballus), and hinnies (Equus caballus
x Equus asinus).
An accurate and systematic identification of archaeological equid
remains is however mandatory for a good understanding of archae-
ological deposits, and for a correct description of animal exploitation
strategies in past human societies. Indeed, domestic equids are known
to have played a key role in the economy of many past civilizations
worldwide, and many historical sources extensively describe their
Roman empire, horses were largely used for riding, hunting and racing
(White, 1970) because of their running ability, whereas donkeys were
acknowledged for their endurance making them particularly suitable
for traction, and as pack animal in farming activities (Hyland, 1990;
Peters, 1998; Toynbee, 1973). Concerning mules, they were renowned
for their vigor and were mostly employed for long distance transport of
persons or goods (Armitage and Chapman, 1979). At the opposite,
hinnies were rarely used, and are described as being of low working
interest (Clutton-Brock, 1992; Loudon, 1825). This broad diversity of
uses demonstrates that an accurate and reliable identification of equid
archaeological remains is of high interest for a better understanding of
socio-economic systems in past societies.
Several methodologies have been developed in order to distinguish
equid species from bone morphology. Approaches based on the ob-
servation of visual macroscopic criteria were first proposed (Arloing,
1882; Rosselli Vilá, 1921) and, more recently, methods using geometric
morphometrics (GMM) have been applied to this question in order to
provide more reliable identifications (Cucchi et al., 2017; Hanot et al.,

∗
Corresponding author.
E-mail addresses: pauline.hanot@mnhn.fr (P. Hanot), bochaton@shh.mpg.de (C. Bochaton).

https://doi.org/10.1016/j.jas.2018.03.012
Received 1 November 2017; Received in revised form 30 March 2018; Accepted 31 March 2018
0305-4403/ © 2018 Elsevier Ltd. All rights reserved.
P. Hanot, C. Bochaton
2017). Nevertheless, GMM approaches remain challenging to imple-
ment, and they currently suffer from a low practicability in routine
zooarchaeological studies due to the relatively long sequence of ana-
lytical protocols they require. On the other hand, the reliability of the
methods based on macroscopic criteria is often called into question
partly because of the surprisingly small amount of remains attributed to
donkeys and hybrids in archaeological sites which contrasts with their
seeming economic importance documented by historical sources
(Albarella et al., 1993; Bökönyi, 1974; Johnstone, 2006; Manconi,
1995; Peters, 1998). Most of the currently used identification criteria
are based on the morphology of teeth enamel (Armitage and Chapman,
1979; Davis, 1980; Eisenmann, 1986; Payne, 1991; Uerpmann, 2002),
and of the skull (Albizuri and Nadal, 1991; Azzaroli, 1978; Eisenmann,
1986, 1980; Groves and Mazák, 1967; Kunst, 2000) but only few cri-
teria are available for postcranial elements (Arloing, 1882; Barone,
1986; Eisenmann and Beckouche, 1986; Peters, 1998; Rosselli Vilá,
1921). Unfortunately, most of these characters are considered as un-
reliable or difficult to observe (Albarella et al., 1993; Baxter, 1998;
Johnstone, 2004; Zeder, 1986). The main issue of these works is often
related to the number of specimens included in the comparative sample
used to describe the criteria. Indeed, the used sample size is often small
and unfitted to a correct consideration of the intraspecific variability
within each taxon. Moreover, these works generally lack preliminary
tests performed on a reference sample in contrary to what has been
done on other taxa (Bochaton et al., 2016; Zeder and Lapham, 2010;
Zeder and Pilaar, 2010); this prevents to quantify the reliability of the
defined identification criteria. These difficulties frequently result in
identifications of archaeological equid remains mainly based on the
largely criticized criterion of simple bone size (Forest, 2008), con-
sidering that horses and mules are larger than donkeys. However, this
size criterion was recently demonstrated as irrelevant by a study of
bone shape data which allowed for the identification of small size horse
bones in archaeological contexts (Hanot et al., 2017). The same study
also revealed that bone size is not appropriate for the distinction of
modern horses, donkeys and hybrids. Another major limitation is the
near-absence of criteria allowing for the identification of hybrids. This
is mainly explained by the fact that they probably concomitantly dis-
play morphological characteristics of both their parents (Forest, 1997).
In addition, the small number of hybrid skeletons available in osteo-
logical collections has likely contributed to restrain the development of
identification methodologies (Johnstone, 2006).
The aim of this study is to try to solve these different issues con-
cerning the identification of archaeological equid remains by proposing
an easily applicable identification method for isolated bones of horses
and donkeys using reliable macroscopic osteological criteria. We also
investigate the morphology of hybrids in order to evaluate their mor-
phological variability and to propose an identification method.
To do so we use a large comparative sample of 107 modern skele-
tons of horses, donkeys, and hybrids in order to assess the reliability of
a set of morphological criteria defined on nine different bones. Eleven
of these characters have been collected in the literature (Barone, 1986;
Eisenmann, 1986; Kunst, 2000; Peters, 1998), but we also describe a set
of 15 new criteria. Finally, in order to demonstrate the relevance of our
approach, we performed an application of our identification metho-
dology on a sample of five equid skeletons discovered on French ar-
chaeological sites.
2. Material and method
2.1. Research of osteological criteria for the distinction between horses and
donkeys
Our modern comparative sample includes 82 complete or sub-
complete skeletons of horses (39 specimens) and donkeys (43 speci-
mens) from several European museum institutions (see online
Supplementary material S1). Horses specimens are represented by the
13
Journal of Archaeological Science 94 (2018) 12–20
two extant caballine species (both belonging to the subgenus Equus;
Groves and Grubb, 2011): 23 domestic horses (Equus caballus Linnaeus,
1758) of various breeds, and 15 Przewalski's horses (Equus przewalskii
Poliakov, 1881). Both males and females are included, in relatively
equal proportions (see online Supplementary material S1). All speci-
mens are adult with fully fused long bones epiphyses in order to discard
the impact of growth on bone morphology.
