Early Evidence for the Extensive Heat Treatment of Silcrete in the Howiesons Poort at Klipdrift Shelter (Layer PBD, 65 ka), South Africa

Abstract

Heating stone to enhance its flaking qualities is among the multiple innovative adaptations introduced by early modern human groups in southern Africa, in particular during the Middle Stone Age Still Bay and Howiesons Poort traditions. Comparatively little is known about the role and impact of this technology on early modern human behaviors and cultural expressions , due, in part, to the lack of comprehensive studies of archaeological assemblages documenting the heat treatment of stone. We address this issue through an analysis of the procedure used for heating and a technological analysis of a lithic assemblage recovered from one Howiesons Poort assemblage at Klipdrift Shelter (southern Cape, South Africa). The resulting data show extensive silcrete heat treatment, which adds a new dimension to our understanding of fire-related behaviors during the Howiesons Poort, highlighting the important role played by a heat treatment stage in the production of silcrete blades. These results are made possible by our new analytical procedure that relies on the analysis of all silcrete artifacts. It provides direct evidence of a controlled use of fire which took place during an early stage of core exploitation, thereby impacting on all subsequent stages of the lithic chaıˆnechaıˆne opé ratoire, which, to date, has no known equivalent in the Middle Stone Age or Middle Paleolithic record outside of southern Africa.

Full text

Early Evidence for the Extensive Heat Treatment of
Silcrete in the Howiesons Poort at Klipdrift Shelter
(Layer PBD, 65 ka), South Africa
Anne Delagnes, Patrick Schmidt, Katja Douze, Sarah Wurz, Ludovic
Bellot-Gurlet, Nicholas J. Conard, Klaus G. Nickel, Karen L. Van Niekerk,
Christopher S. Henshilwood
To cite this version:
Anne Delagnes, Patrick Schmidt, Katja Douze, Sarah Wurz, Ludovic Bellot-Gurlet, et al..
Early Evidence for the Extensive Heat Treatment of Silcrete in the Howiesons Poort at Klipdrift
Shelter (Layer PBD, 65 ka), South Africa. PLoS ONE, Public Library of Science, 2016, 11 (10),
pp.e0163874. <10.1371/journal.pone.0163874>.<hal-01399932>
HAL Id: hal-01399932
http://hal.upmc.fr/hal-01399932
Submitted on 21 Nov 2016

HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-
entific research documents, whether they are pub-
lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destin´ee au d´epˆot et `a la diffusion de documents
scientifiques de niveau recherche, publi´es ou non,
´emanant des ´etablissements d’enseignement et de
recherche fran¸cais ou ´etrangers, des laboratoires
publics ou priv´es.
Distributed under a Creative Commons Attribution 4.0 International License

RESEARCH ARTICLE
Early Evidence for the Extensive Heat
Treatment of Silcrete in the Howiesons Poort
at Klipdrift Shelter (Layer PBD, 65 ka), South
Africa
Anne Delagnes
1,2☯
*, Patrick Schmidt
3☯
, Katja Douze
1,2
, Sarah Wurz
2,4
, Ludovic Bellot-
Gurlet
5
, Nicholas J. Conard
3
, Klaus G. Nickel
6
, Karen L. van Niekerk
4,2
, Christopher
S. Henshilwood
2,4
1PACEA, CNRS—University of Bordeaux, Pessac, France, 2School of Geography, Archaeology and
Environmental Studies and Evolutionary Studies Institute, University of the Witwatersrand, Johannesburg,
South Africa, 3Department of Prehistory and Quaternary Ecology, Eberhard Karls University of Tu¨bingen,
Tu¨bingen, Germany, 4Department of Archaeology, History, Cultural Studies and Religion, University of
Bergen, Bergen, Norway, 5MONARIS, Sorbonne Universite
´s, UPMC Universite
´Paris 6, UMR 8233, Paris,
France, 6Department of Geosciences, Applied Mineralogy, EberhardKarls University of Tu¨bingen,
Tu¨bingen, Germany
☯These authors contributed equally to this work.
*a.delagnes@pacea.u-bordeaux1.fr
Abstract
Heating stone to enhance its flaking qualities is among the multiple innovative adaptations
introduced by early modern human groups in southern Africa, in particular during the Middle
Stone Age Still Bay and Howiesons Poort traditions. Comparatively little is known about the
role and impact of this technology on early modern human behaviors and cultural expres-
sions, due, in part, to the lack of comprehensive studies of archaeological assemblages
documenting the heat treatment of stone. We address this issue through an analysis of the
procedure used for heating and a technological analysis of a lithic assemblage recovered
from one Howiesons Poort assemblage at Klipdrift Shelter (southern Cape, South Africa).
The resulting data show extensive silcrete heat treatment, which adds a new dimension to
our understanding of fire-related behaviors during the Howiesons Poort, highlighting the
important role played by a heat treatment stage in the production of silcrete blades. These
results are made possible by our new analytical procedure that relies on the analysis of all
silcrete artifacts. It provides direct evidence of a controlled use of fire which took place dur-
ing an early stage of core exploitation, thereby impacting on all subsequent stages of the
lithic chaı
ˆne ope
´ratoire, which, to date, has no known equivalent in the Middle Stone Age or
Middle Paleolithic record outside of southern Africa.
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 1 / 27
a11111
OPEN ACCESS
Citation: Delagnes A, Schmidt P, Douze K, Wurz S,
Bellot-Gurlet L, Conard NJ, et al. (2016) Early
Evidence for the Extensive Heat Treatment of
Silcrete in the Howiesons Poort at Klipdrift Shelter
(Layer PBD, 65 ka), South Africa. PLoS ONE 11
(10): e0163874. doi:10.1371/journal.
pone.0163874
Editor: Nuno Bicho, Universidade do Algarve,
PORTUGAL
Received: December 19, 2015
Accepted: September 15, 2016
Published: October 19, 2016
Copyright: ©2016 Delagnes et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: All relevant data are
within the paper and in the Supporting Information
files.
Funding: We acknowledge support by Deutsche
Forschungsgemeinschaft and Open Access
Publishing Fund of University of Tu¨bingen. The
research of PS was funded by the Deutsche
Forschungsgemeinschaft (DFG) through the
research project Heat Treatment in the South
African MSA (Grant Nr: CO 226/25-1, MI 1748/2-1,
NI 299/25-1). Financial support was provided to

Introduction
The intentional heat treatment of silica rocksconstitutes a major technological milestone in
prehistory since the earliest developments of stone tool-making.It provides the first evidence
of a transformativetechnology, i.e. transforming the physicochemicalproperties of a material
for technical purposes, and it marks the emergence of fire engineering as a response to a variety
of needs that largely transcend hominin basic subsistence requirements. Heat treatment of
stone has long been documented in the prehistoric record as an intentional technical process
used to improve the working quality of silica rocks and to enhance the sharpness and straight-
ness the tool edges [1–3]. This technological process was reinvented many times in the Upper
Pleistocene and Early Holocene in various geographical contexts. Its first occurrence was
recently pushed back to more than 70 ka (thousand years) ago in the South African Middle
Stone Age (MSA) sequences from Pinnacle Point [4] and Blombos Cave [5]. The heat-treated
raw material was silcrete, a continental silica rock [6] of rather good quality that acquires even
better knapping quality when heated. The development of a fire-based transformative technol-
ogy adds a new component to the extensive list of inventive solutions introduced in the MSA,
in particular by Still Bay and Howiesons Poort groups [7,8]. However, southern African MSA
heat treatment of stone still remains poorly documented and much of the debate has focused
on the heating methods and on the induced physical transformations of the silcrete [4,9–12].
Conversely, very little is known about the role and impact of this new technology on early
modern human behaviors and cultural expressions. In other words, the role of heat treatment
in the MSA technological repertoire still has to be determined. The question remains whether
this early emergenceof stone heat treatment responds to a new set of specialized technological
skills or whetherit is part of the domestic sphere of activities. In this paper we address this
issue through the analysis of the heating technique and technological strategy developed in a
recently discovered and excavated MSA site: Klipdrift Shelter (KDS) (southern Cape region,
South Africa) [13], a site that indicates the extensive use of fire for the heat treatment of silcrete
within one discrete occupation layer of this site.
To date, evidence for heat treatment of stone has been described for a few South African
MSA sites: PinnaclePoint, Blombos Cave, DiepkloofRock Shelter and Mertenhof Shelter [4,5,
9,11,13,14]. They all are stratified sites with successive occupations devoted to a broad range
of subsistence, technical and symbolic activities [4,5,13,15–27]. Stone-heating is amongst the
multiple technological and domestic uses of fire performed by the Still Bay and Howiesons
Poort groups [28]. The widespread use of fire, including the repeated burning of living floors in
some sites [29,30], results in archaeological deposits that are extensively burnt. Thus, assessing
the role played by fire with more precision is complicated by the fact that a large number of
archaeologicalitems in these sites were incidentally burnt after use. In this context, a crucial
issue is to differentiate the post-discard burning events, resulting in what we call burnt artifacts,
from artifacts that were intentionally heat-treated, referred to as heated artifacts.
Silcrete is the only raw material for which chemical and structural modifications through
heat treatment by MSA groups has been observed in South Africa. While silcrete's thermal
transformations are comparable to the ones of flint, it has been shown experimentally that the
control of temperature and heating speed is significantly less demanding for silcrete than for
flint [11,31]. The presence of tempering-residue on some heated artifacts from Diepkloof fur-
ther indicates direct contact between embers and silcreteduring the heat-treatment and the use
of a fast-heating process [11]. These data provide an interesting alternative to the model of a
complex heating procedure involving the use of a sand-bath heatingenvironment [4,10]. Sil-
crete is a highly diverse category of rock,comprising a variety of typesthat differ in terms of
texture, homogeneity and color. The visual results of silcrete's heat-induced transformations
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 2 / 27
CSH for the excavation of the KDS site and
subsequent analysis by a National Research
Foundation/Department of Science and Technology
funded Chair at the University of the
Witwatersrand, South Africa and by the University
of Bergen, Norway. AD and KD have benefited from
the financial and logistical support of the
Evolutionary Studies Institute (University of the
Witwatersrand, Johannesburg) during their stay in
Cape Town.
Competing Interests: The authors have declared
that no competing interests exist.

