Journal of Archaeological Science 34 (2007) 1824e1829

http://www.elsevier.com/locate/jas

Evaluating bone collagen extraction methods for stable

isotope analysis in dietary studies
Marie Louise S. Jørkov a,*, Jan Heinemeier b, Niels Lynnerup a
a
Laboratory of Biological Anthropology, Institute of Forensic Medicine, University of Copenhagen, Blegdamsvej 3,
DK-2200 N Copenhagen, Denmark
b
AMS 14C Dating Centre, Department of Physics and Astronomy, University of Aarhus, Aarhus, Denmark
Received 6 October 2006; received in revised form 21 December 2006; accepted 22 December 2006

Abstract

The specific purpose of this study was to compare three different collagen extraction methods commonly used in isotope laboratories con-
ducting dietary studies. We evaluated their resultant differences in d13C and d15N, collagen quality and collagen yield. Our study was based on
well-preserved skeletal material from the medieval period in Denmark. Our study shows that there is a systematic significant difference in the
yield and the d13C values between the three methods. Using the method of DeNiro and Epstein [DeNiro, M.J., Epstein, S., 1981. Influence of diet
on the distribution of nitrogen isotopes in animals. Geochimica et Cosmochimica Acta 45, 341e351] with NaOH as cleaning agent, will, ac-
cording to our study, give d13C values that are on average 0.32& more positive than using the ultra-filtration method [Brown, T.A., Nelson,
D.E., Vogel, J.S., Southon, J.R., 1988. Improved collagen extraction by modified Longin method. Radiocarbon 30 (2), 171e177, modified in
Richards, M.P., Hedges R.E.M., 1999. Stable isotope evidence for similarities in the types of marine foods used by late Mesolithic humans
at sites along the Atlantic coast of Europe. Journal of Archaeological Science 26, 717e722]. The third method, which is a modified version
of the second method, excluded the ultra-filtration step. This method seems to give d13C values that lie in between the other methods. Our study
did not show any significant difference in d15N values. Although the differences between the methods are very small, we conclude that the use of
stable isotope analysis in food determination studies requires adherence to routine methods for preparing and measuring samples.
Ó 2007 Elsevier Ltd. All rights reserved.

Keywords: Collagen extraction methods; Palaeodiet; Carbon; Nitrogen

1. Introduction such as humic acids and lipids that might influence the repro-
ducibility of the measurements (Bronk Ramsey et al., 2004;
Stable isotope analyses of d13C and d15N measured in bone Brown et al., 1988; Collins and Galley, 1998; Garvie-Lok
collagen are routinely used for the reconstruction of ancient et al., 2004; Lidén et al., 1995; Nielsen-Marsh and Hedges,
diets and subsistence patterns (e.g., Ambrose, 1993; Boche- 2000; Semal and Orban, 1995).
rens et al., 2006; Honch et al., 2006; Jay and Richards, Generally, following Longin (1971), the extraction methods
2006; Katzenberg, 2000; Richards et al., 1998). Several tech- involve dissolving the mineral matrix in a HCl solution, sub-
niques have been developed to prepare bone samples for iso- sequent solubilisation of collagen at elevated temperature in
tope analysis. Most of these consider and adjust for factors a weak HCl solution, followed by lyophilisation of the remain-
ing collagen. However, there are various modifications includ-
ing the addition of a treatment step with NaOH to remove
* Corresponding author. Tel.: þ45 2063 5054; fax: þ45 3532 7215. humic acids before solubilisation of the collagen (DeNiro
E-mail address: [email protected] (M.L.S. Jørkov). and Epstein, 1981), or the use of ultra-filtration to purify the

0305-4403/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jas.2006.12.020
 M.L.S. Jørkov et al. / Journal of Archaeological Science 34 (2007) 1824e1829 1825