Our study focused on nine bone elements from cranial (skull and
mandible) and postcranial (scapula, humerus, radius/ulna, third me-
tacarpal bone, femur, tibia, and proximal phalanx) skeleton. Between
one and five criteria were found as reliable on each bone, 15 of these
criteria were newly described by the authors and 11 originated from the
literature. These criteria are described in our study following the ana-
tomical nomenclature of Barone (1976). For each criterion, two char-
acter states were defined corresponding to horse (“A”) and donkey
(“B”). When the character state was not clearly recordable on com-
parative specimens, it was registered as such (“A/B”). A criterion which
was impossible to observe (e.g. broken bone, bones kept in anatomical
connection, etc.) was registered as unobservable (“-”). All the ob-
servations were carried out by a single observer (PH).
To synthesize the observations, several quantification tools were
defined. The Number of Observations (NObs) refers to the total number
of times a criterion was observed should it be a clear character state
(“A”, “B”) or not (“A/B”). The Number of Attributions (NA) corresponds
to the number of times a criterion was unambiguously attributed to a
character state (“A” or “B”). The Number of Correct attributions (NC)
corresponds to the number of correctly-classified specimens for each
criterion (sum of the number of horses correctly identified by the “A”
state and of the number of donkeys correctly identified by the “B”
state). This allows us to compute the Percentage of Assessment
(PA=NA/NObs*100) and a Correct Identification Rate (CIR=NC/
NA*100). The PA is supposed to reflect the variability of the criterion
and the difficulty to clearly define the character state. As for the CIR, it
is also supposed to reflect the reliability of the criterion. Only the cri-
teria with CIR above or equal to 90% and PA above or equal to 80%
were retained and are described in this study.
2.2. Research of osteological criteria for the identification of hybrids
Our modern comparative sample includes 25 skeletons of hybrids
between horse and donkey: 8 hinnies (Equus caballus x Equus asinus) and
17 mules (Equus asinus x Equus caballus). The characters found as reli-
able for the identification of horses and donkeys were tested on hybrid
specimens; among them four have been previously described by Peters
(1998) as allowing to identify mules.
2.3. Archaeological application
The applicability of the defined identification criteria to archae-
ological material was assessed using 5 archaeological equid skeletons
(Table 1). All of these specimens are complete, or almost complete, and
come from French sites dating from the 3rd to the 19th century. They
were all recorded as articulated skeletons in the field and were con-
sidered as corresponding to single individuals. The only exception is the
set of bones from Elbeuf - “Rue Guynemer” whose attribution to a single
specimen is mainly related to the fact that they were discovered in the
same archaeological feature (Barme and Clavel, 2015). All the skeletons
have previously been identified (Table 1) on the basis of both visual
macroscopic (Barme and Clavel, 2015; Derbois, 2006; Lepetz, 1996;
Yvinec, 1998), and GMM criteria (Hanot et al., 2017). However, it
should be mentioned that the identification of the specimen from Elbeuf
has proven to be problematic. Indeed, it was primarily supposed to be a
donkey on the basis of visual criteria observed on teeth coupled with
the small size of its bones (Barme and Clavel, 2015); but a subsequent
study using GMM data has suggested it was a mix between horse and
donkey bones (Hanot et al., 2017).
P. Hanot, C. Bochaton Journal of Archaeological Science 94 (2018) 12–20

Table 1
List of the archaeological specimens.
Archaeological site Locality Datation Anatomical connections Previous identification

rd
"Fresnes-lès-Montauban" Fresnes-lès-Montauban (Pas-de-Calais/ 3 c. Preserved Horse (Hanot et al., 2017; Lepetz, 1996)
France)
th
"Le Village" Longueil-Annel (Oise/France) 18 c. Preserved Horse (Hanot et al., 2017)
"La Vieille Église" Baillet-en-France (Val-d’Oise/France) 10–11th c. Preserved Donkey (Hanot et al., 2017; Yvinec, 1998)
"Rue Guynemer" Elbeuf (Seine-Maritime/France) 16th c. Not preserved Donkey/Horse (Barme and Clavel, 2015; Hanot et al.,
2017)
"155 rue Anatole-France - Les Serres" Méru (Oise/France) 18–19th c. Preserved Donkey (Derbois, 2006)

Fig. 1. Distinctive characters of skull and

mandible. A: Skull of donkey (MNHN-ZA-
AC, 1982.128) and horse (MNHN-
UMR7209-E1) in dorsal view; B: Posterior
part of the skull of donkey (MNHN-ZA-AC,
1982.128) and horse (MNHN-UMR7209-E1)
in lateral view; C: Premaxillary part of the
skull of donkey (MNHN-ZA-AC, 1982.128)
and horse (MNHN-ZA-AC, 1980.29) in ven-
tral view; D: Postero-dorsal part of the
mandible of donkey (MNHN-ZA-AC,
1982.128) and horse (MNHN-UMR7209-
ind.) in lateral view; E: Anterior part of the
mandible of donkey (MNHN-ZA-AC,
1982.128) and horse (MNHN-ZA-AC,
1929.35) in lateral view. Abbreviations: c.
p.: coronoid process; e. o. p.: external oc-
cipital process; f.: frontal; i. b.: inter-al-
veolar border; m. f.: mental foramen; n. p.:
nuchal plane; o.: orbit; o. c.: occipital con-
dyle; t. i.: third incisor.

3. Results
The full set of observations performed on each specimen of the
modern sample is included as online Supplementary material S2.
3.1. Reliable criteria for the distinction between horses and donkeys
3.1.1. Skull
SK1 (N Obs = 64, PA = 86%, CIR = 100%): Shape of the zygomatic
process of the frontal bone (Fig. 1-A) – visible in anterior and dorsal
views.
In donkeys, the zygomatic process of the frontal bone is more lat-
erally extended (SK1B) than in horses (SK1A). This difference con-
cerning the morphology of the supra-orbital border of the frontal was
first signaled by Barone (1976) and is also mentioned by Johnstone
(2004).
SK2 (N Obs = 65, PA = 85%, CIR = 100%): Posterior extension of
the external occipital process (Fig. 1-B) – visible in lateral view.