vary significantly from one silcrete type to another, complicating the identification of heat
treatment. This difficulty is probably the reason why most studies on heat-treatment are lim-
ited to small quantities of so-called "diagnostic artifacts". A comprehensive understanding of
the technological behaviors related to the heat treatment of silcrete, its role and impact within
the lithic chaîne opératoire, is thus still missing.
In this paper we present a novel approach applied to all silcrete artifacts (>2 cm) from layer
PDB at KDS. Our study is based on a piece-by-piece comparison, a non-destructive analysis
that allows for the identification of all intentionally heated lithics. The methodological protocol
used to identify the heated pieces is combined with the characterization of the technological
stages or operations related to the heating process. This analysis provides new insights into the
technological use of fire by MSA groups.
Materials and Methods
Archeological context
Silcrete forms a significant component of the raw materials used for tool production in some of
the Howiesons Poort and Still Bay sites located on the western and southern Cape coasts of
South Africa. The reason for this is the proximity of this zone to the so-called silcrete Cape
coastal belt formed by the Cape Fold Mountains, where primary sources are relatively abun-
dant [32]. Fine-grained silcrete outcrops occur in several localities at a minimum distance of 10
km north of KDS.This archeological site ispart of the Klipdrift Complex,a series of sites
located along the Indian Ocean shoreline in the De Hoop Nature Reserve, southern Cape
region (Fig 1). KDS has been excavated since 2011 by C.S. Henshilwood and K.L. van Niekerk
[13] with a permit obtained from Heritage Western Cape, the Provincial Heritage Agency
based in Cape Town, South Africa. The research permits to conduct archaeological excavations
are issued under the National Heritage Resources Act (Act 25 of 1999) and the Western Cape
Provincial Gazette 6061, Notice 298 of 2003. CSH is the permit holder for the relevant permit:
HWC REF No. HM / OVERBERG / CAPE AGULHAS / DE HOOP NATURE RESERVEI /
KLIPDRIFT CAVE Permit No. 2010/06/001. No ethics clearance or permit is required to study
the lithic artifacts from the Klipdrift Complex. The specimens used in this article are in a public
collection that is accessible to other researchers. The lithics are housed and curated by the Iziko
Museums of South Africa in Cape Town, as well as at Wits satellite laboratory in Cape Town.
The lithics are catalogued under the labels: KDS 2011, 2012 and an individual record number
is attributed to each artifact. A permit for transport and destructive analysis of 12 KDS artifacts
studied here (Table 1) was provided by the South African Heritage Resources Agency
(SAHRA) (PERMIT NO: 9/2/079/0008). All destructive and spectroscopic analyses reported in
this paper fall under this permit. All necessary permits were obtained for the described study,
which complied with all relevant regulations.
The c. 1.2 m deep sequenceis dated to between c. 70 and 50 ka by single-grain Optically
Stimulated Luminescence (OSL) [13]. All layers contain well preserved archaeological assem-
blages, including marine and terrestrial faunal remains, organic materials, lithics, ochre and
engraved ostrich eggshells [13]. Layers PCA to PAY relate to the Howiesons Poort industry. Sil-
crete is the main raw material usedfor blade production in the lower part of the sequence (lay-
ers PCA, PBE, PBD). The silcrete is predominantly fine-grained. The other materials with good
knapping quality are fine-grained silica rocks (henceforth called chert) and calcrete. Both are
poorly represented whereas local hydrothermal quartz and quartzite together account for more
than half of the raw material used for knapping [13]. Considering that most hydrothermal
quartz and quartzite cannot be expected to show any significant modification of their internal
structure and knapping quality after heating, silcrete is the only good candidate for assessing
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 3 / 27

Fig 1. Map of the excavated area at KDS and site stratigraphy with OSL dates (illustration credit:
Magnus M. Haaland, university of Bergen).
doi:10.1371/journal.pone.0163874.g001
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 4 / 27

the significance and role of heat treatment at KDS. Layer PBD, dated to 64.6±4.2 ka, was
selected for this analysis because it contains the richest lithic assemblage (c. 2.500 artifacts >2
cm) excavated from a total surface of 4 m
2
including abundant silcrete artifacts. In layer PBD
silcrete accounts for 46% of the knapped component, the remaining raw material spectrum
being composed of quartzite (35.4%), hydrothermal quartz (17.1%), chert and calcrete (1.5%).
Blade production is the unique target of the silcrete reduction sequence (n = 531/862, 61.6% of
blades), while also being well developed on quartz (n = 150/316, 47,5%) but being quite mar-
ginal on quartzite (n = 81/653, 12.4%) (for more details, see [13]). The assemblage includes
end- and by-products from all stages of the chaîne opératoire, resulting from multiple on-site
technological activities ranging from blank production to tool use and discard.
Quantifying the prevalence of heat treatment of silcrete
When silcrete is heated, it undergoes several chemical and physical changes. These changes
include increased brittleness [33], occasional heat fracturing [34], reddening [35] and the loss
of porosity [11]. Thus, past heating of stone may be identified through archeometric tech-
niques. However, identifying these characteristics does not directly imply intentional heat
treatment because unintentional burning and taphonomic processes such as natural fires at the
site cause identical transformations in silcrete. Therefore, recognizinglithic heat treatment
must follow a line of reasoning that aims to show intentionality, for example, by demonstrating
that rocks were systematically and repeatedly knapped after their heat-treatment. In order to
investigate post-heat treatment knapping, the signature of a fracture that took place after heat
treatment (a post-heating scar) has to be recognized and distinguished from a fracture that
occurred in unheated material (a pre-heating scar). In silcrete, this distinction is generally pos-
sible because heat treatment causes structural transformations that allow the subsequent frac-
tures to propagate more evenly [11]. The resulting fracture surfaces show less micro- relief (i.e.
are smoother) than fracture surfaces on unheated silcrete. Hence, in principle, post-heating
removal scars can be distinguished from pre-heating scars by their relatively increased smooth-
ness. The approach to measure “gloss” [4] on silcrete artifacts for determining whether these
artifacts were heat-treated or not is based on the same principle. Gloss on knapped fracture
Table 1. Sample numbers and descriptions of the artifacts analyzed destructively and by IR
spectroscopy.
Accession number Observed heat treatment proxies Performed analysis
KB421 Pre-heating scar, post-heating scar, tempering-residue IR-ATR, Residue section
K2782 Post-heating scar, tempering-residue on natural surface Thin section, IR-ATR
KB565 Pre-heating scar, post-heating scar, HINC Thin section
K2840 Pre-heating scar, post-heating scar, tempering-residue Thin section, IR-ATR
KB576 Pre-heating scar, post-heating scar, HINC Thin section
K1938 Pre-heating scar, post-heating scar Thin section
KB1817 Pre-heating scar, post-heating scar Thin section
K2714 Pre-heating scar, post-heating scar Thin section
KDS-1 (unpl.) Pre-heating scar, post-heating scar Thin section
KDS-2 (unpl.) Pre-heating scar, post-heating scar, HINC Thin section
KDS-3 (unpl.) Pre-heating scar, post-heating scar Thin section
KDS-4 (unpl.) Pre-heating scar, post-heating scar Thin section
KDS-5 (unpl.) Post-heating scar, tempering-residue on natural surface IR-ATR
unpl. = not piece-plotted during excavation. HINC = Heat-induced non-conchoidal fracture
doi:10.1371/journal.pone.0163874.t001
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 5 / 27

surfaces is controlled by several factors, one of whichis the microscopic roughnessof the sur-
faces that causes incident light to be more or less reflected back.The light reflection, or gloss,is
therefore an indirect measure of the micro-topography of fracturesurfaces. However, the
reflection of light is also controlled by the refractive index and the absorption coefficient, two
material properties put into relation by the Fresnel equations [36]. Mineralogical composition
therefore influences the measured gloss values. For example, variations of the anatase (TiO
2
)
concentration, common in South African silcrete [6,32], influence the total light reflection
from the silcretesurface because of the relatively high refractiveindex of anatase. This can be
expected to result in different gloss values from two silcrete types, even though they have simi-
lar surface roughness values. This is why we prefer to directly observe the surface roughness
and use the terms ‘smooth’, as the opposite of ‘rough’ instead of ‘glossy’ and ‘dull’.
The apparent advantage of gloss analysis over the visual estimation of surface roughness is
that it produces absolute values that can be used to demonstrate whether an assemblage is
likely to contain heat-treated artifacts or not. However, single artifacts cannot be identified as
being heat-treated or not heated with this method. The reason for this is the structural het-
erogeneity of South African silcrete types. The silcrete types used for stone knapping in the
MSA range from very fine matrix-supported rocks with almost no inclusions to very coarse
grain-supported types with up to 4% TiO
2
, resulting in very different natural fracture pat-
terns and gloss values, even without heat treatment. Without the knowledge of the exact sil-
crete type, it is therefore not possible to interpret the gloss value obtained on a single artifact.
The same is true for a visual estimation of surface roughness without prior distinction of dif-
ferent silcrete types. Relatively rough silcrete types, even when heat-treated to high tempera-
tures, may display post-heating scars that are rougher than pre-heating scars of finer silcrete
types. The comparison of Fig 2A with Fig 2F illustrates this problem by showing the contrast
between a removal scar on a fine, but not heated piece of silcrete(Fig 2A) and a post-heating
scar on another, coarser but heat-treated piece of silcrete (Fig 2F). Here, the direct measure-
ment of gloss or the estimation of surface roughness, without the comparison with a mean-
ingful reference collection that contains the different silcrete types would have yielded an
erroneous result.
Such problems can be overcome if pre- and post-heating scars are preserved on a single arti-
fact: the most reliableway to identify post-heatingscars is to directly compare their smoothness
with relatively rougher pre-heating scars on the same artifact. However, such pieces are not the
majority in heat-treated assemblages, where a large number of artifacts show post-heating
scars only [9] and consequently no contrast in roughness. In these cases, the artifacts must be
compared to a geological reference sample of silcrete (unheated and experimentally heat-
treated) with the same structure and micro-facies, hence with the same fracture properties as
the artefact, in order to determine whether the observed flake scars are pre- or post-heating
scars. When choosing such reference samples, it is important to take into account the charac-
teristics that drive the different fracture patterns in silcrete: texture type, clast grain-size, crys-
tallography and grain-size of the matrix, inclusions and alterations of the texture, etc. This is
the approach that we apply to the study of the KDS silcrete artifacts.
Archeological reference samples
12 silcrete artifacts that represent the macroscopic variability of silcrete types in layer PBD
were selected for destructive analysis becauseof the presence of distinct markers of intentional
heat treatment. Petrographic thin sections were cut from 11 of them. Accession numbers and
brief descriptionsof the archaeologicalsamples analyzed destructivelyand by infrared(IR)
spectroscopy are summarized in Table 1.
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 6 / 27