solubilised collagen (Brown et al., 1988). These recommenda- Cortical bone samples were taken from the posterior side of
tions are commonly applied as cleaning steps in order to mea- the midshaft of femora and humeri. The surface of the bone
sure the original collagen used in dietary studies. samples was cleaned with a round milling cutter. Bone powder
There are diverse chemical approaches in sample prepara- was then drilled out using a low speed Proxxon MICROMOT
tion used by different stable isotope laboratories. However, 40E drill (drill diameter: 2 mm). The resultant bone samples
whatever the steps, the laboratories use each others results varied between consisting of fine bone powder (grains with
as references and comparison in dietary studies (e.g. Bayliss an approximate diameter of 0.01 mm) (p), a mix of bone pow-
et al., 2004; Jørkov, 2002; Keegan, 1989). Previous studies der and small shavings with a size of 0.1e0.5 mm, to shavings
of these extraction methods have examined the effects of ultra with a size of 0.9e1.0 mm (ps). A further rib fragment and
filtration in radiocarbon-dating and the contribution of lipids shards of a femur and humerus (sample nos. 6, 7 and 8)
on the stable carbon isotope values (e.g. Bronk Ramsey were ground manually in a mortar into coarse powder of
et al., 2004; Lidén et al., 1995). Results have shown that 0.04e0.9 mm (cp). The samples from each bone were subdi-
ultra-filtration may still leave larger contaminating particles vided into three portions, and the three extraction methods
and lipids may alter the carbon signals. We therefore thought were then applied.
it necessary to investigate whether the extraction methods and
cleaning steps may influence the isotopic result and hence po- 2.1. Extraction method A
tentially the interpretation of dietary variation.
The specific purpose of this study was to compare three sam- This extraction method followed the protocol of DeNiro
ple preparation methods on well-preserved skeletal material by and Epstein (1981). Bone samples were demineralised in
evaluating the resultant differences in d13C and d15N, collagen 1 M HCl at 4  C for 1 h, being stirred every 5 min. They
quality and collagen yield. The first method (A) includes the were then rinsed to neutral pH with de-ionised water. 0.2 M
treatment with NaOH, the second method (B) includes both NaOH was added to remove contaminating humic acids. Sam-
ultra-filtration and filtration with Ezee filter separators (5e8 mm) ples were rinsed again to neutral pH with de-ionised H2O. HCl
(Elkay Laboratory Product) before lyophilisation as described was added the sample tubes, obtaining a pH of 2.5. The sam-
in Richards and Hedges (1999) and Müldner and Richards ples were then covered and gelatinised in this weak acid solu-
(2005). The third method (C) is a modified version of method tion at 70  C for 16 h in order to concentrate the protein
B, by excluding the ultra-filtration step. This modified version components. After removing insoluble residues by centrifug-
of method B is also used in laboratories conducting stable iso- ing the samples at 2500 rpm for 10 min, the remaining super-
tope analysis in dietary studies (Honch et al., 2006). natant solution was evaporated at 100  C for 6 h until reaching
ca. 3 ml. The solution was then freeze-dried for 24 h.
2. Materials and methods
2.2. Extraction method B
The bone samples selected for this study were chosen from
Collagen was extracted using the standard procedures by
a large well documented skeletal collection from the medieval
Brown et al. (1988), modified in Richards and Hedges (1999)
cemetery Ahlgade 15e17 in Holbæk, Denmark (Asmussen,
by the use of Millipore Amicon Ultra-4 centrifugal filter
1997). The cemetery was used between ca. 1100e1573 AD,
(30,000 NMWL) prior to lyophilisation, so that molecules
and contained more than 700 skeletons buried in clay soil.
larger than 30 kDa were retained. As with method A, the sam-
The material was chosen because of its excellent state of pres-
ples were demineralised in 1 M HCl, at 4  C for 1.5e10 h, until
ervation (the bones were macroscopically intact with a hard
the release of CO2 could no longer be observed. The use of 1 M
structure and feel (i.e. non-flaky)). Eight bone samples from
HCl is a slight modification to 0.5 M HCl, which is commonly
five individuals were each treated with the three extraction
used (Richards and Hedges, 1999). The samples were then
methods. Details on the samples, including age and sex of
rinsed to neutral pH with de-ionised water. Weak HCl
the individuals are given in Table 1.
solution was added to obtain a pH of 2.5. Samples were then
Table 1 gelatinised at 70  C for 24 h. After collagen solubilisation
Sample details any insoluble residues were removed with a 5e8 mm Ezee
Individual Sample Sex Age Bone element Bone mesh filter (Elkay Laboratory Products). The remaining solu-
no. (years) structure tion was concentrated on the ultra-filters by centrifugation at
EG141 1 Male 45þ Femur (R) ps 2500 rpm. The supernatant of purified ‘‘collagen’’ (>30 kDa),
EG155 2 ? 9 Femur (R) s was then freeze-dried for 48 h.
EG160 3 Female 45þ Femur (R) p
4 Humerus (R) p 2.3. Extraction method C
EG161 5 Female 18e25 Femur (R) p
EG295 6 Male 18e25 Femur (R) cp
7 Humerus (L) cp Method C followed the same protocol as method B, but the
8 Rib cp ultra-filtration step was left out.
(R), right; (L), left. Bone structure when drilled: p, powder; ps, powder and The collagen extractions were carried out in our laboratory.
shavings; s, shavings; cp, coarse powder. All isotopic measurements were performed in at least
1826 M.L.S. Jørkov et al. / Journal of Archaeological Science 34 (2007) 1824e1829

