Introduction
Recent developments in radiocarbon dating (RD) have made this technique increasingly attractive for dating cultural heritage artifacts (Hajdas Reference Hajdas2023; Hajdas et al. Reference Hajdas, Jull, Huysecom, Mayor, Renold, Synal, Hatte, Hong, Chivall, Beck, Liccioli, Fedi, Friedrich, Maspero and Sava2019). However, one of the biggest challenges that still exist in RD of cultural heritage is the presence of exogenous carbon introduced during burial, conservation, restoration, or storage (Brock et al. Reference Brock, Dee, Hughes, Snoeck, Staff and Ramsey2018; Caforio et al. Reference Caforio, Fedi, Liccioli and Salvini2013; Durier Reference Durier2022; Fedi et al. Reference Fedi, Caforio, Liccioli, Mandò, Salvini and Taccetti2014; Hajdas et al. Reference Hajdas, Ascough, Garnett, Fallon, Pearson, Quarta, Spalding, Yamaguchi and Yoneda2021; Hajdas et al. Reference Hajdas, Guidobaldi, Haghipour and Wyss2024; Liccioli et al. Reference Liccioli, Fedi, Carraresi and Mando2017).
The method, developed in 1949 (Arnold and Libby Reference Arnold and Libby1949), is based on measurements of the residual activity of 14C, fixed in organic matter or carbonates, that decays after death or burial. The premise of the method is that the original isotopic signature of the material reflects that of the atmosphere at the time of formation and any measurable changes in the concentration of 14C are due to radioactive decay. The atmospheric 14C concentration and that of contemporary (living) organic matter are on the level of one 14C atom in a trillion 12C atoms (14C/12C = 10–12) and this ratio declines to 10–15 after 50,000 years, which is the time limit of the method. Such low concentrations require a high level of purity of the sample to assure accurate readings. Two factors are important when the contamination is at play: one is the mass of the contaminant relative to the mass of total carbon and the second is the 14C/12C ratio of this exogenous carbon. For more than a century, conservation and restoration treatments that preserved archaeological objects and excavation sites have relied on organic consolidants, adhesives, and coatings (Horie Reference Horie2010). Many of these conservation materials are produced from fossil carbon, which is free of 14C and introduces significant old contamination, due to their high carbon content. The effect has been recognized in various studies (e.g., Donahue et al. Reference Donahue, Olin and Harbottle2002; Donahue and Jull Reference Donahue and Jull1998; Hajdas et al. Reference Hajdas, Bonani, Thut, Leone, Pfenninger and Maden2004). Methods of potential pretreatment were investigated by Dee et al. (Reference Dee, Brock, Bowles and Ramsey2011) and more recently by Brock et al. (Reference Brock, Dee, Hughes, Snoeck, Staff and Ramsey2018) who employed Chromosorb, a silica-based polymer with a high surface area to volume ratio, which was artificially aged before simulation of contamination with various conservation materials. The tests performed by Brock et al. (Reference Brock, Dee, Hughes, Snoeck, Staff and Ramsey2018) were undertaken on 11 substances, and a combination of 17 treatment steps involving solvents was applied. Fourier transform infrared (FTIR) spectroscopy and gas chromatography-mass spectrometry (GC-MS) were used to monitor the efficiency of specific sample pretreatments using solvents. Although performed on the artificial material, with no control of the final 14C signal, both studies hint to the actual problem: the complete removal of some conservation substances, such as polyvinyl acetate (PVA) derivatives and/or cellulose nitrate lacquers, is limited, if not impossible.
Typically, RD of anthropogenically modified wood is challenging due to the presence of substances used to protect the wood from various destructive processes. Two general classes of wood can be subject to 14C investigations: excavated archaeological waterlogged wood (hereafter called archaeological wood), and historical wood (i.e., wood that has not undergone burial on land or in water).
Archaeological waterlogged wood often requires replacing water in cell structures with substances that prevent collapse during the drying phase. This is because such wood has undergone biological degradation during burial, which has weakened the cellular structure (Blanchette et al. Reference Blanchette, Haight, Koestler, Hatchfield and Arnold1994; Kim et al. Reference Kim, Funada and Adya2016). Numerous wood consolidants are known to have been in use over the last century. With the exception of sucrose, which can be used to preserve wood, most consolidants are made of fossil carbon i.e., they contain 14C-free carbon. Other carbon-containing contaminants present in archaeological wood could originate from its use in the past (e.g., fatty acids or proteins in storage containers, tar on ship timbers), burial and post excavation storage (such as carbonates and/or remains from bacteria, fungi, insect or worm attack) or post-excavation conservation treatments (adhesives, varnishes, coatings) (Durier Reference Durier2022, unpublished PhD thesis).