In horses, the external occipital process is weakly posteriorly ex-
tended above the nuchal plane (SK2A) whereas, in donkeys, this process
presents a strong posterior projection that overcomes the posterior
margin of the occipital condyle (SK2B). This last character state was
mentioned by Barone (1976) as characteristic of donkeys, and by
Arloing (1882) as observable on hybrids too. Moreover, the recent
GMM study of Hanot et al. (2017) also signaled it.
SK3 (N Obs = 65, PA = 88%, CIR = 96%): Morphology of the
anterior part of the incisive bone (Fig. 1-C) – visible in ventral view.
In horses, the inter-alveolar borders of the incisive bone are arched
(SK3A) whereas they are more parallel and straight in donkeys (SK3B).
This criterion was previously signaled by Johnstone (2004).

14
P. Hanot, C. Bochaton
SK4 (N Obs = 65, PA = 92%, CIR = 93%): Morphology of the
anterior margin of the temporal fossa (not figured) – visible in dorsal
view.
In horses, the anterior margin of the temporal fossa is bordered by a
sharp crest prolonging the temporal line (SK4A). In donkeys, the
anterior margin of the temporal fossa is thicker and blunted (SK4B).
3.1.2. Mandible
MD1 (N Obs = 56, PA = 88%, CIR = 96%): Morphology of the
coronoid process (Fig. 1-D) – visible in lateral view.
In horses, the dorsal extremity of the coronoid process of the
mandible is rounded or truncated in lateral view (MD1A). In donkeys, it
is more triangular (MD1B). The shape of the coronoid process was also
considered as discriminating by Barone (1976) who described it as in-
curved in posterior direction in donkeys.
MD2 (N Obs = 65, PA = 91%, CIR = 90%): Orientation of the
mental foramen (Fig. 1-E) – visible in lateral view.
In horses, the mental foramen (“foramen mentonnier” sensu Barone,
1976) is oriented in posterior direction resulting in a medial border of
the foramen visible in lateral view (MD2A). In donkeys, the mental
foramen is more oriented in medial direction and so laterally deeper. As
a consequence the medial delimitation of the foramen of donkeys is not
visible in lateral view (MD2B).
3.1.3. Scapula
SC1 (N Obs = 69, PA = 84%, CIR = 97%): Shape of the tuberosity
of the scapular spine (Fig. 2-A) – visible in lateral view.
In horses, the tuberosity of the scapular spine is elongated (SC1A)
whereas it is ovoid in donkeys (SC1B).
3.1.4. Humerus
H1 (N Obs = 74, PA = 85%, CIR = 98%): Shape of the minor tu-
bercle (Fig. 2-B) – visible in medial view.
In horses, the anterior part of the minor tubercle is rounded and
weakly extended in anterior direction (H1A). In donkeys, the minor
tubercle presents a rectangular shape and is well-extended in anterior
direction (H1B).
H2 (N Obs = 74, PA = 91%, CIR = 97%): Morphology of the in-
termediate tubercle (Fig. 2-B) – visible in medial view.
In horses, the intermediate tubercle is well-extended in anterior
direction (H2A). In donkeys, the intermediate tubercle is weakly ante-
riorly extended (H2B). The prominence of the intermediate tubercle
was also signaled by Barone (1976) as a distinctive character between
horses and donkeys.
H3 (N Obs = 75, PA = 96%, CIR = 92%): Shape of articular head
(Fig. 2-C) - visible in dorsal view.
In horses, the most anterior point of the articular head of the hu-
merus is located at the level of the intertubercular groove, between the
lateral and intermediate tubercles (H3A). In donkeys, this point is lo-
cated at the level of the intermediate tubercle (H3B).
H4 (N Obs = 75, PA = 97%, CIR = 90%): Lateral epicondylar crest
(Fig. 2-D) – visible in ventral view.
In horses, the posterior crest of the lateral epicondyle is blunted,
giving to the posterior extremity a rounded shape in ventral view
(H4A). In donkeys, the crest of the lateral epicondyle is sharper, giving
to the posterior extremity a pointed shape in ventral view (H4B).
3.1.5. Radius/ulna
R1 (N Obs = 74, PA = 95%, CIR = 100%): Morphology of the
transversal crest of the radius (Fig. 3-A) – visible in posterior and
ventral view.
In horses, the radius presents a continuous and straight distal
transversal crest (R1A). In donkeys, this transversal crest is dug by a
wide furrow passing through the epiphyseal junction. In ventral view,
this furrow gives a curved shape to the transversal crest (R1B). This last
character state was previously mentioned by Barone (1976) as
15
Journal of Archaeological Science 94 (2018) 12–20
characteristic of donkeys, and by Peters as also occurring on mules
(1998).
R2 (N Obs = 74, PA = 81%, CIR = 98%): Shape of the lateral tu-
berosity of the olecranon (Fig. 3-B) – visible in lateral view.
In horses, the lateral tuberosity of the olecranon for the insertion of
the lateral head of the triceps brachii is of rounded shape (R2A). In
donkeys, this tuberosity is more expended and elongated in ventral
direction (R2B).
R3 (N Obs = 74, PA = 95%, CIR = 91%): Ossification of the in-
terosseous proximal ligament (Fig. 3-B) – visible in lateral view.
In horses, the interosseous proximal ligament is not ossified re-
sulting in a gap between radius and ulna bones above the interosseous
space (R3A). In donkeys, the interosseous proximal ligament tends to be
ossified making the ulna completely fused to the radius above the in-
terosseous space (R3B). This character was first signaled by Barone
(1976).
3.1.6. Third metacarpal
MC1 (N Obs = 69, PA = 91%, CIR = 97%): Shape of the distal ex-
tremity (Fig. 3-C) – visible in posterior view.
In horses, the posterior intermediate projection of the distal condyle
is well extended in dorsal direction making curvilinear the postero-
dorsal border of the distal epiphysis of the bone (MC1A). In donkeys,
the posterior intermediate projection of the distal condyle is weakly
extended in dorsal direction making straighter the postero-dorsal
border of the distal epiphysis of the bone (MC1B).