duplicate with a GV Instruments Isoprime stable isotope mass

6.30
7.30
4.50
4.30
4.20
4.20
d13C and d15N measured in duplicate at the AMS 14C Dating Centre, Department of Physics and Astronomy, University of Aarhus, Denmark. The maximum analytical error (1s) for d13C and d15N were 0.15& and
18.43
20.63
spectrometer combined with a Eurovector elemental analyzer

C
(continuous flow) at the AMS Laboratory at the Institute of
Physics and Astronomy, University of Aarhus, Denmark.

2.79
5.70
1.80
2.48
1.20
1.32
2.10
3.50
B
Bulk collagen from each sample was weighed in duplicates
to between 215e250 mg. To examine the accuracy and preci-

% Yield

6.60
7.60
4.47
3.50
13.50
16.70

10.20
15.60
sion of analytical methods, a working standard gelatine mate-

A
rial and an AMS standard reference (a whale bone produced
with the preparation method A) (Heinemeier, 2005, personal
communication) with known d13C and d15N were analysed

3.2
3.1
3.3
3.3
3.3
3.3
3.1
3.3
C
in tandem with samples of bone collagen. These secondary
standards are calibrated against approximately ten internation-

3.4
3.3
3.3
3.4
3.3
3.4
3.3
3.3
B
ally recognised isotopic standards. The maximum analytical
error (1s) for d13C and d15N were 0.15& and 0.3&

C:N

3.3
3.3
3.2
3.1
3.2
3.3
3.1
3.2
respectively.

A
Comparative statistical analyses following the procedures
outlined by Bland and Altman (1986) were used to assess

15.8
16.5
14.0
16.2
14.9
14.9
15.9
15.0
the three preparation methods. Friedman tests (FTS) and

C
paired t-tests were applied to test for difference in C:N atomic
ratio, d13C and d15N and % collagen yield between the

16.0
16.8
16.4
15.9
16.2
15.5
15.7
15.5
B
methods (Altman, 1999).

3. Results

15.7
15.8
15.0
16.0
15.9
15.1
16.6
15.7
%N
A
The results of the stable isotope measurements and compar-

12.40
12.20
11.18
11.30
12.28
13.47
12.79
14.11
ative analysis are summarised in Tables 2e4 and Figs. 1 and 2.
C
3.1. Collagen preservation
12.77
12.51
11.26
11.25
12.47
13.47
13.36
12.79
On the basis of the quality indicators for bone collagen ex-
B

tracts (Ambrose, 1990, 1993; Schoeninger et al., 1989; van

Klinken, 1999), the collagen yield, C:N atomic ratio and %C
12.43
12.04
10.92
10.63
12.26
13.39
13.17
13.30
d15N

and %N were considered comparable for each of the three

A

methods although the collagen yields did vary greatly between

the methods (Table 2).
44.0
44.5
39.5
45.5
42.3
42.4
42.4
42.1
C

The lowest percentage collagen yield, ranging from 1.2% to

5.7%, was obtained using method B. Method A resulted in
yields ranging from 3.5% to 16.7%, and method C resulted
46.6
47.5
45.9
46.2
46.2
45.5
44.5
44.3
B

in yields ranging from 4.2% to 20.6%. However, %C by

weight of the collagen samples was fairly consistent (Table 2):
44.4
44.7
41.1
42.5
43.0
42.1
43.3
43.3
%C

41.1e44.7% (mean: 43.1%) for method A, 44.3e47.5%

Isotopic results of samples extracted by method A, B and C

(mean: 45.8%) for method B, and 39.5e45.5% (mean:

42.8%) for method C. This was also the case for the %N:
19.28
19.26
20.40
20.36
19.13
18.59
18.67
18.65

15.0e16.6% (mean: 15.7%) for method A, 15.5e16.8%

(mean: 15.9%) for method B, and 14.0e16.5% (mean:
C

15.4%) for method C.