Geological reference samples
In order to identify and to assess the presence/absence of flake scars from before and after heat
treatment on the PBD artifacts, we built a reference collection of unheated and experimentally
heat-treated flakes made from comparable silcrete types (S1A and S1B Fig). For this, we
Fig 2. Comparison between fracture surfaces from before and after heat treatment. (a, b, c) = before heat treatment, (d, e, f) = after heat treatment.
SK-13-03C before (a) and after (d) heat treatment; SK-13-04C before (b) and after(e) heat treatment; SK-13-03B before (c) and after (f) heat treatment.
Note that, regardless of the initial roughness of the silcrete type, heat treatment results in smoother fracture patterns but pre-heating fracture scars on fine-
grained silcrete (a) may be smoother than post-heating scars on coarser silcrete types (f). All fracture surfaces were photographed at identical magnification
and in similar raking light conditions.
doi:10.1371/journal.pone.0163874.g002
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 7 / 27

sampled different silcrete types in the vicinity of KDS. We collected 31 samples from six differ-
ent locations (Fig 3) and cut petrographic thin sections from each specimen in order to com-
pare them with the thin sections of the archaeological samples shown in Table 1. The purpose
of this sampling project was not to propose a raw material provisioning scheme for KDS but to
collect geological reference material with a comparable structure, and hence, with comparable
thermal behavior, for building the reference collection.
First, we compared the microfacies observed in the archaeological thin sectionswith the
microfacies observed in the 31 geological thin sections. Criteria for this comparison were: tex-
ture type [6], clast grain size and sorting; degree of rounding, dissolution and overgrowth;
cloudy iron oxide-,clay- or anatase-inclusions;presence/absence of indicators of Illuviation
like colloform features [6] or clay skins. Only if all these criteria are identical in two distinct
pieces of silcrete, do we expect that they contain similar amounts of molecular and chemically
bound water [11] and hence, that they show similar thermal behavior. We removed one flake
from each of thesereference samples and thenheat-treated the rest of the sample in an electri-
cal furnace at 450°C for 3h with a ramp of 4°C/min. 450°C is near the upper limit of experi-
mentally determined temperatures of heat treatment [9]. The atmosphere in the oven is
Fig 3. Map of the KDS area showing the six sampling locations for our reference collection (illustration credit: Gauthier Devilder,
PACEA/CNRS-university of Bordeaux). The black spots correspond to primary silcrete and ferricrete outcrops georeferenced by Roberts
(2003). The red x of source 5 indicates that this outcrop was not georeferenced by Roberts.
doi:10.1371/journal.pone.0163874.g003
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 8 / 27

expected to have no effect on the thermal transformations [11,37,38]. The ramp rate was cho-
sen arbitrarily because it does not influence the thermal transformations as long as the sample
is held at maximum temperature for >1h [39], it only controls the probability of heat-induced
fracturing [40]. A second set of flakes was removed from each sample after they had cooled to
room-temperature. The flakes produced were used to recognize pre-heating and post-heating
removal scars of the KDS PBD artifacts by directly comparing the latter with the corresponding
silcrete types of the reference collection (as determinedon the basis of macroscopic similarity).
Fig 4 shows three examples of matching thin section micrographs used to identify meaning-
ful reference samples. Because some of the geological samples showed almost identical
Fig 4. Micrographs of archaeological samples and geological samples with similar petrographic textures.
(a, c, e) = Archaeological samples and (b, d, f) = geological samples. Cross polarized light. (a) K2782 and (b) its
geological counterpart SK-13-01A, (c) K2714 and (d) its geological counterpart SK-13-03B, (e) K576 and (f) its
geological counterpart SK-13-04C.
doi:10.1371/journal.pone.0163874.g004
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 9 / 27

microfacies, we selected 17 of them that correlated best to the 11 artifacts from PBD (Table 1).
Control flakes were removed from these 17 reference blocks before they were heat-treated.
Two of the blocks shattered during heating while the other 15 did not break and were perfectly
suitable for knapping after the procedure. The fracture surfaces of the 15 flakes removed after
heat treatment differ from the fracture surfaces of the unheated control flakes in that they are
distinctively smoother. Fig 2 illustrates the transformation of the fracture pattern after heat
treatment for three silcrete samples with different grain-size.
Thus, our experimental reference collection contains 30 flakes from 15 reference samples
that are representative of the different silcrete types worked in PBD, i.e. one control flakewith
pre-heating scars and one heated flake with post-heating scars for each of the 15 samples. Dur-
ing the identification of heat treatment proxies on KDS lithics, these reference samples were
laid-out on a table and each artifact was matched to one of the 30 flakes on macroscopic
grounds. HINC fractures were identified on the basis of the criteria described in Schmidt et al.
[9] and the two geological samples that shattered during heat treatment.
Macroscopic identification of heat treatment proxies on artifacts
The identification of the heating proxies was performed for all silcrete pieces from layer PBD
by direct comparison with the experimental referencecollection. As proposed by Schmidt etal.
[9], we identified the following proxies (Fig 5):
Fig 5. Macroscopic heating proxies. A: pre-heating scar distinguishable by the contrast between relatively
rough and smooth surfaces, B: heat-induced non-conchoidal (HINC) fracture, characterized here by a non-
conchoidal fracture plane with scalar features, C: post-heating scars, D: potlid fractures.
doi:10.1371/journal.pone.0163874.g005
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 10 / 27

1. Pre-heating scar (Fig 5A): a relatively rough fracture surface corresponding to the removal
of a flake from unheated silcrete.
2. Post-heating scar (Fig 5C): a relatively smooth fracture surface corresponding to the removal
of a flake from heat-treated silcrete.
3. Heat-induced-non-conchoida l (HINC) fracture (Fig 5B): a fracture surface produced by fail-
ure during heating (sometimes also termed ‘overheating’ [40]). HINC fracture surfaces can
be recognized due to their strong surface roughness, the presence of scalar features on the
surface [9] and concave morphologies that often contain angular features. We only identify
this type of fracture surface as HINC when it is clearly cross-cut by at least one post-heating
removal. This technological relationship indicates that the failure occurred during heat
treatment, i.e. that the reduction was continued afterwards. In the opposite case, when such
a fracture surfaceis not cross-cut by a flake removal, it may result from fracturingat any
stage, e.g. during accidental burning after discard, and no technological information con-
cerning heat treatment can be retrieved from it. It is noteworthy that HINC fractures more
likely occur during fast heating and that they are almost absent when silcrete is heat-treated
with slow heating rates like in a sand-bath [9,10,40].
4. Tempering-residue: a black organic tar (wood tar) produced by dry distillation of plant exu-
dations that was deposited on the silcrete surface during its contact with glowing embers
during burning [9]. We only identify a black organic residue as tempering-residue if it cov-
ers a pre-heating removal scar or a natural surfaceand when it is cross-cut by a post-heating
removal scar, otherwise, the residue may result from taphonomic factors after discard of the
artifacts.
In addition to these proxies related to an intentional heating process, we also distinguish
potlid fractures (Fig 5D) or micro-cracks which are formed after the discard of the lithics and
are thus characteristic of incidental events of burning.
Infrared spectroscopy and microstructure of the identified residues
In order to verify that the residue we observed on KDS PBD artifacts is indeed an organic tar,
we conducted infrared analyses on four artifacts containing the residue. We used non-destruc-
tive micro-ATR infrared spectroscopy on their surfaces in order to test whether the black KDS
residue is an organic compound (Table 1). For this purpose, we acquired spectra in the spectral
region that contains the specific absorption bands of C-H bonds (C-H stretching bands)
because such bonds are characteristic for organic compounds. A Bruker IRscope II microscope
with a 20X germanium-ATR objective connected to an FT-IR Equinox 55 spectrometer was
used for this infrared ATR surface analysis (maximum size of the analyzed area 100 μm
2
, spec-
tra acquired between 2650 and 3150 cm
-1
, resolution 2 cm
-1
).
After determining the organic nature of the black residue, we analyzed its microstructure
and reflectance properties to confirm that it is wood tar. Because tempering-residue, as
described by Schmidt et al. [9], results from the distillation of plant exudations, it is deposited
on the silcrete surface as a hot liquid. This mode of deposition results in degassing pores that
may be preservedin the hardenedblack residue andthe residue may include micrometre-sized
charcoal inclusions due to the direct contact of the silcrete with glowing embers. The reflectiv-
ity of wood tar is also characteristic [41] and can be recognized in reflected light microscopy.
For analyzing these features, we cut sample KB421 perpendicularly to its pre-heating surface
covered with the black residue. The so obtained section was then embedded in resin and
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 11 / 27