Historical wooden objects may be painted, varnished, treated with pesticides, glued, or contain minerogenic oils or grease, which were applied to hinges and locks of antique doors, furniture, or caskets. The wood used in buildings has been impregnated with substances that protect against weathering, fire, or insect pests (pesticides). In most cases, the origin of potential contamination is unknown to the radiocarbon laboratory.
The first application of solvents at the ETH laboratory dates to 1998, with the analysis of wood submitted by Herbert Haas (Radiocarbon Laboratory, Las Vegas, USA). At that time, Herbert Haas, who was the head of a conventional (beta counting) laboratory, collaborated with Georges Bonani and ETH/PSI AMS laboratory. Samples with a small amount of material, insufficient for the counter technique, were submitted to our laboratory. At the time, samples could be easily transferred between countries, enabling radiocarbon dating of materials from Egypt. Among the samples submitted by Herbert Haas were multiple samples of wood from the boat of the pharaoh Khufu. Some of the samples returned expected ages but a few were extremely old (up to 20 kyrs). Solvent treatment has been applied which reduced the effect of contamination but did not entirely remove it (Hajdas et al. Reference Hajdas, Bonani, Thut, Leone, Pfenninger and Maden2004). The results remain unpublished, but the experience and that observation has inspired this study.
Materials
Samples of wood were chosen from well-dated archaeological context from the collections of the Department of Archaeology, Canton Thurgau, Frauenfeld, Switzerland for this study. Wet (waterlogged) wood samples, which have been excavated from three sites of Gachnang-Niederwil, Eschenz and Arbon Bleiche (Switzerland) are well dated thanks to the archaeological context and dendrochronology (Benguerel et al. Reference Benguerel, Brem, Giger, Leuzinger, Pollmann, Schnyder, Schweichel, Steiner and Streit2014; Hasenfratz and Raemaekers Reference Hasenfratz and Raemaekers2007; Leuzinger Reference Leuzinger2000; Ruoff and Seifert Reference Ruoff and Seifert2018; Schweichel et al. Reference Schweichel, Leuzinger and Brem2017). Different approaches to conservation of the wood have been applied by several conservation laboratories at Department of Archaeology, Canton Thurgau and Swiss National Museum, Affoltern am Albis.
Samples
No. 1 Sample 1962.027 Hg 810, Alnus, preserved with Arigal C (Melamine-formaldehyde MF)
A wooden vessel was excavated in 1962 from the Neolithic site Gachnang-Niederwil, near Egelsee. The site has been dated using a wiggle matching of radiocarbon ages and dendrochronology between ca. 3660 and 3585 BCE (Benguerel et al. Reference Benguerel, Brem, Hasenfratz, Keller, Leuzinger, Stehrenberger and Steiner2010).
No. 2 Sample 1991.004 H 34, Quercus, preserved with PEG (submerged)
Wooden frame of latrine excavated in 1991 at a Roman house (Hs. Rebmann)/Eschenz. The wood has been dated by dendrochronology to ca. 160 CE (Benguerel et al. Reference Benguerel, Brem, Fazer, Giger, Hartmann, Leuzinger, Mayer, Müller, Schnyder, Schoch, Schweichel and Steiner2011; Jauch Reference Jauch1997).
No. 3 Sample 1994.006 H 96 S, Fagus, preserved with Sucrose
Wooden frame of water reservoir recovered from area Zatti-Landolt/Eschenz (Canton Thurgau) in 1994 dated by dendrochronology to the end of 1st century CE (Jauch Reference Jauch1997).
No. 4 Sample 1995.001 H 3006, Tilia, preserved with Ethanol-Ether
Wood from a log ladder found in Arbon Bleiche 3 (Canton Thurgau) in 1995. The famous Neolithic site of lake dwellers (Jacomet et al. Reference Jacomet, Leuzinger and Schibler2004; Leuzinger Reference Leuzinger2000) has been dated using dendrochronology to 34th century BCE (Schweichel et al. Reference Schweichel, Leuzinger and Brem2017).
No. 5 Sample 1995.006 H 3358, Quercus, PEG freeze-dried (FD_Q)
Stave, a wooden part of a barrel from the site Arbon Bleiche 3 (Canton Thurgau), excavated in 1995.
No. 6 Sample 2002.051 H 2145, Alnus, PEG freeze-dried (FD_A)
Barrel stave from Eschenz/Tasgetium (Canton Thurgau) dated to the end of the 1st century CE or beginning of the 2nd century (Benguerel et al. Reference Benguerel, Brem, Ebneter, Ferrer, Hartmann, Leuzinger, Müller, Rast-Eicher, Rühling, Schweichel and Spangenberg2012).
No. 7 Sample 2010.006 H 27 S, Quercus, room dried (Air)
Wooden post from a Roman bridge Eschenz/Tasgetium (Canton Thurgau) dated from the middle of 1st century to the 3rd century CE (Benguerel et al. Reference Benguerel, Brem, Fazer, Giger, Hartmann, Leuzinger, Mayer, Müller, Schnyder, Schoch, Schweichel and Steiner2011).