19.25
19.38
20.41
20.54
19.20
18.86
19.04
18.99

3.2. d13C values

B

The results of the d13C measurements show that samples

18.99
19.02
20.25
20.23
18.82
18.48
18.57
18.60

prepared following method A have values that are consistently

d13C

less negative than samples prepared by method B, with method

A

0.3& respectively.

C having values that lie between the two (Table 2, Fig. 1). The
differences are statistically significant (FTS ¼ 14.250,
Sample no.

p ¼ 0.001). A comparative analysis of the differences between

Table 2

the methods showed that there is a statistically significant

difference between all three methods (Table 3), albeit this
1
2
3
4
5
6
7
8
 M.L.S. Jørkov et al. / Journal of Archaeological Science 34 (2007) 1824e1829 1827

Table 3
Comparative analysis of the difference in d13C values between the methods
d13C Mean difference 2s t p
Method A vs. B 0.32& 0.22& 8.195 <0.001
Method A vs. C 0.15& 0.20& 4.311 0.004
Method B vs. C 0.17& 0.30& 3.140 0.016

difference is numerically small: the biggest difference is

between Methods A and B, with a mean difference of
0.32  0.22& (2s).

3.3. d15N values

Fig. 1. d13C & of the samples prepared by method A ¼ C, B ¼ 
15
The analysis of the d N measurements between the and C ¼ :. All samples have marked maximum analytical error of
methods and of their difference was completed in a similar 1s ¼ 0.15&.
manner. There was no statistically significant difference in
d15N values between any of the methods (Table 4, Fig. 2). peptides should be improved as contaminants are expected
to be of lower molecular weight.
It was anticipated that collagen yields would be somewhat
4. Discussion
lower for the samples extracted with method B, since the use
of ultra-filters lowers the overall yield by removing molecules
When working with archaeological bone collagen, it is es-
that are smaller than 30 kDa. We performed an analysis of the
sential to be familiar with the mechanisms that can alter the
collagen loss when ultra-filtrating. The analysis showed that
isotopic signal as well as with the quality indicators that are
the collagen yield is reduced by up to 86% (data not shown).
available to assess collagen preservation. The sources of con-
The lowest collagen yield was obtained from bone sample no.
taminants that can act to decompose the bone mineral (i.e. mi-
5, with a collagen yield of 1.2% using method B and 4.47% using
crobial attack, pH, groundwater activity and temperature etc.)
method A (Table 2). This sample was taken from a femur, and
are depended on the burial environment and vary both geo-
produced fine powder during drilling of the bone sample. The
graphically and temporally (Hedges, 2002; Nielsen-Marsh
powdery nature of the drilled sample could indicate that the col-
and Hedges, 2000). Before applying the preferred preparation
lagen was more poorly preserved and the low yield would seem
method, one should therefore consider the possible sources of
to confirm this. In contrast, bone sample no. 2 produced shavings
diagenesis and contaminants from the burial environment that
when drilling. In our experience the quality of bone shavings
may have altered the original collagen isotopic signal. Today
seem to be an indication of well preserved collagen, although
most isotopic studies (including dietary studies) follow the
this would be difficult to quantify between different sites. This
same criteria suggested by Ambrose (1990, 1993) and van
assumption was confirmed by the high collagen yield for all
Klinken (1999) for what is considered well preserved colla-
three methods. The method that gave the lowest yield of the
gen. Methods B and C are widely used in radiocarbon labora-
tories and laboratories conducting stable isotope analysis in
dietary studies (Bronk Ramsey et al., 2004; Honch et al.,
2006; Richards et al., 2006; Richards and Hedges, 1999).
Since several extraction protocols exist, it is important that
in the end the protocols yield comparable results. The quality
of the collagen should therefore also be compared as it may
influence the result. The % collagen yield is one criteria, but
not in itself the dominating factor. Of course, the ideal extrac-
tion method should maximise collagen yield while minimising
the degradation of the extracted protein remnants and remov-
ing contaminants. By using method B, the yield of larger

