Journal of

Archaeological
SCIENCE
Journal of Archaeological Science 30 (2003) 1393–1399
http://www.elsevier.com/locate/jas

A practical approach to the identification of low temperature

heated bone using TEM
H.E.C. Koon*, R.A. Nicholson, M.J. Collins
Ancient Biomolecules Group, Drummond Building, University of Newcastle, Newcastle upon Tyne NE1 7RU, UK
Received 29 November 2002; received in revised form 7 February 2003; accepted 13 February 2003

Abstract

Unlike burnt bone, cooked bone leaves no obvious trace. The ability to identify cooked bone has far reaching implications within
the fields of palaeo-anthropology, archaeology and forensic science. There is, however, only limited literature on low temperature
heated bone. This is not due to a lack of significance, but from an absence of any means of identifying such heating at temperatures
insufficient to cause charring. The situation is further complicated by the fact that diagenetic alteration to bone may mimic heat
induced changes.
A previous Transmission Electron Microscopy (TEM) study showed that heat induced morphological changes to the collagen
fibrils occur after low temperature heating of fish bone. This paper investigates whether these findings could be replicated on
mammal bone. A series of experiments were carried out using sheep humeri. These results were compared with cooked and
uncooked bones recovered from experimental burials, representing a variety of different environments (moorland, woodland and
garden soil). Morphological changes to the fibrils were seen following only very mild heating events, such as short-term roasting of
fleshed bone. However, similar changes were observed in unheated bone which had been buried in a low pH (3.5–4.5) soil for 7 years.
Within a given burial cooked and uncooked bone was easily distinguishable. The technique, therefore, has direct application in
forensic studies and may be of value in distinguishing heated from unheated bone within a given archaeological assemblage.
 2003 Elsevier Ltd. All rights reserved.

Keywords: Bone collagen fibrils; IR; Low temperature heating; TEM; Diagenesis; cooked bone