polished to obtain a surface suitable for reflected light microscopy. We made microscopic
observations of the sections at magnifications ranging from 100x to 500x, using a Leitz petro-
graphic microscope setup for reflection and oil immersion.
Technological description of the archeological artifacts
Our analysis of the heat-treated silcrete artifacts from KDS is based on a piece-by-piece macro-
scopic identification of the heating proxies described above, in combination with technological
description. The identification of heating proxies was combined with a descriptive analysis of
each artifact based on technological and size attributes. The technological description allows
establishing the chronology of all knapping stages, from core preparation and blank production
to tool retouch or use,while determining the chronological position of the heating proxies in
this reduction sequence. The method developed here presents three key advantages: 1) it is
based on objectiveand easily observablecriteria; 2) it can be applied to complete assemblages;
3) it ensures reliable identification, by combining different heat treatment-related proxies with
the technological stage to which each lithic artifact corresponds. Our study thus provides a
complete characterization of the heated lithics from layer PBD at KDS.
Results
Scope of the silcrete heat treatment
In layer PBD, 92% of the silcrete pieces (n = 793/862 >2 cm) show clear evidence of heating,
while the remaining 8% (see Table 2) are mostly (7.2%) non-diagnostic pieces and 0.8% are
unheated. The non-diagnostic pieces comprise artifacts incidentally burnt after discard as well as
pieces corresponding to silcrete types for which the thermal transformations are not macroscop-
ically distinguishable because of the absence of the respective silcretetype in our reference collec-
tion (S1A and S1B Fig). We can thus deduce that silcrete in layer PBD was extensively, if not
entirely, heat-treated, which strongly argues for the intentionality of the heating process. By con-
trast, evidence of incidental burning is very low, with only 4% of artifactsburnt after discard
(n = 31/793 heated artifacts). The latter are characterized by heat-induced alterations in the form
of micro-cracks or potlid fractures, i.e. circular non-conchoidal hollows formed after the produc-
tion of the artifacts and/or rough non-conchoidal surfaces covering the entire blanks. The burnt
artifacts likely correspond to artifacts abandoned close to or in fireplaces after use.
The heated component in layer PBD (Table 2) includes all categories of end-products and
by-products. Blades (Fig 6) form the main category of end-product, and they are the exclusive
target of the silcrete reduction sequence [13]. Their production conforms with what has already
been described for other South African Howiesons Poort sites [42–46]. It is based on a
Table 2. Comparative frequencies of the heated vs unheated or non-diagnostic component and of the heating proxies for the basic technological
type-products.
All silcrete Blades Flakes Cores Chunks
N % N % N % N % N %
Post-heating removals 531 67 377 71 134 52.7 7 25 13 26.5
Pre-+post-heating removals 198 25 116 21.8 64 25.2 10 35.7 8 16.3
Heat-induced fracture 55 6.9 16 3 18 7.1 7 25 14 28.6
Tempering Residue 9 1.1 4 0.8 4 1.6 0 0 1 2.1
TOTAL heated 793 92 513 96.6 220 86.6 24 85.7 36 73.5
TOTAL unheated/non-diagnostic 69 8 18 3.4 34 13.4 4 14.3 13 26.5
TOTAL 862 100 531 100 254 100 28 100 49 100
doi:10.1371/journal.pone.0163874.t002
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 12 / 27

marginal percussion technique using a soft hammer for the extraction of unidirectional blades
from cores that are predominantly exploited on their larger flat surface. Blades are either
unmodified and likely used as such, or more rarely transformed into a variety of formal tools
(n = 24, 4.5% of all silcrete artifacts >2 cm): backed tools and segments (n = 6, Fig 6: 17–19),
blades with slight continuous retouch on one lateral edge (n = 8, Fig 6: 22, 23), notched blades
(n = 2, Fig 6: 20, 21), borers (n = 2), pièces esquillées (n = 2) and miscellaneous tools (n = 4). A
large majority of flakes are by-products of blade production (n =218, 86%), resulting from
core volume preparation, flaking surface and platform preparation, while only 14% (n = 36)
are elongated flakes with blade negatives on their dorsal face and belong to a blade production
stage. Chunks are waste products that result from knapping accidents either due to bad control
of the force and/or locationof the blow or to internal cracks in the raw material that have led to
core fragmentation.
Fig 6. Heated blades and tools on blades from KDS, layer PBD. 1–16: silcrete blades showing their diversity in terms of size attributes and visual
transformations after heating, 17–19: backed tools including one fragment of backed tool (17), one bi-truncated tool (18), one segment (19); 20,21:
notched blades; 22,23: blades with slight continuous retouch on one lateral edge.
doi:10.1371/journal.pone.0163874.g006
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 13 / 27

The number of cores (n = 28), together witha core/blade ratio of 1/19and a significant pro-
portion of cortical elements (23.5%), suggest that the knapping sequence was performed on
site, at least in part.A stage of core heating occurredduring this sequence, as 24 of the 28 cores
show evidence of heating before their reduction (Figs 7and 8and S2A–S2D Fig).
Among the diagnostic heating proxies observable on the artifacts, post-heating scars alone
are by far the most frequent (531/793, 67% of all heated artifacts). The technological composi-
tion of this group is consistent with the overall composition of the heated artifacts (Table 2). It
is indicative of long and extensive core reduction sequences that continue well beyond the first
reduction stages which removed the surfaces initially heated. Post-heating scars thus encom-
pass all stages of production, including core preparation (n = 69/531, 13%), blade production
(n = 407/531, 76.5%) and retouch (26/531, 5%). The intensity of core reduction is also
expressed by the fact that post-heatingremovals cover more than one half of the core flaking
surfaces for a large majority of the cores (19/24 cores). These data are consistent with an exten-
sive heating process that took place during an early stage of core exploitation and that impacted
all subsequent stages, basically targeted at producing blades.
Chronology of the heat treatment
Artifacts combining pre-heating and post-heating scars represent only 25% of all heated prod-
ucts, confirming the importance of not focusing solely on this category for assessing the role
played by the heat treatment in an assemblage. They are nevertheless highly informative for
assessing the chronological position of the heating stage relative to the whole lithic chaîne opér-
atoire, as they provide information on the stages that directly precede and follow the heating
stage (Table 2). The information here is more qualitative than quantitative as a significant pro-
portion of the pre-heating surfaces are technologically undeterminable(n = 134/189, 71% of all
artifacts with both pre-heating and post-heating scars, excluding the artifacts with pre-heating
surfaces corresponding to cortex). This is due to the very small size of the surface areas or to
the rough aspect of the pre-heating surfacesthat prevent any accurate characterization of the
corresponding scars. When readable, the pre-heating surfaces preserved on the products
belong to a bladeor flake production sequence(n = 41/189, 21.5%) or to a core preparation
stage (n = 11, 6%), which means that some cores were flaked beforeheat treatment. The analy-
sis of the cores confirmsthis observation. Some cores show pre-heating surfacesthat corre-
spond to blade or core preparation removals (Fig 7B and Fig 8C).Thus, in some cases the
heating stage occurred after an initial stage of reduction, involving a segmented chaîne opéra-
toire in which the heat treatment occurred during an early reduction stage, but not necessarily
during the first stage of core exploitation.
Technological advantages of heat-induced fracturing
Heat-inducedfractures are quite frequent (7%) in layer PBD, and are observedon all categories
of artifacts, including seven cores (Fig 7A and 7C and Fig 8A, 8B and 8C). A distinction has to
be made between heat-induced fractures that occurred at an early stage of core exploitation
and fractures that split cores or blanks into fragments at an advanced stage of transformation.
While the latter may be interpreted as accidentaland indicative of a total failure of the heating
process, the former may not necessarily compromise the knapping process. In layer PBD, all
artifacts with heat-induced fractures show negatives of removals which cross-cut the heat-
induced fracture planes, which means that the flaking stages were subsequent to the heat-
induced fracture of the core. Only a portion of the fragmented blocks actually preserves heat-
induced fractures, and in some cases knapping must have removed them, which suggests that
the real importance of heat-induced fractures is largely under-estimated in our accounts. A
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 14 / 27

Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 15 / 27

fairly large number of the raw material blocks must have therefore fractured during heat treat-
ment, which strongly supports a procedure involving fast heating rates, although the exact
quantitative pattern remains to be further experimentally tested.
The presence of silcrete cores that were exploited for blade production after an initial stage
of heat-induced fracturing, combined with the total absence of cores abandoned right after a
heat-induced fracturing, confirm that heat-induced fractures were not an obstacle for the KDS
artisans. Heat-induced fractures also led to the production of large angular fragments suitable
for initiating blade production with minimal core preparation. Portions of heat-induced frac-
ture planes are visible on the flakingsurface of the cores or on the platform preparation surface.
The knappers took advantage of the fracture planes to rapidly start the blade production
sequence, either by using the fracture plane as a platform for blade extraction without any fur-
ther preparation (Fig 8C) or by directly exploiting the large surfacecreated by the fracture for
producing blades after a preliminary stage of platform preparation (Fig 7A and 7C). Fracturing
of the block during heat treatment was thus a controlled risk, fully integrated in the core reduc-
tion sequence.
An additional advantage is that heat-induced fracture minimizes the risk of incidental frac-
ture of the cores through internal heterogeneities or inclusions at a more advanced knapping
stage. Iron oxide-hydroxide concentrations are common inclusions in the silcrete matrix [6],
locally causing lower coherence between quartz grains that can induce knapping accidents.
Such oxide-hydroxides produce H
2
O steam that may lead to preferentially fracturing the block
at the zones where they are highly concentrated [9,11]. The fragmentation may therefore be
seen as resulting in a ‘cleaning’ of the blocks from unwanted inclusions. The controlled frag-
mentation of lithic raw material in a fire is a technical process also described in ethnographic
records from North America [45,46], the Andaman Islands [47] and Zimbabwe [48]. In the
archaeological record, intentional heat-induced fragmentationof raw materials is also men-
tioned in the Sauveterrian culture of the French Mesolithic [49]. The intention of heat fractur-
ing in these other contexts was to split the blocks for further knapping rather than to optimize
the mechanical properties of the material.
At KDS, it is impossible to assess whether the fragmentation of the silcrete blocks was inten-
tional or whether the artisans opportunistically took advantage of the incidental fracturing of
the blocks prior to their exploitation. The fact that knappers did not stop the knapping process
after the core broke during heat-treatment suggests that it was, at least, an accepted risk. Heat-
induced fragmentation of the blocks results in smaller proportionsof chunks due to subsequent
breakages and consequently higherblade productivity as observed when comparing the com-
position of the heated group with the unheated or non-diagnostic silcrete group (Table 2). It
has been experimentally shown that larger volumes of heated raw material are associated with
higher risks of fracture during the heating process [50,51]. Core sizes are heterogeneous (Fig
9A and 9B), and range from very small sizes (maximum length <30 mm) to small sizes (maxi-
mum length >40 mm), with maximum length values around 50 mm. No direct correlation
between size attributes and heat-induced fractures can be shown from the small sample of
cores that show evidence for heat-induced fracturing (Fig 9A and 9B). As core dimensions only
document the block volumes at their stage of discard, blade dimensions provide useful and
Fig 7. Heated blade cores from layer PBD, KDS. Caption for drawings: 1. knapping platform preparation, 2. convexity
preparation, 3. blade removal without initiation, 4. blade removal with initiation, 5. indeterminate removal, 6. cortex, 7. pre-
heating surface, 8. heat-induced non-conchoidal fracture (HINC), 9. post-heating removal. Note that all three cores (A, B, C)
show a sequence of core exploitation (preparation and blade production) that follows a heat treatment which has resulted for A
and C in heat-induced fractures, and which was preceded for A and B by a first stage of core exploitation.
doi:10.1371/journal.pone.0163874.g007
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 16 / 27

Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 17 / 27

complementary information (Fig 9C and 9D). Blade dimensions confirm the high heterogene-
ity of the lithic production in terms of size attributes. The blades range from very small, c. 12–
13 mm maximal length, to large blades longer than 40 mm, with mean values of 26.4 mm (sd.
8.2), 11.3 mm (sd. 3.8)and 3.6 mm (sd.1.7) for length,width and thickness of the heated blades
respectively. The size distribution of the unheated blades does not differ significantly (Fig 9C
and 9D). Blades thus belong to a unique heterogeneous population, and no distinct bladelet
group can be seen, although a significant proportion of blades (30%) are smaller than 20 mm
in length. The diversity of blades is the result of a flexible system of blade production, as
observed from the cores that include a variety of types: unifacial cores, multifacial cores, semi-
rotating cores and cores exploited on their narrow face. We can deduce from this data that core
heat treatment corresponds to the objectives of a flexible but nonetheless highly purposeful
lithic production.
Fig 8. Heated blade cores from layer PBD, KDS. Caption for drawings: 1. knapping platform preparation, 2. convexity
preparation, 3. blade removal without initiation, 4. blade removal with initiation, 5. indeterminate removal, 6. cortex, 7. pre-
heating surface, 8. heat-induced non-conchoidal fracture (HINC), 9. post-heating removal, 10. patina. Note that all three
cores present heat-induced non-conchoidal fractures; the heat-induced fracture visible on core B produced two refitted
fragments, the biggest fragment was exploited as a core whereas the smallest was discarded. Core C shows a pre-heating
knapping stage that includes core preparation and blade production. The initial blanks of cores B and C are flakes
(V = ventral face, D = dorsal face).
doi:10.1371/journal.pone.0163874.g008
Fig 9. Size distribution of the cores and of the heated/unheated blades from KDS, layer PBD. A, B: length/width and length/
thickness distribution of the cores showing heat-induced fractures compared with the cores showing only pre- and post-heating
surfaces, C, D: length/width and length/thickness distribution of the heated blades comparedwith the unheated or non-diagnostic
ones.
doi:10.1371/journal.pone.0163874.g009
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 18 / 27

Tempering residue
A small number of artifacts (1.1%), comprising blades, flakes and chunks, exhibit a black resi-
due on pre-heating or natural surfaces that is visually similar to the tempering-residue
described in Schmidt et al [9]. The residual deposit is located either on a portion of the outer
natural surface of the initialblank or on a pre-heating surface. In either case, the residue-cov-
ered surfaces are cross-cut by post-heating removals. There is no evidence of residue on the
post-heating surfaces. The residue was thus deposited on these surfaces after an initial stage of
knapping and before the sequence of knapping that succeeded heat treatment, i.e. during the
heating stage itself. If the nature of this black residue is confirmed to be tempering-residue, it
indicates that silcreteblocks in KDS were heated in directcontact with embers in open-air
domestic fires, as recently reported at Diepkloof Rock Shelter [9].
Infrared spectroscopy and reflection microscopy of the tempering-residue. All four
infrared spectra of the black tempering-residue, covering pre-heating removal scars and natural
surfaces on the four residue bearing artifacts, clearly show C-H stretching bands (Fig 10). The
C-H band positions are not identical in all four samples but shift within a range of 10 cm
-1
,
revealing structural differences in the CH bearing molecules. These differences may be due to
various factors (temperature and duration of burning, differential taphonomic factors like oxi-
dation, different types of plants used for fuel, etc.) and their interpretation lies beyond the
scope of this work. However, the infrared spectra of the black deposit leave no doubtthat the
tempering residue observed on KDS artifacts is an organic compound.
Fig 10. Micro-ATR infrared spectra and reflection photo-micrographs of KDS tempering-residue. Left: infrared spectra in the CH-stretching region:
(a) K2782, (b) KB421, (c) K2840, (d) KDS-5. Spectra vertically offset for clarity. Right: reflection micrographs of a section of tempering residue on KB421.
Note the low reflectance value of the residue (appearing as grey), indicating that it is formed by tar, and the pores close to the silcrete surface, indicate melt
degassing during the formation of the tar as a hot liquid. Reflected white light and oil immersion.
doi:10.1371/journal.pone.0163874.g010
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 19 / 27

In the polished KDS section, the residue appears to be a 1-to-20 μm-thick film on the sil-
crete surface (Fig 10). The low reflectance values of the material composing this film are close
to the values known from wood tar [41]. Pores due to melt degassing are observable near the
silcrete surface, indicating a deposition of the film on the silcrete surface as a hot liquid.
Together with the organic nature, this microstructuresupports the model of formations of
tempering-residue as proposed by Schmidt et al. [9], i.e. the deposition of tar on the silcrete
surface through dry distillation of the plant exudations (gum or resin) of the wood that is
burned in the fire. Thus, our microscopic analysis of the residue on the KDS PBD artifacts sup-
ports that it is similar to the tempering residue on Diepkloof artifacts. However, no micro-
charcoal inclusions can be observedin the tar as is the case for the Diepkloof samples. Whether
this is due to differences in the plant species burnedat KDS or due to variable locations of the
silcrete within the hearths must await further microscopic analysis of supplementary artifacts
with tempering residue.
Discussion
Methodological progress
Our approach of using a reference collection of silcrete with similar microfacies, and therefore
similar thermal behavior, to identify heat treatment proxies on artifacts is a powerful non-
destructive tool. Its application to the KDS PBD assemblage has proven successful as it allows
the accurate characterization of all silcrete lithicsin an assemblage. The results obtained by this
technique are expected to be rather reliable because of the direct comparison between heated
and not heated pieces of the same silcrete type. This approach is especially useful when the
direct comparison between pre-heating and post-heating scars on a single artifact is not possi-
ble because onlyone of these types of removal scars is visible (i.e.the majority of artifacts).In
view of the variabilityof silcrete types in South Africa,this approach can be expected to be
more reliable than the use of a gloss meter [4].
However, 7.2% of all silcrete artifacts remained unidentified after our procedure. The reason
for this is that these silcrete types did not macroscopically match one of the silcrete types in our
reference collection, illustrating the need for exhaustive sampling of all silcrete types present in
the studied archaeological assemblage. Another potential limitation of this approach would be
long distance transport of silcrete [52]. Imported exogenous silcrete cannot be evaluated by our
protocol becauseof the lack of appropriate reference samples.
Technique used for heat treatment at KDS
During our comparison,we identified tempering-residue on some of the KDS artifacts. Struc-
tural and chemical analyses indicate that this residue has a similar origin as tempering-residue
in Diepkloof [9]: an organic wood tar deposited on the silcrete surface as a hot liquid during
heat treatment. The KDS PBD residue is associated with pre-heating removal scars or natural
surfaces, which wereboth part of the outer surface of the silcreteblocks during their heat treat-
ment. Post-heating-scars are free of black residue. Surfaces covered by the tempering residue
are cross-cut by post-heating removal scars, indicating that the silcrete was knapped after the
formation of the residue. This relation shows that the KDS tempering-residues do not result
from taphonomic burning after discard. Although the overall frequency of tempering-residue
appears low in layer PBD, it must be kept in mind that even if the total surfaceon a piece of sil-
crete is in contact with embers, only a small portion of this surface may be covered be temper-
ing residue [9]. Thus, a relatively small percentage of artifactsshowing traces of tempering
residue may still be indicative of a large number of silcrete pieces heat-treated in embers. The
model of ember-heat treatment is supported by the c. 7% of artifacts showing remnants of
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 20 / 27