Table 4
Comparative analysis of the difference in d15N values between the methods
d15N Mean difference 2s t p
Method A vs. B 0.22& 0.68& 1.814 0.112
Fig. 2. d15N & of the samples prepared by method A ¼ C, B ¼ 
Method A vs. C 0.20& 0.77& 1.462 0.187
and C ¼ :. All samples have marked maximum analytical error of
Method B vs. C 0.019& 1.16& 0.092 0.930
1s ¼ 0.30&.
1828 M.L.S. Jørkov et al. / Journal of Archaeological Science 34 (2007) 1824e1829

three for this bone sample was method B. Again, this was likely method A with NaOH as cleaning agent will according to
to be due to the loss of collagen through the ultra-filter. However, our study give d13C values that are on average 0.32& more
powder (‘‘p’’) is not the only indication of poor collagen preser- positive than using the ultra-filtration method (method B).
vation in bone. Bone samples prepared from coarse powder Our study did not show any significant difference in d15N
(‘‘cp’’) have just as low a collagen yield. Schoeninger et al. values. The variation in d15N between the methods of bone
(1989) reported that they achieved a much higher collagen yield sample no. 8 is, however, noteworthy. This sample was taken
if they avoided powdering, but they also stated that the superfi- from a rib (i.e. trabecular bone). The other seven samples were
cial appearance, as well as the collagen yield or the C:N atomic from the cortical part of long bones (i.e. compact bone). Since
ratio can not be used to predict how well collagen is preserved. trabecular bone is more easily contaminated than compact
However, in samples of poor preservation, preparation of chunks bone, the variation in d15N between the methods might per-
rather than powder should be achieved, as it has a greater chance haps indicate that method C does not remove certain contam-
of producing original intact collagen because the collagen fibre inants as well as methods A and B, which have more similar
structure is preserved. Collins and Galley (1998) also showed d15N values for this bone sample.
that grinding of bone could damage the collagen. The signifi- Samples extracted using method B should have had un-
cance should be less in collagen which is already damaged, as wanted non-protein contaminants removed in the ultra filtra-
many of the long rigid collagen fibres have already been cut tion step. In our study, the treatment with NaOH seems to
up. The high %C and %N values measured in the large mole- remove the non protein contaminants (base-soluble contami-
cules of method B could indicate two things: Either the original nants such as humic acids) better than the ultra-filter on
bone collagen is still contaminated, or the original collagen has material that has molecules less than 30 kDa. It produces
molecular sizes which are smaller than 30 kDa and what is left is a residue that is mostly derived from collagen, but can also
contamination. contain extraneous organic and inorganic matter. The effect
The ultra-filters were pre-rinsed with 0.1 N NaOH and of those seems insignificant on skeletal material which is
centrifuged twice with de-ionised water as suggested by the well preserved.
manufacturer. Contaminants that could interfere should be re- The d13C results could be biased by lipids as they are
moved at this stage. Studies at the Oxford Radiocarbon Accel- known to have d13C values more negative from those of pro-
erator Unit (Bronk Ramsey et al., 2004) have shown that this tein (Lidén et al., 1995). However, the degree of interference
pre-rinsing may not be sufficient to remove contaminants and is depended upon the amount of lipid preserved in bone. In
that the suggested cleaning protocol by manufacturer might this case we are analysing well preserved archaeological
cause the C:N atomic ratios to be higher than the original col- bone material from the medieval period and the chances of
lagen. A C:N atomic ratio within the range of 2.9e3.6 is an having lipids present is not so high as in modern bone samples,
indicator of good collagen preservation (Ambrose, 1990). As although cholesterol is known to be quite robust in archaeolog-
with our C:N results (Table 2), the original C:N atomic ratio ical time scales (Stott and Evershed, 1996). According to
for each of the cleaning methods were all within the accept- Lidén et al. (1995), the d13C in non-lipid-extracted collagen
able limits because the absolute proportion of contamination may be as high as 1.8& more negative than in samples of lipid
was very small. It is not all laboratories which are able to con- extracted collagen. In order to avoid variability they therefore
duct such thorough cleaning protocol of the ultra-filters since suggest removing lipids entirely in the extraction process by
it is both time consuming and costly. incorporating a methanol-chloroform solvent wash step.
The samples treated with NaOH (A) had C:N atomic ratios Method B selects the high molecular weight material. As
of 3.1e3.3 and the ultra-filtrated samples had slightly higher lipid molecules have a smaller weight than 30 kDa, the
C:N atomic ratios between 3.3e3.4. The largest difference be- Ultra-filtration should remove any lipids that may have been
tween the two methods was for sample no. 4 with C:N ratio of present in the sample. Method A only removes the humic
3.1 and 3.4 for A and B, respectively. acids. Lidén et al. (1995) show that treatment with NaOH re-
Why samples extracted using method C have lower %C and sults in a decrease in collagen yield, contrary to the results of
%N values when the samples have only been Ezee filtered can- this study.
not be explained at this point. According to the Oxford 14C da-
tabase (van Klinken, 1999) the %C and %N values of intact 5. Conclusion
collagen should be around 34.8  8.8 and 11e16 wt%, respec-
tively. Higher values might indicate addition of organic carbon From this study we conclude that the use of stable isotope
with small amounts of inorganic matter can be expected in the analysis in palaeodietary studies requires adherence to routine
extracts (Ambrose, 1990). And since the rest of the quality methods for preparing and measuring samples. How far apart
markers suggest good collagen, the collagen should be not re- measurements can be without causing variation in the dietary
jected for analysis. It seems as if the use of NaOH in the A interpretation is in part a question of judgement. According to
method removes the non protein contaminants as the samples Lovell et al. (1986) a normal variation within a population is
have %C and %N values within the accepted range of 41.1e up to 0.3&. Any difference in d13C larger than 0.3& may
44.7% and 14e16%, respectively. cause the interpreter to suggest a difference in subsistence. De-
There is a systematic and significant difference in the d13C Niro and Schoeninger (1983) showed an intra-individual vari-
values between the methods although it was small. Using ation of up to 2&; however, those studies were conducted on
 M.L.S. Jørkov et al. / Journal of Archaeological Science 34 (2007) 1824e1829 1829