1. Introduction therefore remains undetected. If minimal heating of

The social and practical implications of the tech-
nological breakthrough heralded by the first use of fire
cannot be underestimated [17]. Evidence of heated bone
in later pre-history can be used not only to imply human
influence, but also to elucidate dietary habits, cooking
techniques [24] and socio-cultural or funerary practices
[16]. Unfortunately, it is only possible to identify the
presence of heated bone if charred, from abrasive
features (pot polish) or by inference from the presence
of associated features, such as hearths [8]. This means
that there is a wealth of evidence not being utilised from
boiled and roasted bone, which though heated, does
not reach a temperature that will induce charring and
* Corresponding author. Tel.: +44-78137-13205.
E-mail address: hannah.koon@ncl.ac.uk (H.E.C. Koon).
0305-4403/03/$ - see front matter  2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0305-4403(03)00034-7
bone could be identified perceptions of the past might
change. As an example, establishing cannibalism in the
archaeological record is a topic which remains of great
interest and debate especially with regard to sites in the
American Southwest [15,24] and Fiji [4]. Recognising
cooked human bone could not only help to distinguish
war related mutilation or secondary burial from canni-
balism [7] but the reported incidence of cannibalism
would arguably increase [6].
The importance of identifying heated bone from
archaeological sites has led to a number of methods of
identification. There are three main stages through
which a bone passes as it is heated; first water is boiled
away, followed by combustion of the organic matter and
finally modification and decomposition of the mineral.
Methods have identified changes in the appearance of
the bone as well as detecting thermal alteration within
1394 H.E.C. Koon et al. / Journal of Archaeological Science 30 (2003) 1393–1399
the organic and mineral phases. At a macroscopic level,
studies have focused on the observation of thermally
induced colour changes and patterns of warping and
cracking on the bone surface [11,20]. Within the bone
microstructure, the chemical modification and loss to
the organic fraction of bone has been investi-
gated through analytical techniques monitoring carbon,
hydrogen, and nitrogen (CHN) concentrations [13,22],
as well as glycine/glutamic acid (gly/glu) ratios and
ammonia (NH3) levels [23]. Changes to the mineral have
also been identified using x-ray diffraction [20] and
infrared spectroscopy [22].
The situation however is complicated by the fact that
weathering processes cause similar alteration to bone,
such as cracking, cortical exfoliation and colour change
to the bone surface [1], an increase in crystal size and
organisation within the bone mineral [22,25], a loss of
collagen and modification to the remaining amino acid
composition [9,26]. Indeed a recent study failed to
discriminate between boiled and buried bone [19].
Furthermore, attempting to detect evidence of cook-
ing provides an additional problem in that when heating
food the objective is to retain the moisture. This being
the case it is unlikely that the bone would reach even the
initial stage of thermal alteration where water is boiled
out of the bone especially if the surrounding flesh acts as
an insulator from the external temperature. Conse-
quently, the aforementioned changes used to identify
heated bone will only occur at temperatures greatly
beyond those achieved during domestic roasting or
boiling.
In this study we explore an alternative approach,
namely monitoring heat-induced changes to the packing
of the collagen fibril. Evidence from previous studies of
mineralised sheep tendon [21] and fish bone [18] suggest
that mild heating leads to disorganisation of mineralised
collagen fibrils, observable using TEM. It was unclear if
mammal bone collagen would react in a similar way, nor
if the alterations could be unequivocally attributed to
heating, or could also be caused by diagenesis (cf. [19]).
This paper comprises an investigation to monitor
changes to collagen fibrils in sheep bones subjected to
low temperature heating, burial for 7 years, or both.
Primarily TEM analysis was used to identify alterations
to the collagen component, supplemented with infrared
investigation to monitor any corresponding changes to
the mineral component.
2. Materials and methods
To obtain fresh heated bone samples the humeri of
fleshed sheep forelimbs of approximately equal size were
used. The specimens (1–8) were placed on a preheated
domestic oven and exposed to temperatures of 180 (C
and 220 (C, over different time durations. The bone
temperatures were measured using a thermocouple
attached between the meat and the bone. After heating
all of the samples were left to cool for 15 min before
being reweighed and then defleshed using a scalpel.
Six samples were also taken from sheep metapodials
which had been previously buried. The samples utilised
for this study came from three different post deposition
environments; A1 and A2 were buried in a garden soil, a
worked clay loam which had been previously limed to
raise the pH to neutral; B1 and B2 were buried in a
deciduous woodland, in a moderately drained stagno-
gleic brown earth soil, with pH 3.5–4.5; C1 and C2 were
buried in the Yorkshire Moors into a humus iron-pan
stagnopodzol with pH 3.5–4.5. Three of these samples
(A2, B2 and C2) had been subjected to the same heating
regime: they were all boiled for one hour (including
20 min for water to boil), prior to burial for 7 years. The
three remaining samples were not heated, but had also
been buried for the same 7-year period [12].
Approximately 5 mm3 sections were removed using a
hacksaw, from both the experimentally heated and
previously buried specimens. Finally, the samples were
placed in a solution of 0.5 M HCl and stirred con-
tinuously on a Spiro-mixer at 4 (C. This low tempera-
ture was employed to minimise destruction of the
collagen during demineralisation, caused by hydrolysis
from the low pH of the acid [2]. To control for deminer-
alisation comparative preparations were also conducted
using 0.1 M EDTA with the pH adjusted to neutral
(pH 7) using 0.5 M NaOH.
Once demineralised the insoluble ‘collagen’ residue
was washed thoroughly with distilled water, to remove
salt and soluble gelatine before being cut into 2 mm3
cubes with a scalpel. TEM preparation was adapted
from Richter [18]. Samples were placed in a boiling tube
with approximately 3 ml phosphowolframacid, which
had been adjusted to pH 7 with 0.5 M NaOH. In order
to liberate the collagen fibrils the boiling tube was placed
in a beaker of ice and the solution macerated (Ultra-
Turrax) for 3 min. In order to maintain a low tempera-
ture maceration was conducted in 30 s bursts with 30 s
intervals between. The head of the macerator and the
forceps used to transport the samples were washed
thoroughly with distilled water before processing each
new sample, to minimise cross contamination.
The resulting solutions were transferred to Falcon
tubes and centrifuged on a Minstral 3000 centrifuge at
3000 g and 4 (C for 15 min. Afterwards the super-
natants were discarded and the remaining pellets were
re-suspended in 1 ml of phosphowolframacid and mixed
using an Autovortex mixer. A 0.5 ml drop from each of
the solutions was pipetted on to three formwar grids and
allowed to air dry for 5 min, after which time the excess
liquid was removed and the grids dried with filter paper.
A piece of filter paper was placed on to a petri dish
and dampened with a 50:50 solution of an ethanol/water
mix. A portion of dental wax was placed on the filter
 H.E.C. Koon et al. / Journal of Archaeological Science 30 (2003) 1393–1399 1395