heat-induced fractures. As explained earlier, 7% of all heat-treated artifacts that show remnants
of HINC fractures can be expected to indicate that a large number of the silcrete blocks brought
to the site heat-fractured during heat treatment. Because such heat-induced fracturing during
heat treatment is associated with fast heating rates like the ones produced in an open-air fire
[9,10] and because it ismuch rarer during slower sand-bath heating, thefinding of HINC frac-
tures on the PBD silcrete strongly suggests the use of a fast heating technique at KDS. It thus
appears highly likely that heat treatment in the embers of open-air fires was a technique com-
monly used in layer PBD. Additionally, heat treatment in embers is currently the only archaeo-
logically documented technique in the South African MSA [53] and at our given state of
knowledge it also appears to be the best model for explaining heat treatment at KDS.
Role and benefits of the heat treatment of silcrete at Klipdrift Shelter
The analysis performed on the lithics from layer PBD at KDS provides evidence of the exten-
sive and structured use of fire as a transformative technology for silcrete knapping by some
Howiesons Poort groups. It is a systematic procedure that results in more than 90% of the
assemblage being intentionally heated. The heating stage occurred non-randomly, in an early
stage of core exploitation, which was sometimes preceded by an initial knapping stage. As a
consequence, the whole chaîne opératoire, from core preparation to blade production and tool
manufacturing, benefitedfrom the advantages of the heating process. The heat treatment of a
lithic raw material from the earliest stages of lithic reduction, thereby impacting the entire lithic
chaîne opératoire, is a practice which appears again much later in the prehistoric record, at the
end of the Late Pleistocene [54].
The advantages of heat treating cores are multiple. As pointed out previously, heat treat-
ment greatly facilitates silcrete knapping. The reason for this is similarto that for flint and
chert whose heat treatment leads to a measurable reduction of the material’s fracture toughness
[50] and to an increased hardness [51]. This means that less force is needed to detach a flake or
blade after heat treatment, resulting in better control and precision during percussion. Improv-
ing the knapping quality of lithic raw material is a feature whichseems common to allprehis-
toric cultures that intentionally heated silica rocks[3,50,55]. It was also very likely a major
focus for some KDS artisans, as the extraction of blades using a soft hammer marginal percus-
sion requires high striking precision. Additional advantages which seem specific to this assem-
blage relate to the heat-induced fracturing of the blocks at an early stage of core exploitation.
These advantages include: 1) fragmentation, which results in the elimination of internal hetero-
geneities (iron oxide inclusions, remnants of illuviation features), which could have caused the
incidental breakage of the core at an advanced stage of reduction; 2) the production of angular
fragments with suitable angles and surfaces that can be directly exploited for knapping without
further preparation; 3) fewer constraints on the selection of the volumes to be heat-treated.
This heating procedure echoes the flexibility of the blade production system developed by the
knappers. The flexibilityis evident in the diversity of blade production methods and in the vari-
ability of blade sizeattributes. Our analysis of the lithics from layer PBD therefore shows that
optimal control of the heating procedure, which would include avoiding breakage, was not a
prerequisite for the KDS artisans (e.g.[10]).
Layer PBD relates to the intermediate Howiesons Poort phase as initially definedat Diepk-
loof Rock Shelter [22]. At KDS, this phase differs significantly from the late Howiesons Poort
phase in tool types and raw material frequencies [13]. Silcrete is the predominant raw material
and notched tools are the most common tools in the intermediate Howiesons Poort phase,
while hydrothermal quartz is dominant in the late Howiesons Poort phase, in association with
higher proportions of backed tools [13]. However, many elements of continuity exist from one
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 21 / 27

layer to another at KDS. In particular blade production remains the primary objective of the
lithic production in the whole Howiesons Poort sequence and silcrete heating appears as a con-
stant feature, as observed from our preliminary analysis of the layers below (PCA, PBE) and
above (PBC, PBA/PBB, PAZ, PAY) layer PBD. As the proportion of silcrete artifacts decreases
in the upper part of the KDS sequence, it might be expected that heat treatment also becomes
less prevalent.
Heat treatment of silcrete during the MSA
The extensive heat treatment of silcrete in the Howiesons Poort is not only found at KDS. The
recent analyses of the intermediate Howiesons Poort assemblages from Diepkloof Rock Shelter
also indicate high proportions of heat-treated artifacts [9], over 90%, and thus similar to those
observed at KDS in layer PBD. Evidence of heated cores and products in the Howiesons Poort
layers of Pinnacle Point (PP5-6 site) suggests that silcrete heat treatment was also used at this
site for blade production [4,20,56]. While detailed data are still required for in-depth compar-
ative analyses of theheating practices and related technological behaviors, it seems clearthat
some Howiesons Poort groups who occupiedthese sites routinely heat-treated cores for pro-
ducing blades. The KDS and Diepkloofrecords together suggest that the Intermediate Howie-
sons Poort phase, to which layer PBD belongs, would correspond to a period of extensive
development of silcrete heat treatment for blade production.Considering that heatingof lithic
raw materials has only recently been described [4], further analyses are still required to test the
presence of heating practices in a widerrange of sites and time periods.
Silcrete heating by Still Bay artisans comprises a strategy that is different to that in the
Howiesons Poort. At Blombos Cave, it was shown that the heating stage focused on tool
manufacturing, inparticular for the pressure retouch of the apical part of some Still Bay points
[5]. Experiments have shown that silcrete pressure retouching cannot be performed without
prior heat treatment of the raw material. Here, the benefit of heating is to allow and ensure the
successful application of a specifictechnique that contributes to the final retouch of the apical
part of points. Compared to the heating process in layer PBD at KDS, the heating process in
the Still Bay at Blombos Cave is far more targeted. Although significantly different, the Still Bay
and Howiesons Poort heating practices share the common feature of improving the knapping
qualities of silcretefor the productionof specific sought-afterend-products, such as bifacial
points and blades. Stone heating in the MSA is embedded in the on-site domestic activities that
revolve around fire. It is worth noting that, at KDS, silcrete was collected from within a maxi-
mum radius of a few kilometers to a few tens of kilometers from the site whereas at Blombos
Cave the silcrete sources were, according to our knowledge of present sources, located further
away.
Heating stone in prehistoric times: a recurrent and discontinuous
practice
The Howiesons Poort heatingpractices, as definedin layer PBDat KDS, at DiepkloofRock
Shelter and Mertenhof Shelter, are unique and cannot be directly correlated with stone-heating
practices documented in other prehistoric contexts. It is only from c. 20 ka that the heat treat-
ment of stone developedin Asia and Europe [57]. Both in terms of heating method and techno-
logical achievement, the MSA heating practices differ significantly from the heating practices
developed in the late Upper Pleistocene. In the Eurasian late Upper Pleistocene record, the
emergence of flint heat treatment most likely implied setting up more complex heating struc-
tures, allowing a strict control of the heating temperature and rate [54,57–61]. Beyond the
recurrent association between heat treatment and industries targeting normalized end-
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 22 / 27

products, such as blades, micro-blades or thin bifacial points, there is a significant technologi-
cal gap between the MSA fire-based industries and the late Upper Pleistocene/Holocene fire-
based cultures in Eurasia.
It has not been demonstrated so far that the MSA groups used complex heating methods.
The presence of tempering residue, indicating heat treatment in direct contact with glowing
embers, suggests that this procedure was often realized with a fast process using open fires.
Another proxy strengthening the hypothesis of a fast heating technique is the fact that some of
the artifacts show signs of fracturing during heat treatment (heat-induced fractures), because
thermal fracturing more likely occurs during fast heating [10,40]. The fire-based technology
developed at KDS did not require highly specialized skills and was likely performed as part of
on-site domestic activities. In the context of southern Africa, data are still too limited for
exploring potential patterns of continuity or evolution across the MSA.
Conclusion
Our data add a new dimension to the understanding of fire-related behaviors during the
Howiesons Poort by demonstrating the major role played by heat treatment in the production
of silcrete blades in layer PBD at KDS. It provides the first direct evidence of the intentional
and extensive use of fire applied to a whole lithic chaîne opératoire, based on an analytical
approach that has allowed the analysis of all heated lithics from layer PBD. The heat treatment
of silcrete in this layer has impacted all stages of core reductionand all subsequent operations
of tool manufacturing.For the artisans, the benefitsof heat treatment performedin an early
stage of the chaîne opératoire were multiple. The Howiesons Poort groups considerably devel-
oped and optimized a technology that had possibly emerged from the early MSA [4], and con-
tinued in the Still Bay (c. 77–72 ka) in relation to the Still Bay point manufacturing process[5].
Although further evidenceis required, our analysis indicates that the heat treatment of sil-
crete was a major asset for someof the MSA groups who occupied theCape region, by facilitat-
ing blade production in an area where the most abundant fine-grained raw material was
difficult to exploit for such purposes.The heat treatment of silcretethus reflects an innovative
adaptation based on optimal use of local resources. The adaptable and innovative attitude was
likely one factorthat favored the widespreaddistribution of the Howiesons Poort materialcul-
ture in southern Africa. Fire-based domestic activities were particularly well-developed in
Howiesons Poort sites, and relate to various activities including site maintenance, hearth clean-
ing and burning of bedding [29,30]. The regular occurrence of domestic hearths in the Howie-
sons Poort probably had a causal effect on the significant development of fire-based
technological activities including heating for the preparation of compound-adhesives used for
hafting tools [62–65] and making composite tools and stone-tipped arrows [66,67]. Addition-
ally, the heat treatment of silcrete, applied to the entire chaîne opératoire process, can be con-
sidered a significant innovation and was part of a package of new adaptive skills using fire that
has no equivalent in the earlier Middle Stone Age or in the Middle Paleolithic.
Supporting Information
S1 Fig. Sample of silcrete reference collection used for comparison. A, B: illustration of
experimental flakes struck before and after the heating of each block.
(PDF)
S2 Fig. Silcrete cores from layer PBD. A, B, C, D: Picture and technological drawing of each
core. Caption for drawings: 1. knapping platform preparation, 2. convexity preparation, 3.
blade removal without initiation, 4. blade removal with initiation, 5. indeterminate removal, 6.
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 23 / 27