MC2 (N Obs = 72, PA = 86%, CIR = 94%): Depression on the distal
posterior area (Fig. 3-C) - visible in posterior view.
In horses, the distal posterior area of the third metacarpal is flat (A).
In donkeys, this area is depressed (MC2B). This last character state was
previously signaled by Arloing (1882) as characteristic of donkeys and
mules, and by Peters (1998) as occurring on mules.
3.1.7. Proximal phalanx
P1 (N Obs = 71, PA = 80%, CIR = 96%): Imprint of the distal
middle sesamoidean ligament (Fig. 3-D) – visible in palmar and lateral
views.
In horses, the imprint of the distal middle sesamoidean ligament
(“ligamentous insertions” sensu Barone, 1976) is weakly developed
with a blunt margin (P1A). In donkeys, this imprint is strongly marked
with a sharp margin (P1B). This last character state was signaled as
occurring on mules by Peters (1998).
3.1.8. Femur
F1 (N Obs = 74, PA = 97%, CIR = 100%): Occurrence of a crest
linking the third to the greater trochanter (Fig. 4-A) – visible in lateral
view.
In horses, the femur bears a well-marked sharp crest linking the
third trochanter to the greater trochanter (F1A). In donkeys, this crest is
absent and this area is smooth (F1B).
F2 (N Obs = 74, PA = 91%, CIR = 99%): Medial margin of the in-
tertrochanteric crest (Fig. 4-B) – visible in medial view.
In horses, the medial margin of the intertrochanteric crest deli-
miting the posterior limit of the trochanteric fossa is thick and blunted
(F2A). In donkeys, the medial margin of the intertrochanteric crest is
thin and sharp (F2B).
F3 (N Obs = 74, PA = 96%, CIR = 97%): Morphology of the
greater trochanter (Fig. 4-A) – visible in lateral view.
In horses, the dorsal margin of the greater trochanter is of triangular
shape (F3A). In donkeys, this structure is rounded and shorter (F3B).
F4 (N Obs = 74, PA = 95%, CIR = 97%): Morphology of the con-
vexity of the greater trochanter (Fig. 4-C) – visible in dorsal view.
In horses, the convexity of the greater trochanter presents a well-
delimited dorsal tuberosity that extends along an antero-posterior axis
(F4A). In donkeys, this tuberosity presents a medial extension on the
anterior margin of the epiphysis (F4B).
P. Hanot, C. Bochaton
F5 (N Obs = 74, PA = 82%, CIR = 95%): Dorsal extension of the
lesser trochanter (Fig. 4-B) – visible in medial view.
In horses, the lesser trochanter extends dorsally toward the head of
the femur in the form of a well-marked crest whose dorsal limit nearly
reaches the head of the femur (F5A). In donkeys, the dorsal extension of
the lesser trochanter is either absent or marked by a weakly defined
crest whose dorsal limit reaches a maximum of two-third of the distance
between the tuberosity and the head of the femur (F5B).
3.1.9. Tibia
T1 (N Obs = 74, PA = 91%, CIR = 99%): Shape of the distal ar-
ticular surface (Fig. 5-A) – visible in ventral view.
In horses, the posterior margin of the distal extremity of the tibia is
quite straight providing a nearly rectangular shape to the articular
surface (T1A). In donkeys, the articular surface is extended in postero-
lateral direction and presents a more trapezoidal shape (T1B). This last
character state was previously signaled by Peters (1998) as occurring
on mules.
T2 (N Obs = 75, PA = 91%, CIR = 96%): Shape of the medial
condyle of the proximal articular surface (Fig. 5-B) – visible in dorsal
view.
In horses, the medial articular surface of the tibia presents a
rounded posteromedial margin (T2A). In donkeys, the medial articular
16
Journal of Archaeological Science 94 (2018) 12–20
Fig. 2. Distinctive characters of scapula and
humerus. A: Scapula of donkey (MNHN-ZA-
AC, 1982.128) and horse (MNHN-ZA-AC,
1929.35) in lateral view; B: Dorsal part of
the humerus of donkey (MNHN-ZA-AC,
1982.128) and horse (MNHN-ZA-AC,
1962.228) in lateral view; C: Humerus of
donkey (MNHN-ZA-AC, 1982.128) and
horse (MNHN-ZA-AC, 1929.35) in dorsal
view; D: Humerus of donkey (MNHN-ZA-AC,
1982.128) and horse (MNHN-ZA-AC,
1929.35) in ventral view. Abbreviations: a.
h.: articular head; i. t.: intermediate tu-
bercle; l. e.: lateral epicondyle; l. t.: lateral
tubercle; m. t.: minor tubercle; s. s.: scap-
ular spine; t. s. s.: tuberosity of the scapular
spine.
surface of the tibia does not extend to the medial and posterior margins
of the bone, and its posteromedial margin forms an angle (T2B).
T3 (N Obs = 75, PA = 92%, CIR = 94%): Shape of the lateral
condyle of the proximal articular surface (Fig. 5-B) – visible in dorsal
view.
In horses, the posterior extremity of the lateral condyle of the tibia is
enlarged (T3A). In donkeys, this extremity presents a more constricted
and pointed shape (T3B).
T4 (N Obs = 75, PA = 87%, CIR = 92%): Morphology of the tu-
berosity of the tibial crest (Fig. 5-C) – visible in anterior view.
In horses, the tuberosity for attachment of the semitendinosus muscle
located on the tibial crest is relatively weakly developed (T4A). In
donkeys, the tuberosity is more developed (T4B). This last character
state was previously highlighted by Barone (1976) and Arloing (1882)
as a characteristic of donkeys.