animals raised on controlled diets and are not a satisfactory Garvie-Lok, S.J., Varney, T.L., Katzenberg, M.A., 2004. Preparation of
way to estimate baseline variability in human isotope ratios. bone carbonate for stable isotope analysis: The effects of treatment
time and acid concentration. Journal of Archaeological Science 31,
Although a difference in collagen quality using the different 763e776.
methods was seen in this study and were in fact statistically Hedges, R.E.M., 2002. Bone diagenesis: An overview of processes. Archaeo-
significant, we believe it would have no influence on the over- metry 44, 319e328.
all interpretation of the isotope results. They do, however, in- Honch, N.V., Higham, T.F.G., Chapman, J., Gaydarska, B., Hedges, R.E.M.,
dicate a need for consistency in the preparation method used 2006. A paleodietary investigation of carbon (13C/12C) and nitrogen
(15N/14N) in human and faunal bones from the Copper Age cemeteries
and an awareness of the difference that they may give. Natu- of Varna I and Duankulak, Bulgaria. Journal of Archaeological Science
rally, if laboratories make their own modifications to the ex- 33 (11), 1493e1504.
traction procedures, these should be duly noted. Jay, M., Richards, M.P., 2006. Diet in the Iron Age cemetery population at
We are aware that our study comprised only eight samples Wetwang Slack, East Yorkshire, UK: carbon and nitrogen stable isotope
of well preserved bone from one Medieval cemetery in Den- evidence. Journal of Archaeological Science 33, 653e662.
Jørkov, M.L.S., 2002. Dining with the Romans during the Imperial period. A
mark. We therefore think that a future study would benefit study based on stable isotope analysis. Unpublished M.Sc. dissertation,
by including poorly preserved bone as well as material from University of Bradford.
different sites and time periods, to see how effective the three Katzenberg, M.A., 2000. Stable isotope analysis: a tool for studying past diet,
methods are for excluding contaminants. demography, and life history. In: Katzenberg, M.A., Saunders, S.R. (Eds.),
Biological Anthropology of the Human Skeleton. Wiley-Liss, New York,
pp. 305e327.
_xcan, M.Y.,
Keegan, W.F., 1989. Stable isotope analysis of prehistoric diet. In: Is
References Kennedy, K.A.R. (Eds.), Reconstruction of Life from the Skeleton. Alan R.
Liss, New York, pp. 223e236.
Altman, D.G., 1999. Practical Statistics for Medical Research. Chapman & van Klinken, G.J., 1999. Bone collagen quality indicators for palaeodietary
Hall/CRC. and radiocarbon measurements. Journal of Archaeological Science 26,
Ambrose, S.H., 1990. Preparation and characterization of bone and tooth 687e695.
collagen for isotope analysis. Journal of Archaeological Sciences 17, Lidén, K., Takahashi, C., Nelson, D.E., 1995. The effects of lipids in stable
431e451. isotope analysis and the effects of NaOH treatment on the composition
Ambrose, S.H., 1993. Isotopic analysis of palaeodiets: Methodological and of extracted bone collagen. Journal of Archaeological Science 22, 321e
interpretive considerations. In: Sandford, M.K. (Ed.), Investigations of 326.
Ancient Human Tissue: Chemical Analysis in Anthropology. Gordon and Longin, R., 1971. New method of collagen extraction for radiocarbon dating.
Breach Science Publishers, Langhorne, Pennsylvania, USA, pp. 59e130. Nature 230, 241e242.