Table 1
A summary table representing the number of fibrils counted and classified (excluding any ambiguous fibrils) once the samples had been
demineralised. It also shows the results of the infrared analyses and a brief description of the conditions that each sample was subjected to prior
to analyses

Sample Brief description Unaltered fibrils Altered fibrils SF C/P

Beaded Dumbbell
1 Unheated fresh control 85 2 0 2.72 0.38
2 Unheated fresh control 73 2 0 – –
3 Heated at 220 (C to 60 (C over 6 min 79 6 0 – –
4 Heated at 180 (C to 60 (C over 14 min 43 26 1 – –
5 Heated at 220 (C to 80 (C over 16 min 28 10 0 – –
6 Heated at 180 (C to 80 (C over 18 min 13 13 1 2.68 0.45
7 Heated at 180 (C to 96 (C over 60 min 26 72 2 2.64 0.51
8 Heated at 220 (C to 100 (C over 30 min 12 67 4 2.59 0.63
A1 Unheated and buried in garden soil pH 7 87 0 0 – –
A2 Heated and buried in garden soil pH 7 50 28 7 – –
B1 Unheated and buried in woodland soil pH 3.5–4.5 83 17 0 – –
B2 Heated and buried in woodland soil pH 3.5–4.5 7 4 2 – –
C1 Unheated and buried in moors pH 3.5–4.5 42 26 2 – –
C2 Heated and buried in moors pH 3.5–4.5 1 38 8 – –

paper and a 0.5 ml drop of uranyl acetate (2% solution
in 50:50 ethanol/water mix) was suspended on top. The
formwar grids were then floated on this solution, speci-
men side down, for 30 min, during which time the
system was covered to protect it from light. Each set of
three grids (per sample) were finally analysed under a
JEOL 1200EX transmission electron microscope.
Infrared analysis was also carried out on four samples
from the heating experiments. These were chosen as they
displayed the most alteration to the collagen fibrils
under TEM. An unheated control sample was also
analysed.
Bone samples were crushed into powder using a pestle
and mortar, before being mixed with potassium bromide
and pressed into discs [5]. Analysis was carried out using
Fourier Transform Infrared Spectroscopy. The infrared
‘splitting factor’ (SF), a measure of apatite crystallinity,
was calculated using the splitting ratio of the Phosphate

4 doublet at 567 cm
1 and 605 cm
1 after Weiner and
Bar-Yosef [26]. The Carbonate/Phosphate Ratio (C/P)
was also calculated from the infrared spectrum using the
peaks at 1415 cm
1 (CO2
3) and 1035 cm
1 (PO3
4)
[27].
3. Results and discussion
3.1. Infrared analyses
Infrared analyses were unable to discriminate
between the samples, as none had crystallinity (SF)
values higher than 2.8 (Table 1), the value associated
with fresh modern bone [26]. This suggests that no
alteration had occurred to the mineral component of
these bone samples as a result of low temperature
heating. Extensive heating leads to an increase in SF and
a decrease in the carbonate/phosphate peak ratio
(C/P) [19].
3.2. Observed changes to the collagen fibril
TEM analyses of the heated fresh bone samples
identified unaltered collagen fibrils, together with fibrils
exhibiting localised areas of swelling both along the
fibrils and at the ends. These are the areas described by
Richter [18] as ‘melted’ and represent regions where
localised unpacking of the fibril has occurred. Fibrils
were identified as collagen from their distinct staining
pattern with a characteristic periodicity, and fibril
diameters of between 55–65 nm. Statistical comparison
was performed on the most intact and cleanest of each of
the three prepared grids. Fibrils were classed using the
following criteria, adapted from observations made by
Richter [18].
3.2.1. Unaltered
A collagen fibril with no evidence of ‘melting’. The
fibrils should have an equal diameter throughout
(Fig. 1A).
3.2.2. Altered, beaded or dumbbell
There are two categories of alteration utilised in
this study to describe a modified fibril. Both have
amorphous areas connected by strands that retain a
normal collagen-banding pattern. A beaded fibril
(Fig. 1B) has one or more localised areas of melting both
along the fibril and at the end. A dumbbell (Fig. 1C)
describes a short (typically <3 µm) collagen fibril that
has bulbous areas of melting at both ends. This probably
1396 H.E.C. Koon et al. / Journal of Archaeological Science 30 (2003) 1393–1399