3.2. Characterization of hybrids specimens
We found no reliable visual qualitative character for the identifi-
cation of hybrids on the basis of osteology. According to our results,
hybrids have proved to be strongly morphologically variable in regard
of their character states with specimens displaying from 100% of
donkey criteria to 95% of horse criteria (see online Supplementary
P. Hanot, C. Bochaton Journal of Archaeological Science 94 (2018) 12–20

Fig. 3. Distinctive characters of radius/ulna,

third metacarpal and proximal phalanx. A:
Radius/ulna of donkey (MNHN-ZA-AC,
1982.128) and horse (MNHN-ZA-AC,
1929.35) in ventral view; B: Proximal part
of the radius/ulna of donkey (MNHN-ZA-
AC, 1982.128) and horse (MNHN-ZA-AC,
1929.35) in lateral view; C: Distal part of the
third metacarpal of donkey (MNHN-ZA-AC,
1982.128) and horse (MNHN-ZA-AC,
1929.35) in posterior view; D: Proximal
phalanx of donkey (MNHN-ZA-AC,
1893.634) and horse (MNHN-ZA-AC,
1975.124) in lateral and ventral views.
Abbreviations: a. f. h.: articular facet with
the humerus; d. p. a.: depression of the
distal posterior area; d. t.: dorsal tuberosity
of the olecranon; i. s. l.: imprint of the se-
samoidean ligament; i. p.: intermediate
projection; i. s.: interosseous space; l. t.:
lateral tuberosity of the olecranon; o.: ole-
cranon; t. c.: transversal crest.

material S3). However, it should be mentioned that the hybrid bones donkey on the basis of its predominant attribution in the case it exceeds
globally display more identification criteria recorded as “B” state 85% of the recorded criteria (among at least 13 criteria recorded as “A”
(corresponding to a donkey character - 47%) than to “A” (horse - 33%) or “B” states); 2) a specimen presenting less than 85% of the criteria
or “AB” (undefined - 20%) states. This means that donkey character corresponding to one of the parent species cannot be undoubtedly
states are more frequently recorded than horses ones on hybrid speci- identified as a horse or donkey and is possibly a hybrid. In our com-
mens. This remark is valid for the criteria from the literature (R1, MC2, parative sample, only hybrid specimens present less than 80% of the
T1, P1) signaled as occurring on hybrid specimens (see online criteria of their predominant attribution.
Supplementary material S4) and that we observed in donkeys. How-
ever, our overall results show that hybrid skeletons present, for most of
3.3. Archaeological application
them, a patchwork of criteria we found as characteristic of horses and
donkeys. This can appear as potentially useful for the identification of
Between 15 and 24 identification criteria have been observed (with
mostly complete skeletons of hybrids using the identification criteria of
clear “A” or “B” character state) on each archaeological skeleton
their parent species.
(Table 1). The detail of the criteria observed on each specimen is pro-
In order to take advantage of this specificity, we considered the
vided in online supplementary material (see online Supplementary
most complete skeletons of our sample (i. e. specimens on which more
material S5).
than half of the 26 retained criteria have been recorded as “A” or “B”
Most (more than 88%) of the criteria recorded on the bones from
states). Among the 63 most complete of our horse and donkey skele-
Fresnes-lès-Montauban and Longueil-Annel correspond to the “A” state.
tons, 62 present more than 85% of character states corresponding to
These two specimens can thus be identified as horses on the basis of our
their species (see online Supplementary material S3). At the opposite,
criteria. Conversely, the skeletons from Baillet-en-France and Meru
among the 21 most complete hybrid skeletons, only 3 present more
predominantly display the morphological features corresponding to the
than 85% of character states corresponding to one of their parents (see
“B” state (respectively recorded in 87% and 92% of the cases) and can
online Supplementary material S3). These data show that the combi-
be considered as donkey skeletons. For these specimens, although the
nation of character states is useful to identify the hybrids of our sample
identification would also have been possible on isolated bones, only the
considering that: 1) a specimen could be identified as a horse or a
study of their complete skeleton allows us to discard the possibility that

17
P. Hanot, C. Bochaton
they could be hybrid specimens.
The results obtained on the specimen from Elbeuf are more am-
biguous. Indeed, the “A” state is predominant but was recorded in only
67% of the cases. This result would indicate that this specimen would
be a hybrid specimen but only at the condition that the skeleton cor-
responds to a single individual.
4. Discussion
We found evidence of 26 osteological criteria allowing for a reliable
discrimination of horse and donkey bones. Among these criteria, four
were reliable in 100% of the cases: the shape of the zygomatic process
of the frontal bone, the morphology of the posterior extension of the
external occipital process, the morphology of the transversal crest of the
radius, and the occurrence of a crest linking the third to the greater
trochanter of the femur. The other criteria were, for 14 of them, reliable
on 95–99% of the horse and donkey skeletons, and the height last cri-
teria provided correct identification on 90–95% of the cases. Several of
these criteria were already signaled in the literature but were never
tested on a large sample of specimens making their reliability difficult
to assess. Thus, our results show that a robust identification of domestic
equids is possible using isolated bones on the basis of macroscopic
criteria.
However, the identification of hybrid specimens has proved to be far
more difficult considering we have been unable to find any typical
osteological character allowing for their identification on isolated
bones. This means that, following our results, hybrids are impossible to
18
Journal of Archaeological Science 94 (2018) 12–20
Fig. 4. Distinctive characters of femur. A:
Proximal part of the femur of donkey
(MNHN-ZA-AC, 1982.128) and horse
(MNHN-ZA-AC, 1929.37) in lateral view; B:
Proximal part of femur of donkey (MNHN-
ZA-AC, 1893.634) and horse (MNHN-ZA-
AC, 1929.37) in medial view; C: Proximal
part of the femur of donkey (MNHN-ZA-AC,
1893.634) and horse (MNHN-ZA-AC,
1980.29) in antero-dorsal view.
Abbreviations: d. e. l. t.: dorsal extension
of the lesser trochanter; g. t.: greater tro-
chanter; h. f.: head of the femur; i. c.: in-
tertrochanteric crest; l. t.: lesser trochanter;
t. t.: third trochanter; t. g. t.: tuberosity of
the greater trochanter; t. f.: trochanteric
fossa.
identify on the basis of osteological criteria observed on isolated bones.