Asmussen E. (Ed.), 1997. Ahlgade 15e17, Holbæk. En arkæologisk og historisk Lovell, N.C., Nelson, D.E., Schwarcz, H.P., 1986. Carbon isotope ratios in
undersøgelse fra 1200 til nutiden. Aarbøger for Nordisk Oldkyndighed og palaeodiet: Lack of age or sex effect. Archaeometry 28, 51e55.
Historie 1994e95. Det Kongelige Nordiske Oldskriftselskab, København. Müldner, G., Richards, M.P., 2005. Fast or feast: reconstructing diet in later
Bayliss, A., Shepherd Popescu, E., Beavan-Athfield, N., Bronk Ramsey, C., Medieval England by stable isotope analysis. Journal of Archaeological
Cook, G.T., Locker, A., 2004. The Potential significance of dietary offsets Science 32 (1), 39e48.
for the interpretation of radiocarbon dates: an archaeologically significant Nielsen-Marsh, C.M., Hedges, R.E.M., 2000. Patterns of diagenesis in bone II:
example from medieval Norwich. Journal of Archaeological Science 31, Effects of acidic acid treatment and the removal of diagenetic CO2 3 . Jour-
563e575. nal of Archaeological Science 27, 1151e1159.
Bland, J.M., Altman, D.G., 1986. Statistical methods for assessing agreement Richards, M.P., Hedges, R.E.M., 1999. Stable isotope evidence for similarities
between two methods of clinical measurement. Lancet 8476, 307e310. in the types of marine foods used by Late Mesolithic humans at sites along
Bocherens, B., Mashkour, M., Drucker, D.G., Moussa, L., Billiou, D., 2006. the Atlantic coast of Europe. Journal of Archaeological Science 26, 717e
Stable isotope evidence for palaeodiets in southern Turkmenistan during 722.
Historical period and Iron Age. Journal of Archaeological Science 33, Richards, M.P., Hedges, R.E.M., Molleson, T.I., Vogel, J.C., 1998. Stable
253e264. isotope analysis reveals variation in human diet at the Poundbury Camp
Bronk Ramsey, C., Higham, T., Bowles, A., Hedges, R., 2004. Improvements cemetery site. Journal of Archaeological Science 25, 1247e1252.
to the pre-treatment of bone at Oxford. Radiocarbon 46 (1), 155e163. Richards, M.P., Fuller, B.T., Molleson, T.I., 2006. Stable isotope palaeodietary
Brown, T.A., Nelson, D.E., Vogel, J.S., Southon, J.R., 1988. Improved colla- study of humans and fauna from the multi-period (Iron Age, Viking and
gen extraction by modified Longin method. Radiocarbon 30 (2), 171e177. Late Medieval) site of Newark Bay, Orkney. Journal of Archaeological
Collins, M.J., Galley, P., 1998. Towards an optimal method of archaeological Science 33, 122e131.
collagen extraction: The influence of pH and grinding. Ancient Biomole- Schoeninger, M.J., Moor, K.M., Murray, M.L., Kingston, J.D., 1989. Detection
cules 2, 209e222. of bone preservation in archaeological and fossil samples. Applied Geo-
DeNiro, M.J., Epstein, S., 1981. Influence of diet on the distribution of nitro- chemistry 4, 281e292.
gen isotopes in animals. Geochimica et Cosmochimica Acta 45, 341e351. Stott, A.W., Evershed, R.P., 1996. d13C analysis of cholesterol preserved in
DeNiro, M.J., Schoeninger, M.J., 1983. Stable carbon and nitrogen isotope archaeological bones and teeth. Analytical Chemistry 68, 4402e4408.
ratios of bone collagen: variations within individual, between sexes, and Semal, P., Orban, R., 1995. Collagen extraction from recent and fossil bones:
within populations raised on monotonous diets. Journal of Archaeological Quantitative and qualitative aspects. Journal of Archaeological Science 22,
Science 10, 199e203. 463e467.