Fig. 1. Four TEM images highlighting the different forms of alteration observed in collagen fibrils after a heating event. A, an unheated collagen
fibril. B, a beaded shaped heated collagen fibril. C, a dumbbell shaped heated collagen fibril. D, amorphous material likely to be denatured collagen.

arises from a parting of the fibril at the beaded region

and thus is a more advanced state of degradation.
Amorphous material was also observed in samples
from the heating experiments (Fig. 1D). Although there
was no clear evidence of banding indicative of native
collagen similar material was identified by Richter [18]
as fully denatured collagen.
The pattern of deterioration observed in these exper-
iments confirms the earlier findings by Richter [18], with
an apparent progression from beaded fibrils to dumb-
bells and amorphous collagen with greater exposure to
heat (Figs. 2 and 3).

3.3. Fresh heated bone

In all cases the fleshed bone reached a substantially

lower temperature than that achieved by the oven itself
(Fig. 2a). Also on visual inspection no signs of thermal
alteration (warping, cracking or colour change) were
apparent on the bone surface. Nevertheless a clear
relationship is observed between the heating profile and
Fig. 2. A. Temperature profiles recorded from roasted bone, the
numbers 3–8 refer to the fresh heated bone samples, for further details
see Table 1. B. Changing proportion of altered collagen fibrils plotted
against integrated heating time (assuming an activation energy of
173 kJ mol
1). Again the numbers 3–8 refer to the fresh heated bone
samples and 2 refers to an unheated fresh control, for further details
see Table 1.
 H.E.C. Koon et al. / Journal of Archaeological Science 30 (2003) 1393–1399
Fig. 3. A stacked bar chart showing the relationship between increased temperature and the percentage of altered bone collagen fibrils, using TEM
analyses of samples heated until the bone temperature reached 60 (C (3–4), 80 (C (5–6), 96 (C (7) and 100 (C (8). Two unheated fresh samples (1–2)
are also included as controls.
the rise in the number of damaged fibrils. This is
demonstrated in Fig. 3 where samples have been
arranged in ascending order on the basis of the maxi-
mum bone temperature that each attained. To confirm
this the total amount of thermal damage (estimated by
integrating the heating curve and assuming an activation
energy of 173 kJ mol
1; Collins unpublished data) was
calculated for each experiment (Fig. 2b). These values
were then normalised to a constant 100 (C and reported
as times@100 (C. This relationship indicates that the
observed alteration of fibrils is not an artefact of sample
preparation and confirms that the deterioration of col-
lagen is highly temperature dependent (e.g. [3,10,14]).
The rapidity of thermal alteration detectable by TEM is
remarkable, given that Roberts et al. [19] conclude
that in order to fashion measurable change in cooked
bone (such as change in mineralogy, loss of collagen)
prolonged (>9 h) boiling would be required.
Despite the many similarities in the morphological
changes observed by this study and that of Richter [18],
the rates of change in experimentally heated sheep bone
are somewhat slower than those of fish (Pleuronectes
platessa). In our study several unaltered fibrils were
identified after roasting for 60 min during which time
the bone temperature reached in excess of 95 (C
(w323s@100 (C). However, Richter found that at 60 (C
after 30 min (w2s@100 (C) most of the fibrils exhibited
alteration and at 100 (C after the same duration all the
collagen had ‘melted’ [18]. This is anticipated due to the
lower thermal stability of fish bone collagen [28].
1397
3.4. Buried bone
Fig. 4 shows the results of the TEM analyses from six
weathered bone samples. Significantly, two of the un-
heated weathered samples (B1 and C1, both from acid
soil) displayed the same types of alteration to their
collagen fibrils as that seen in the fresh heated samples.
This is perhaps not surprising; we estimate that burial
for 7 years at 10 (C is thermally equivalent to short
boiling (w15s@100 (C). However, the unheated sample
buried in a neutral pH soil for the same 7-year period