In spite of this limitation, we found that hybrids are, for most of them, a
patchwork of the criteria of both parent species. This strong osteolo-
gical variability was also mentioned about hybrids from other species:
indeed, hybrids are described as more often displaying morphological
closeness with their parents rather than specific hybrid traits (Bochaton
et al., 2016; Evin et al., 2015; Forest, 1997; McDade, 1990; Rieseberg
et al., 1993). We nevertheless found that the identification of hybrids
can be attempted using complete or mostly complete skeletons on
which more than half of our criteria can be unambiguously recorded.
Indeed, almost all of the hybrid skeletons of our sample present less
than 85% of the criteria of one of their parent species, whereas nearly
all modern horse and donkey skeletons present more than 85% of the
character states of their respective species. Our results also show that
most of the criteria precisely recorded (“A” or “B” states) on hybrid
bones are associated to “B” states (corresponding to donkey; 56%) ra-
ther than to “A” states (corresponding to horse; 44%). This remark is
valid for the four criteria defined by Peters (1998) as characterizing
mules (R1, MC2, T1, P1) which, in spite of being retained in our study
as donkey characters, were also recorded in most of our hybrid speci-
mens (see online Supplementary material S4).
These results obtained on the archaeological sample allow for the
unambiguous identification of four of the five archaeological equid
skeletons included in this study as horses (two specimens) or donkeys
(two specimens). The only exception concerns the specimen from the
site of Elbeuf - “Rue Guynemer” which presents criteria of both horse
(ten criteria) and donkey (five criteria). This result could suggest that it
P. Hanot, C. Bochaton
could be a hybrid specimen displaying a mix of characteristics from
both parents. However, the fact that this specimen was signaled as not
discovered in anatomical connection does not allow us to confirm it. In
addition, a previous study using GMM (Hanot et al., 2017) suggested
this specimen would be likely a composite assemblage of bones from
horses and, to a lesser extent, donkeys. This assumption mainly resulted
from the absence of bones providing reliable attribution to hybrids
following the GMM specific classification. In the later study, the authors
suggested that cranial remains could belong to a donkey and most of the
postcranial ones could be horse bones. This hypothesis appears as quite
concomitant with our results and with dental and field data (Barme and
Clavel, 2015). This example highlights the crucial importance of a good
knowledge of the archaeological context and of collecting reliable field
observations in order to allow for a proper study of archaeological
skeletons.
The case of this specimen also demonstrates the interest to use
several methodological approaches jointly in order to obtain reliable
identifications. However, in spite of being globally less powerful than
GMM approaches (particularly on isolated bones), the identification
method we propose presents the advantage of being simpler to use
routinely by zooarchaeologists. Moreover, the fact that our identifica-
tion criteria have been tested on a large sample of comparative speci-
mens enabled us to quantify their reliability and to increase the ro-
bustness of the approach. Our results allow us to raise a final concern
about the comparative samples used in identification methodologies,
especially concerning hybrids which seem to present an important
morphological variability. Indeed, using a small number of comparative
specimens can lead to misleading identifications, no matter the method
used, because it does not allow taking into account the full morpholo-
gical diversity of the species. In summary, we demonstrated that visual
criteria can be considered as a reliable tool for a first study of equid
remains although they may be afterward completed by other metho-
dological approaches (e. g. GMM, Zonkey or ancient DNA when pre-
served; Cucchi et al., 2017; Hanot et al., 2017; Orlando et al., 2015;
Schubert et al., 2017) allowing to secure the identifications and obtain
additional information about the bone remains.
19
Journal of Archaeological Science 94 (2018) 12–20
Fig. 5. Distinctive characters of tibia. A:
Distal part of the tibia of donkey (MNHN-
ZA-AC, 1982.128) and horse (MNHN-ZA-
AC, 1929.35) in ventral view; B: Proximal
part of the tibia of donkey (MNHN-ZA-AC,
1982.128) and horse (MNHN-ZA-AC,
1929.35) in dorsal view; C: Proximal half of
the tibia of donkey (MNHN-ZA-AC,
1982.128) and horse (MNHN-ZA-AC,
1929.35) in lateral view. Abbreviations: l.
c.: lateral condyle of the tibia; m. c.: medial
condyle of the tibia; t. c.: tibial crest; t. t. c.:
tuberosity of the tibial crest.
5. Conclusion
In this study, we demonstrated that domestic horses, donkeys and,
to a certain extent, hybrids can be identified on the basis of visual
criteria using their bone remains. The provided method could allow for
a better assessment of the respective presence of the horses, donkeys
and their hybrids in archaeological sites. It could thus contribute to
improve our knowledge about their place in human past societies.
Further studies allowing for identification of the extinct Equus hy-
druntinus (Regalia, 1907) should also be carried out. Indeed, the
chronology of extinction of the European wild ass is poorly understood
and their remains could have possibly persisted in archaeological re-
cords until historical periods (Jourdan, 1976; Marquet and Poulain,
1985; Willms, 1989).
Our identification tool can be used routinely and could initiate the
development of more argumentative identification procedures. Indeed,
we proposed a reproducible and refutable identification method based
on criteria that have been tested on a large sample of modern skeletons.
This kind of approach is mandatory in order to provide reliable iden-
tifications of archaeological remains that are potentially questionable
and challengeable. Identification methods are a major issue of many
zooarchaeological studies across the world, especially in places where
important taxonomic diversity occurs in the archaeological deposits.
Therefore, many works concerning the skeletal anatomy of vertebrate
taxa have still to be conducted in order to perform more reliable and
challengeable taxonomic identifications of the bone remains.
Funding
This work was supported by the “ATM blanche”/MNHN.
Conflicts of interest
The authors declare that they have no conflict of interest.