cortex or patina, 7. pre-heating surface, 8. heat-inducedfracture, 9. post-heating removal, 10.
burnt after discard, potlids, 11. retouch.
(PDF)
Acknowledgments
We acknowledge support by Deutsche Forschungsgemeinschaft and Open Access Publishing
Fund of University of Tübingen. The research of PS was funded by the Deutsche Forschungsge-
meinschaft (DFG) through the research project Heat Treatment in the South African MSA
(Grant Nr: CO 226/25-1, MI 1748/2-1, NI 299/25-1). Financial support was provided to CSH
for the excavation of the KDS site and subsequent analysis by a National Research Foundation/
Department of Scienceand Technology funded Chair at the University of the Witwatersrand,
South Africa and by the University of Bergen, Norway. AD and KD have benefited from the
financial and logistical support of the Evolutionary Studies Institute (University of the Witwa-
tersrand, Johannesburg) during their stay in Cape Town. AD also wants to thank the Labex
LaScArBx, which is a research programme supported by the ANR (ANR-10-LABX-52), for its
support. We all warmly thank Petro Keene, Lisa Hulett and Samantha S. Mienies for their con-
stant help and assistance at the Wits satellite laboratory in Cape Town during the analytical
work, as well as Jens Axel Frick (University of Tübingen) for his help with knapping the refer-
ence collection, Simone Kaulfuß (University of Tübingen) for preparing the polished section
for our microscopic analysis, Magnus Mathisen Haaland (university of Bergen) and Gauthier
Devilder (PACEA—university of Bordeaux) for their last minute assistance in the achievement
of the figures. The content of this article has also greatly benefited from the comments of Peter
Hiscock, Yamandu H. Hilbert and of three anonymous reviewers.
Author Contributions
Conceptualization:AD PS KD SW.
Data curation: AD PS.
Formal analysis: AD PS KD SW.
Funding acquisition: PS NJC CSH.
Investigation: AD PS KD SW KLVN CSH.
Methodology: AD PS.
Project administration: CSH.
Resources: NJC CSH.
Supervision: CSH.
Visualization: AD PS KD.
Writing – original draft: AD PS.
Writing – review & editing: AD PS KD SW LBG NJC KGN KLVN CSH.
References
1. Crabtree DE, Butler BR. Notes on experiments on flint knapping. Tebiwa. 1964; 7(1):1–6.
2. Flenniken JJ, Garrison EG. Thermally Altered Novaculite and Stone Tool Manufacturing Techniques.
Journal of Field Archaeology. 1975; 2:125–31. doi: 10.2307/529623
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 24 / 27

3. Mandeville MD, Flenniken JJ. A Comparison of the Flaking Qualities of Nehawka Chert Before and
After Thermal Pretreatment. Plains Anthroplogist. 1974; 19:146–8.
4. Brown KS, Marean CW, Herries AIR, Jacobs Z, Tribolo C, Braun D, et al. Fire As Engineering Toolof
Early Modern Humans. Science. 2009; 325:859–62. doi: 10.1126/science.1175028 PMID: 19679810
5. Mourre V, Villa P, Henshilwood CS. Early Use of Pressure Flaking on Lithic Artifacts at Blombos Cave,
South Africa. Science. 2010; 330:659–62. doi: 10.1126/science.1195550 PMID: 21030655
6. Summerfield MA. Petrography and diagenesis of silcrete from the Kalahari Basin and Cape coastal
zone, Southern Africa. Journal of Sedimentary Research. 1983; 53(3):895–909. doi: 10.1306/
212f82e2-2b24-11d7-8648000102c1865d
7. Henshilwood CS. Late Pleistocene techno-traditions in southern Africa: a review of the Still Bay and
Howiesons Poort, c. 75–59 ka. Journal of World Prehistory. 2012; 25:205–37. doi: 10.1007/s10963-
012-9060-3
8. McBrearty S, Brooks A. The revolution that wasn’t: a new interpretation of the origin of modern human
behavior. Journal of Human Evolution. 2000; 39:453–563. doi: 10.1006/jhev.2000.0435 PMID:
11102266
9. Schmidt P, Porraz G, Bellot-Gurlet L, February E, Ligouis B, Paris C, et al. A previously undescribed
organic residue sheds light on heat treatment in the Middle Stone Age Journal of Human Evolution.
2015; 85:22–34. doi: 10.1016/j.jhevol.2015.05.001 PMID: 26073074
10. Wadley L, Prinsloo L. Experimental heat treatment of silcrete implies analogical reasoning in the Mid-
dle Stone Age. Journal of Human Evolution. 2014; 70:49–60. doi: 10.1016/j.jhevol.2013.11.003 PMID:
24709137
11. Schmidt P, Porraz G, Slodczyk A, Bellot-Gurlet L, Archer W, Miller CE. Heat treatment in the South
African Middle Stone Age: temperature induced transformations of silcrete and their techological impli-
cations. Journal of Archaeological Science. 2013; 40:3519–31. doi: 10.1016/j.jas.2012.10.016
12. Bentsen SE. Controlling the heat: an experimental approach to Middle Stone Age pyrotechnology.
South African Archaeological Bulletin. 2013; 68(198):137–45.
13. Henshilwood CS, van Niekerk KL, Wurz S, Delagnes A, Armitage SJ, Rifkin RF, et al. KlipdriftShelter,
southern Cape, South Africa: preliminary report on the Howiesons Poort layers. Journal of Archaeolog-
ical Science. 2014; 45:284–303. doi: 10.1016/j.jas.2014.01.033
14. Schmidt P, Mackay A. Why Was Silcrete Heat-Treated in the Middle Stone Age? An Early Transforma-
tive Technology in the Context of Raw Material Use at Mertenhof Rock Shelter, South Africa. PLoS
ONE. 2016; 11(2):e0149243. doi: 10.1371/journal.pone.0149243 PMID: 26867145
15. Henshilwood CS, d’Errico F, Marean CW, Milo RG, Yates R. An early bone tool industry from the Mid-
dle Stone Age at Blombos Cave, South Africa: implications for the origins of modern human behaviour,
symbolism and language. Journal of Human Evolution. 2001; 41:631–78. doi: 10.1006/jhev.2001.0515
PMID: 11782112
16. Henshilwood CS, d’Errico F, Yates R, Jacobs Z, Tribolo C, Duller GAT, et al. Emergence of modern
human behavior: Middle Stone Age engravings from South Africa. Science. 2002; 295:1278–80. doi:
10.1126/science.1067575 PMID: 11786608
17. d’Errico F, Henshilwood C, Vanhaeren M, Van Niekerk K. Nassarius kraussianus shell beads from
Blombos Cave: evidence for symbolic behaviour in the Middle Stone Age. Journal of Human Evolution.
2005; 48(1):3–24. doi: 10.1016/j.jhevol.2004.09.002 PMID: 15656934
18. Henshilwood CS, d’Errico F, Watts I. Engraved ochres from the Middle Stone Age levels at Blombos
Cave, South Africa. Journal of Human Evolution. 2009; 57(1):27–47. doi: 10.1016/j.jhevol.2009.01.005
PMID: 19487016
19. Henshilwood CS, d’Errico F, Van Niekerk KL, Coquinot Y, Jacobs Z, Lauritzen SE, et al. A 100,000-
year-old ochre-processing workshop at Blombos Cave, South Africa. Science. 2011; 334(6053):219–
22. doi: 10.1126/science.1211535 PMID: 21998386
20. Brown KS, Marean CW, Jacobs Z, Schoville BJ, Oestmo S, Fisher EC, et al. An early and enduring
technology originating 71,000 years ago in South Africa. Nature. 2012; 491:590–3. doi: 10.1038/
nature11660 PMID: 23135405
21. Archer W, Gunz P, Van Niekerk K, Henshilwood C, McPherron SP. Diachronic change within the Still
Bay at Blombos Cave, South Africa. PLoS ONE. 2015; 10(7). doi: 10.1371/journal.pone.0132428
PMID: 26134976
22. Porraz G, Texier PJ, Archer W, Piboule M, Rigaud JP, Tribolo C. Technological successions in the
Middle Stone Age sequence of Diepkloof Rock Shelter, Western Cape, South-Africa. Journal of
Archaeological Science. 2013; 40(9):3376–400. doi: 10.1016/j.jas.2013.02.012
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 25 / 27