displayed no detectable alteration. The combined data
from these three experiments indicated that collagen
fibril deterioration is not solely controlled by time/
temperature but is also influenced by the burial environ-
ment. Indeed it is striking that the relative degree of
collagen fibril deterioration mirrors the observed surface
degradation of the buried bone itself. Bone buried in
neutral clay loam soil lacked surface alteration, whilst
considerable cortical exfoliation and fungal damage was
evident in bones buried in both the brown earth (pH
3.5–4.5) and the humus iron-pan stagnopodzol soil (pH
3.5–4.5) [12].
Roberts et al. (2002) [19] state that more than nine
hours of boiling would be needed to show physical or
chemical alteration in bone. In our burial experiments
however, bones were only boiled for one hour. Our
method reveals significant degradation of the fibrils in
all of the boiled samples. This finding is anticipated by
the results from our roasting experiments, the most
1398 H.E.C. Koon et al. / Journal of Archaeological Science 30 (2003) 1393–1399
Fig. 4. A stacked bar chart comparing the effect of burial and of heating and burial, on the percentage of alteration to bone collagen fibrils. This
is achieved using TEM analyses of sets of heated weathered (A2, B2 and C2) and unheated weathered (A1, B1 and C1) bone samples, from three
different deposition environments, for further sample details see Materials and methods.
intense of which was only equivalent to boiling for
10 min. It is not clear why the collagen fibrils for the
bones which had been boiled prior to burial were not
significantly more damaged than those from the heated
fresh bones, considering they were exposed to 100 (C
for significantly longer. This needs further investigation
and may reflect different effects induced by boiling and
roasting, or preferential loss of the damaged fibrils
over the period of 7 years in the burial environment.
Comparison of the paired boiled and unheated buried
samples (Fig. 4) showed that in all cases TEM analysis
could discriminate the heated bones.
4. Conclusion
This study provides an intriguing insight into the
initial deterioration of a collagen fibril. Changes are
detected by TEM as unpacking and fragmentation of the
collagen fibril, but as yet it is not known what under-
lying process drives this change. The alterations occur
during the mild heating which accompanies cooking and
well before any gross collagen loss or change in mineral
crystallinity are apparent. The study has also highlighted
the problem of identifying bone that has been heated
prior to burial. The reason for this is that similar
alteration occurs to bone as a consequence of both
prolonged or extreme subsurface diagenesis and heating
processes. As a result, TEM analyses of collagen fibrils
would be of value in discriminating between heated
and unheated bone from a single (well-preserved)
archaeological assemblage. Within a forensic timescale,
however, heated bone could be identified without the
need for comparative material.
Acknowledgements
This paper results from work carried out for an MSc
dissertation at the University of Bradford. We wish to
acknowledge the help and support of all those involved
in this project. In particular, we wish to thank J.
Fearnley of the Department of Biomedical Sciences,
University of Bradford, for his time and assistance in
using the TEM; C. Smith and S. Roberts of the Ancient
Biomolecules Research Group and the Chemical
Analytical Services Unit, University of Newcastle upon
Tyne, for preparing and running SF and C/P analyses.
References
[1] A.K. Behrensmeyer, Taphonomic and ecologic information from
bone weathering, Paleobiology 4 (1978) 150–162.
[2] M.J. Collins, P. Galley, Towards an optimal method of archaeo-
logical collagen extraction: the influence of pH and grinding,
Ancient Biomolecules 2 (1998) 209–222.
[3] M.J. Collins, M.S. Riley, A.M. Child, G. Turner-Walker, A basic
mathematical simulation of the chemical degradation of ancient
collagen, Journal of Archaeological Science 22 (1995) 175–183.
[4] D. Degusta, Fijian cannibalism and mortuary ritual: bioarchaeo-
logical evidence from Vunda, International Journal of
Osteoarchaeology 10 (2000) 76–92.