P. Hanot, C. Bochaton
Acknowledgments
We are grateful to an anonymous reviewer who helps us to improve
the quality of this work. We also would like to thank the researchers,
curators and collection technicians who allowed us to access to the
reference specimens: Luc Vives, Joséphine Lesur, Christine Lefèvre,
Aurélie Verguin, Céline Bens and Karyne Debue (MNHN-Paris); Benoît
Clavel and Sébastien Lepetz (CNRS/MNHN-Paris); Jean-Hervé Yvinec
(INRAP/CRAVO-Compiègne); Gaëtan Jouanin and Maude Barme
(CRAVO-Compiègne); Benoît Mellier (MSN-Angers); Wim Van Neer,
Georges Lenglet, Sébastien Bruaux and Terry Walschaerts (IRSNB-
Bruxelles); Michael Hiermaier (ZSM-Munich); Henriette Obermaier
(SAPM-Munich); Renate Schafberg (MLU/ZNS/H-Halle/Saale); Erich
Pucher (NHM-Wien). We also want to thank the excavation directors
who recovered the archaeological specimens used in this study: Yves
Desfossés (DRAC Champagne-Ardenne), Stéphane Gaudefroy (INRAP),
François Gentili (INRAP), Bénédicte Guillot (INRAP) and Martine
Derbois (INRAP).
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://dx.
doi.org/10.1016/j.jas.2018.03.012.
References
Albarella, U., Ceglia, V., Roberts, P., 1993. S. Giacomo degli schiavoni (molise): an early
fifth century ad deposit of pottery and animal bones from central adriatic Italy. Pap.
Br. Sch. Rome 61, 157–230.
Albizuri, S., Nadal, J., 1991. Estudi de l’èquid aparegut en relació amb l’estructura E10 de
l'Hort d’en Grimau. Olerdulae 3, 112–117.
Arloing, S., 1882. Caractères ostéologiques différentiels de l’âne, du cheval et de leurs
hybrides. Bull. Société Anthropol. Lyon 236–284.
Armitage, P.L., Chapman, H., 1979. Roman mules. Lond. Archaeol. 3, 339–346.
Azzaroli, A., 1978. On a late pleistocene ass from tuscany, with notes on the history of
asses. Palaeontogr. Ital. Pisa 71, 27–47.
Barme, M., Clavel, B., 2015. La pratique urbaine de l’équarrissage à la charnière du
Moyen-Âge et de l’époque moderne : l’exemple d'Elbeuf (Seine-Maritime).
Archéopages 41, 30–39.
Barone, R., 1986. Anatomie comparée des mammifères domestiques: Ostéologie. Vigot,
Paris, France.
Barone, R., 1976. Anatomie comparée des animaux domestiques: Ostéologie. Vigot, Paris,
France.
Baxter, I.L., 1998. Species identification of equids from Western European archaeological
deposits: methodologies, techniques and problems. In: Anderson, S., Boyle, K. (Eds.),
Current and Recent Research in Osteoarchaeology. Oxbow, Oxford, pp. 3–17.
Bochaton, C., Grouard, S., Breuil, M., Ineich, I., Tresset, A., Bailon, S., 2016. Osteological
differentiation of the Iguana laurenti, 1768 (squamata: iguanidae) species: Iguana
iguana (Linnaeus, 1758) and Iguana delicatissima laurenti, 1768, with some comments
on their hybrids. J. Herpetol. 50, 295–305. https://doi.org/10.1670/14-170.
Bökönyi, S., 1991. The earliest occurrence of domestic asses in Italy. In: Equids in the
Ancient World, pp. 178–216.
Bökönyi, S., 1974. History of Domestic Mammals in Central and Eastern Europe.
Akadémiai Kiadó, Budapest.
Clutton-Brock, J., 1992. Horse Power: a History of the Horse and the Donkey in Human
Societies. Harvard University Press, Cambridge.
Cucchi, T., Mohaseb, A., Peigné, S., Debue, K., Orlando, L., Mashkour, M., 2017. Detecting
taxonomic and phylogenetic signals in equid cheek teeth: towards new palaeonto-
logical and archaeological proxies. Open Sci. 4 (160997). https://doi.org/10.1098/
rsos.160997.
Davis, S.J., 1980. Late pleistocene and holocene equid remains from Israel. Zool. J. Linn.
Soc. 70, 289–312. https://doi.org/10.1111/j.1096-3642.1980.tb00854.x.
Derbois, M., 2006. Méru. ADLFI Archéologie Fr. - Inf. Picardie. https://doi.org/10.4000/
adlfi.4196.
Eisenmann, V., 1986. Comparative osteology of modern and fossil horses, halfasses and
asses. In: Meadow, R.H., Uerpmann, H.-P. (Eds.), Equids in the Ancient World.
Ludwig Reichert Verlag, Wiesbaden, pp. 67–116.
Eisenmann, V., 1980. Les Chevaux (Equus sensu lato) fossiles et actuels ;: crânes et dents
jugales supérieures, CNRS. In: Cahiers de Paléontologie. Paris, France.
Eisenmann, V., Beckouche, S., 1986. Identification and discrimination of metapodials
from Pleistocene and modern Equus, wild and domestic. In: Meadow, R.H.,
Uerpmann, H.-P. (Eds.), Equids in the Ancient World. Ludwig Reichert Verlag,
Wiesbaden, pp. 117–163.
20
Journal of Archaeological Science 94 (2018) 12–20
Evin, A., Dobney, K., Schafberg, R., Owen, J., Vidarsdottir, U.S., Larson, G., Cucchi, T.,
2015. Phenotype and animal domestication: a study of dental variation between
domestic, wild, captive, hybrid and insular Sus scrofa. BMC Evol. Biol. 15 (6).
https://doi.org/10.1186/s12862-014-0269-x.
Forest, V., 2008. Equidés de La Tène finale et de la période romaine en Gaule : approche
ostéométrique. In: L'exploitation agricole dans son environnement à la fin de l'Âge du
Fer : Nouvelles approches méthodologiques. Presented at the Rencontre de Saint-
Julien : 18-19 novembre 2004. Archives d’écologie préhistorique, Toulouse, pp.
61–71.
Forest, V., 1997. Alimentation carnée dans le Languedoc médiéval : les témoignages
archéozoologiques des vertébrés supérieurs. Archéologie Midi Médiév. 14–15. pp.
141–160. https://doi.org/10.3406/amime.1997.1321.
Groves, C., Grubb, P., 2011. Ungulate Taxonomy. JHU Press.