23. Texier PJ, Porraz G, Parkington JE, Rigaud JP, Poggenpoel C, Miller C, et al. A Howiesons Poort tradi-
tion of engraving ostrich eggshell containers dated to 60,000 years ago at Diepkloof Rock Shelter,
South Africa. PNAS. 2010; 107(14):6180–5. doi: 10.1073/pnas.0913047107 PMID: 20194764
24. Dayet L, Texier PJ, Daniel F, Porraz G. Ochre resources from the Middle Stone Age sequence of
Diepkloof Rosk Shelter, Western Cape, South Africa. Journal of Archaeological Science. 2013; 40
(9):3492–505. doi: 10.1016/j.jas.2013.01.025
25. Igreja M, Porraz G. Functional insights into the innovative Early Howiesons Poort technology at Diepk-
loof Rock Shelter (Western Cape, South Africa). Journal of Archaeological Science. 2013; 40(9):3475–
91. doi: 10.1016/j.jas.2013.02.026
26. Marean CW, Bar-Matthews M, Bernatchez J, Fisher E, Goldberg P, Herries AIR, et al. Early human
use of marine resources and pigment in South Africa during the Middle Pleistocene. Nature. 2007;
449:905–8. doi: 10.1038/nature06204 PMID: 17943129
27. Villa P, Soressi M, Henshilwood CS, Mourre V. The Still Bay points of Blombos Cave (South Africa).
Journal of Archaeological Science. 2009; 36(2):441–60. doi: 10.1016/j.jas.2008.09.028
28. Bentsen SE. Using pyrotechnology: fire-related features and activities with a focus on the African Mid-
dle Stone Age. Journal of Archaeological Research. 2014; 22:141–75. doi: 10.1007/s10814-013-9069-
x
29. Miller CE, Goldberg P, Berna F. Geoarchaeological investigations at Diepkloof Rock Shelter, Western
Cape, South Africa. Journal of Archaeological Science. 2013; 40:3432–52.
30. Goldberg P, Miller CE, Schiegl S, Ligouis B, Berna F, Conard N. Bedding, hearths, and site mainte-
nance in the Middle Stone Age of Sibudu Cave, KwaZulu-Natal, South Africa. Archaeological and
Anthropological Sciences 2009; 1:95–122. doi: 10.1007/s12520-009-0008-1
31. Mercieca A, Hiscock P. Experimental insights into alternative strategies of lithic heat treatment. Journal
of Archaeological Science. 2008; 35(9):2634–9. doi: 10.1016/j.jas.2008.04.021
32. Roberts DL. Age, genesis and significance of Soutth African coastal belt silcretes. Pretoria: Council for
Geoscience; 2003.
33. Domanski M, Webb JA. Effect of heat treatment on siliceous rocks used in prehistoric lithic technology.
Journal of Archaeological Science. 1992; 19(6):601–14. doi: 10.1016/0305-4403(92)90031-w
34. Mercieca A. Burnt and broken: An Experimental Study of Heat Fracturing in Silcrete. Australian
Archaeology 2000; 51:40–7. doi: 10.1080/03122417.2000.11681679
35. Schindler DL, Hatch JW, Hay CA, Bradt RC. Aboriginal Thermal Alteration of a Central Pennsylvania
Jasper: Analytical and Behavioral Implications. American Antiquity. 1982; 47(3):526–44. doi: 10.2307/
280233
36. Bass M. Handbook of optics. Volume 1, Geometrical and physical optics, polarized light, components
and intruments. 3 ed. New York: McGraw-Hill; 2009. 1251 p.
37. Schmidt P, Badou A, Fro
¨hlich F. Detailed FT near-infrared study of the behaviour of water and hydroxyl
in sedimentary length-fast chalcedony, SiO2, upon heat treatment. Spectrochimica Acta Part A: Molec-
ular and Biomolecular Spectroscopy. 2011; 81(1):552–9. doi: 10.1016/j.saa.2011.06.050 PMID:
21763183
38. Schmidt P, Masse S, Laurent G, Slodczyk A, Le Bourhis E, Perrenoud C, et al. Crystallographic and
structural transformations of sedimentary chalcedony in flint upon heat treatment. Journal of Archaeo-
logical Science. 2012; 39(1):135–44. doi: 10.1016/j.jas.2011.09.012
39. Schmidt P, Paris C, Bellot-Gurlet L. The investment in time needed for heat treatment of flint and chert.
Archaeological and Anthropological Sciences 2015:1–10. doi: 10.1007/s12520-015-0259-y
40. Schmidt P. What causes failure (overheating) during lithic heat treatment? Archaeological and Anthro-
pological Sciences 2014; 6(2):107–12. doi: 10.1007/s12520-013-0162-3
41. Taylor GH, Teichmu¨ller DA, Diessel CFK, Littke R, Robert P. Organic Petrology. Berlin: Gerbru¨ der
Borntraeger; 1998. 704 p.
42. Soriano S, Villa P, Wadley L. Blade technology and tool forms in the Middle Stone Age of South Africa:
the Howiesons Poort and post-Howiesons Poort at Rose Cottage Cave. Journal of Archaeological Sci-
ence. 2007; 34:681–703. doi: 10.1016/j.jas.2006.06.017
43. Villa P, Soriano S, Teyssandier N, Wurz S. The Howiesons Poort and MSA III at Klasies River main
site, Cave 1A. Journal of Archaeological Science. 2010; 37:630–55. doi: 10.1016/j.jas.2009.10.028
44. Wurz S. Variability in the Middle Stone Age Lithic Sequence, 115,000–60,000 Years Ago at Klasies
River, South Africa. Journal of Achaeological Science. 2002; 29:1001–15. doi: 10.1006/jasc.2001.
0799
45. Hester TR. Ethnographic Evidence for the Thermal Alteration of Siliceous Stone. Tebiwa. 1972;
15:63–5.
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 26 / 27

46. Mandeville MD. A consideration of the thermal pretreatment of chert. Plains Anthropologist. 1973;
18:177–202.
47. Man EH. On the Aboriginal Inhabitants of the Andaman Islands. (Part I.). The Journal of the Anthropo-
logical Institute of Great Britain and Ireland. 1883; 12:69–116. doi: 10.1038/032409a0
48. Robinson TR. A Survival of Flake-technique in southern Rhodesia. Man. 1938; 38(224).
49. Guilbert R. Les systèmes de de
´bitage de trois sites sauveterriens dans le sud-est de la France. Bulletin
de la Socie
´te
´Pre
´historique Franc¸aise. 2003; 100(3):463–78.
50. Domanski M, Webb JA, Boland J. Mechanical properties of stone artefact materials and the effect of
heat-treatment. Archaeometry. 1994; 36:177–208. doi: 10.1111/j.1475-4754.1994.tb00963.x
51. Schmidt P, Masse S, Laurent G, Slodcyk A, Le Bourhis E, Perrenoud C, et al. Crystallographic and
structural transformations of sedimentary chalcedony in flint upon heat treatment. Journal of Archaeo-
logical Science. 2012; 39:135–44. doi: 10.1016/j.jas.2011.09.012
52. Nash DJ, Coulson S, Staurset S, Ullyott JS, Babutsi M, Hopkinson L, et al. Provenancing of silcrete
raw materials indicates long-distance transport to Tsodilo Hills, Botswana, during the Middle Stone
Age. Journal of Human Evolution. 2013; 64(4):280–8. doi: 10.1016/j.jhevol.2013.01.010 PMID:
23453438
53. Schmidt P. The ’sand-bath’ and lithic heat treatment in the South African Middle Stone Age: myth or
reality? African Archaeological Review. 2016.
54. Flenniken JJ. The Paleolithic Dyuktai pressure blade technique of Siberia. Artic Anthropology. 1987;
24(2):117–32.
55. Tixier J, Inizan ML. L’e
´mergence des arts du feu: le traitement thermique des roches siliceuses.
Pale
´orient. 2000; 26(2):23–36. doi: 10.3406/paleo.2000.4707
56. Brown KS. The Sword in the Stone: Lithic Raw Material Exploitation in the Middle Stone Age at Pinna-
cle Point Site 5–6, Southern Cape, South Africa. Cape Town: University of Cape Town; 2011.
57. Bordes F. Traitement thermique du silex au Solutre
´en. Bulletin de la Socie
´te
´Pre
´historique Franc¸aise.
1969; 66(7):197. doi: 10.3406/bspf.1969.10404
58. Collins MB. Observations on the Thermal Treatment of Chert in the Solutrean of Laugerie Haute,
France. Proceedings of the Prehistoric Society. 1973; 39:461–3. doi: 10.1017/s0079497x00011774
59. Binder D, Perlès C. Strate
´gies de gestion des outillages lithiques au Ne
´olithique. Pale
´o. 1990; 2:257–
83.
60. Le
´a V. Raw, Pre-heated or ready to use: discovering specialist supply systems for flint industriesin
mid-Neolithic (Chasse
´en) communities in southern France. Antiquity. 2005; 79:51–65.
61. Aubry T, Almeida M, João Neves M, Walter B. Solutrean Laurel Leaf point production and raw material
procurement during the Last Glacial Maximum in southern Europe: two examples from central France
and Portugal. In: Soressi M, D H.L., editors. Multiple approaches to the study of bifacial technologies.
Philadelphia: University of Pennsylvania Museum of Archaeology and Anthropology; 2003. p. 165–82.
62. Wadley L. Compound-adhesive manufacture as a behavioural proxy for complex cognition in the Mid-
dle Stone Age. Current Anthropology. 2010; 51(Suppl. 1):S111–S9. doi: 10.1086/649836
63. Charrie
´-Duhaut A, Porraz G, Cartwright CR, Igreja M, Connan J, Poggenpoel C, et al. First molecular
identification of a hafting adhesive in the Late Howiesons Poort at Diepkloof Rock Shelter (Western
Cape, South Africa). Journal of Archaeological Science. 2013; 40:3506–18. doi: 10.1016/j.jas.2012.
12.026
64. Lombard M. Direct evidence for the use of ochre in the hafting technology of Middle Stone Age tools
from Sibudu Cave. Southern African Humanities. 2006; 18(1):56–67.
65. Wadley L, Hodgskiss T, Grant M. Implications for complex cognition from the hafting of tools with com-
pund adhesives in the Middle Stone Age, South Africa. Proceedings of the National Academy of Sci-
ences. 2009; 106(24):9590–4. doi: 10.1073/pnas.0900957106. PMID: 19433786
66. Lombard M, Haidle MN. Thinking a bow-and arrow set: Cognitive implications of Middle Stone Age
bow and stone-tipped arrow technology. Cambridge Archaeological Journal. 2012; 22(2):237–64. doi:
10.1017/s095977431200025x
67. Lombard M, Phillipson L. Indications of bow and stone-tipped arrow use 64,000 years ago in KwaZulu-
Natal, South Africa. Antiquity. 2010; 84:635–48.
Extensive Heat Treatment of Silcrete at 65 ka
PLOS ONE | DOI:10.1371/journal.pone.0163874 October 19, 2016 27 / 27