[5] M.J. DeNiro, S. Weiner, Chemical, enzymatic and spectroscopic
characterisation of ‘collagen’ and other organic fractions from
prehistoric bones, Geochimica et Cosmochimica Acta 52 (1988)
2197–2206.
 H.E.C. Koon et al. / Journal of Archaeological Science 30 (2003) 1393–1399
[6] S.A. Hurlbut, The taphonomy of cannibalism: a review of the
anthropogenic bone modification in the American southwest,
International Journal of Osteoarchaeology 10 (2000) 4–26.
[7] P.M. Lambert, B.R. Billman, B.L. Leonard, Explaining varia-
bility on mutilated human bone assemblages from the
American southwest: a case study from the Southern Piedmont of
Sleeping Ute Mountain, Colorado, International Journal of
Osteoarchaeology 10 (2000) 49–64.
[8] J.G. McDonnell, Pyrotechnology, in: D.R. Brothwell, A.M.
Pollard (Eds.), Handbook of Archaeological Sciences, Wiley and
Sons, Chichester, 2001, pp. 493–505.
[9] P.M. Masters, Preferential preservation of non-collagenous pro-
teins during bone diagenesis: Implications of chronometric and
stable isotopic measurements, Geochimica et Cosmochimica Acta
51 (1987) 3209–3214.
[10] C.A. Miles, A.J. Bailey, Thermally labile domains in the collagen
molecule, Micron 32 (2001) 325–332.
[11] R.A. Nicholson, A morphological investigation of burnt animal
bone and an evaluation of its utility in archaeology, Journal of
Archaeological Science 20 (1993) 411–428.
[12] R.A. Nicholson, Bone degradation, burial medium and species
representation: debunking the myths, and experiment-based
approach, Journal of Archaeological Science 23 (1996) 513–533.
[13] R.A. Nicholson, Bone degradation in a compost heap, Journal of
Archaeological Science 25 (1998) 393–403.
[14] C.M. Nielsen-Marsh, R.E.M. Hedges, T. Mann, M.J. Collins, A
preliminary investigation of the application of differential
scanning calorimetry to the study of collagen degradation in
archaeological bone, Thermochimica Acta 365 (2000) 129–139.
[15] S.A. Novak, D.D. Kollmann, Perimortem processing of human
remains among the Great Basin Fremont, International Journal
of Osteoarchaeology 10 (2000) 65–75.
[16] T. Oestigaard, Sacrifices of raw, cooked and burnt humans,
Norwegian Archaeological Review 33 (2000) 41–58.
[17] C. Renfrew, P. Bahn, Archaeology: Theories, Methods and
Practice, Thames and Hudson, London, 1996, pp. 240–241.
1399
[18] J. Richter, Experimental study of heat induced morphological
changes in fish bone collagen, Journal of Archaeological Science
13 (1986) 477–481.
[19] S.J. Roberts, C.I. Smith, A. Millard, M.J. Collins, The
taphonomy of cooked bone: characterizing boiling and its
physico-chemical effects, Archaeometry 44 (2002) 485–494.
[20] P. Shipman, G.F. Foster, M. Schoeniger, Burnt bones and teeth:
an experimental study of colour, morphology, crystal structure
and shrinkage, Journal of Archaeological Science 11 (1984)
307–325.
[21] J.M. Snowden, J.F. Weidemann, A morphological and biochemi-
cal examination of the hydrothermal denaturation of collagen,
Meat Science 2 (1976) 1–18.
[22] M.C. Stiner, S.L. Kuhn, S. Weiner, O. Bar-Yosef, Differential
burning, recrystallisation, and fragmentation of archaeological
bone, Journal of Archaeological Science 22 (1995) 223–237.
[23] R.E. Taylor, P.E. Hare, T.D. White, Geochemical criteria for
thermal alteration of bone, Journal of Archaeological Science 22
(1995) 115–119.
[24] C.G. Turner, J.A. Turner, Man Corn: Cannibalism and Violence
in the Prehistoric American Southwest, University of Utah Press,
Salt Lake City, 1999, pp. 16–21.
[25] N. Tuross, A.K. Behrensmeyer, E.D. Eanes, L.W. Fisher, P.E.
Hare, Molecular preservation and crystallographic alterations
in a weathering sequence of wildebeest bones, Applied
Geochemistry 4 (1989) 261–270.
[26] S. Weiner, O. Bar-Yosef, States of preservation of bones
from Prehistoric sites in the Near East: a survey, Journal of
Archaeological Science 17 (1990) 187–196.
[27] W.E. Wright, H.P. Schwarcz, Infrared and isotopic evidence for
diagenesis of bone apatite at Dos Pilas, Guatemala: Palaeodietry
implications, Journal of Archaeological Science 23 (1996)
933–944.
[28] J. Woodhead-Galloway, Collagen: the Anatomy of a Protein.
Studies in Biology no. 117. Edward Arnold, London, 1980.