Groves, C.P., Mazák, V., 1967. On some taxonomic problems of Asiatic wild asses; with
the description of a new subspecies (Perissodactyla; Equidae). Z. Für Säugetierkd 32,
321–355.
Hanot, P., Guintard, C., Lepetz, S., Cornette, R., 2017. Identifying domestic horses,
donkeys and hybrids from archaeological deposits: a 3D morphological investigation
on skeletons. J. Archaeol. Sci. 78, 88–98. https://doi.org/10.1016/j.jas.2016.12.002.
Hyland, A., 1990. Equus: the Horse in the Roman World. B.T. Batsford, London.
Johnstone, C.J., 2006. Those elusive mules: investigating osteometric methods for their
identification. In: Mashkour, M. (Ed.), Equids in Time and Space : Papers in Honour of
Véra Eisenmann. Proceedings of the 9th Conference of the International Council of
Zooarchaeology. Oxbow, Oxford, pp. 183–191.
Johnstone, C.J., 2004. A Biometric Study of Equids in the Roman World. University of
York, York.
Jourdan, L., 1976. La Faune du Site Gallo-Romain et Paléo-Chrétien de la Bourse,
Marseille: Espèces Domestiques et Espèces Sauvages. Centre National de la Recherche
Scientifique.
Kunst, G.K., 2000. Archaeozoological evidence for equid use, sex structure and mortality
in a Roman auxiliary fort (Carnuntum-Petronell, lower Austria). Anthropozoologica
31, 109–118.
Lepetz, S., 1996. L’animal dans la société gallo-romaine de la France du nord. Rev.
Archéologique Picardie n° spécial 12, 1–172. https://doi.org/10.3406/pica.1996.
1906.
Loudon, J.C., 1825. An Encyclopædia of Agriculture. London, UK.
Manconi, F., 1995. Equidi in Sardegna tra 2 sec. a.C. e il 7 sec. d.C. In: Collana Di Studi
Monograjici Del Centro Polesano Di Study Storici, Archaeologici Ed Etnograjici
Rovigo. Atti Del I Convegno Di Archeozoologia, Rovigo 5-7 Marzo 1993. Padusa
Quaderni n.1, Rovigo, pp. 319–329.
Marquet, J., Poulain, T., 1985. Un fossé dépotoir de La Tène III à Vernou-sur-Brenne
(Indre-et-Loire). III-Faune et restes humains. Rev. Archéologique Cent. Fr 24, 68–74.
McDade, L., 1990. Hybrids and phylogenetic systematics I. Patterns of character ex-
pression in hybrids and their implications for cladistic analysis. Evolution 44,
1685–1700. https://doi.org/10.2307/2409347.
Orlando, L., Gilbert, M.T.P., Willerslev, E., 2015. Reconstructing ancient genomes and
epigenomes. Nat. Rev. Genet. 16, 395–408. https://doi.org/10.1038/nrg3935.
Payne, S., 1991. Early holocene equids from tall-I-mushki (Iran) and can hasan III
(Turkey). In: Meadow, R.H., Uerpmann, H.-P. (Eds.), Equids in the Ancient World. Dr
Ludwig Reichert Verlag, Wiesbaden, pp. 132–164.
Peters, J., 1998. Römische Tierhaltung und Tierzucht: eine Synthese aus
archäozoologischer Untersuchung und schriftlich-bildlicher Uberlieferung. Leidorf,
Rahden, Allemagne.
Rieseberg, L.H., Ellstrand, N.C., Arnold, D.M., 1993. What can molecular and morpho-
logical markers tell us about plant hybridization? Crit. Rev. Plant Sci. 12, 213–241.
https://doi.org/10.1080/07352689309701902.
Rosselli Vilá, M., 1921. Extrait de Contribution à l'Ostéologie comparée du Cheval et de
l'Ane et Zootechnie de la Race Asine Catalane. Editorial Catalana, S. A., Barcelona.
Schubert, M., Mashkour, M., Gaunitz, C., Fages, A., Seguin-Orlando, A., Sheikhi, S.,
Alfarhan, A.H., Alquraishi, S.A., Al-Rasheid, K.A.S., Chuang, R., Ermini, L., Gamba,
C., Weinstock, J., Vedat, O., Orlando, L., 2017. Zonkey: a simple, accurate and sen-
sitive pipeline to genetically identify equine F1-hybrids in archaeological assem-
blages. J. Archaeol. Sci. 78, 147–157. https://doi.org/10.1016/j.jas.2016.12.005.
Toynbee, J.M.C., 1973. Animals in Roman Life and Art. Thames and Hudson, London, UK.
Uerpmann, H.-P., 2002. Dental Morphology of Horses, Donkeys and Mules. Presented at
the Horse. Donkey and Co., Basel, Switzerland.
White, K.D., 1970. Roman Farming. Cornell University Press, Ithaca, N.Y.
Willms, C., 1989. Zum Aussterben des europäischen Wildesels L’extinction de l'Ane sau-
vage européen. Germania 67, 143–148.
Yvinec, J.-H., 1998. Étude de la faune. In: Baillet-En-France (Val d'Oise), « La Vieille
Église » Habitat Rural Du Haut Moyen Âge. Document Final de Synthèse de Sauvetage
Urgent. SRAIF, Saint-Denis.
Zeder, M.A., 1986. The equid remains from Tal-e Malyan, Southern Iran. In: Meadow,
R.H., Uerpmann, H.-P. (Eds.), Equids in the Ancient World. Ludwig Reichert Verlag,
Wiesbaden, pp. 366–412.
Zeder, M.A., Lapham, H.A., 2010. Assessing the reliability of criteria used to identify
postcranial bones in sheep, Ovis, and goats, Capra. J. Archaeol. Sci. 37, 2887–2905.
https://doi.org/10.1016/j.jas.2010.06.032.
Zeder, M.A., Pilaar, S.E., 2010. Assessing the reliability of criteria used to identify
mandibles and mandibular teeth in sheep, Ovis, and goats, Capra. J. Archaeol. Sci.
37, 225–242. https://doi.org/10.1016/j.jas.2009.10.002.