Methods
We have chosen to proceed with standard methods of treatment samples for radiocarbon analysis. We also combined or modified methods to test their effectiveness. For comparison all samples were also analyzed without treatment.
Radiocarbon dating—pretreatment methods
-
1. No treatment – ca. 3 mg of wood were packed/wrapped in aluminum cups for combustion in the Elemental Analyzer (EA).
-
2. Acid-Base-Acid (ABA) 60ºC – samples were treated for an hour with 0.5 M HCl, 0.1 M NaOH, 0.5 M HCl at 60ºC.
-
3. Soxhlet – a sequence of one-hour washes each in solvents hexane, acetone, and ethanol was performed in a Soxhlet apparatus (Hajdas et al. Reference Hajdas, Bonani, Thut, Leone, Pfenninger and Maden2004).
-
4. Soxhlet +ABA 60ºC – ABA 60ºC followed the Soxhlet solvent wash.
-
5. Soxhlet + BABAB – the two methods were applied; cellulose separation using the BABAB method (Nemec et al. Reference Nemec, Wacker, Hajdas and Gaggeler2010) followed the Soxhlet solvents wash.
-
6. Cellulose BABAB – cellulose was separated from original sample (Nemec et al. Reference Nemec, Wacker, Hajdas and Gaggeler2010).
-
7. ABA 60ºC Strong – samples were treated for an hour with 1 M HCl, 0.1 M NaOH, 1 M HCl at 60 ºC (modified after Brock et al. Reference Brock, Higham, Ditchfield and Ramsey2010).
-
8. Soxhlet + ABA 60ºC Strong – the two methods were applied; ABA 60ºC Strong followed the Soxhlet solvent wash.
The cleaned samples were air dried at 60ºC and weighed into aluminum cups for the combustion in EA and graphitization (Nemec et al. Reference Nemec, Wacker, Hajdas and Gaggeler2010). Graphite samples were pressed into the aluminum cathodes and analyzed at the MICADAS (Synal et al. Reference Synal, Stocker and Suter2007).
FTIR (Fourier transform infrared spectroscopy) analysis
Monitoring sample contamination using FTIR was not possible prior to AMS analysis. The spectra have been collected on the remaining archived material (when available). FTIR spectra were recorded on an infrared spectrometer by using the ATR (Attenuated Total Reflectance) method (Spectrum-2 PerkinElmer instrument). Spectra were collected between wavenumber 4000 and 450 cm–1 with a resolution of 4 cm–1 and in 8 runs.
Results and discussion
The results of all the AMS analysis (F14C, RA, δ13C) are reported in Table S1 (Supplementary Material). In general, the treatment methods that involve the separation of cellulose provided little material, and some of them were not sufficient for AMS analysis. We were unable to measure the 14C content of the substances used for conservation. The effect of the contamination could only be illustrated by analysis of original samples (without chemical pretreatment, method #1). However, the measured values are only for single samples and do not reflect the degree of contamination in the whole wood object as this is most probably heterogenous.
Figure 1 shows the offset from the expected 14C ages (eRA) (based on true age) detected for samples without chemical pretreatment. The sample eRA was calculated with R_Simulate of OxCal 4.4 and the archaeological ages (Table S2). From this analysis we estimate that the only “Sucrose” and “Ethanol-Ether” conservation treatments used components with F14C>1, and the measured RA are younger than the expected (negative offset). The use of Arigal C and PEG in conservation resulted in depletion of 14C content (apparent old ages = positive offset of RA). As expected, the “air-dried” wood was free of contamination.
Offset of RA measured in wood samples that were not chemically treated in the laboratory, i.e., portion of the original samples were directly combusted and graphitized.
The methods applied in this study worked differently for different conservation treatments.
Our results show that the conservation effect on the 14C ages is difficult to reverse with the standard pretreatment protocols. Only analysis of the air-dried wood (sample No. 7) returned expected age and the difference between measured and expected RA (Figure 2).
Effect of various treatment methods illustrated as a difference between measured (mRA) and expected ages (eRA) of wood samples conserved with Arigal C (Squares), PEG (Filled circle), Sucrose (Diamonds), Ethanol-Ether (Triangles, PEG-FD_Quercus (Open Diamond), PEG-FD_Alnus (Cross), and Air-Dried (Open circles).
No. 1 Sample 1962.027 Hg 810, Alnus, preserved with Arigal C (Melamine-Formaldehyde, MF)
The degree of contamination introduced by the conservation with Arigal C varies for wood treated with different protocols. Apparently, the offset is at the lowest for untreated sample (#1). However, because the AMS analysis is sensitive to minute amounts of carbon, the high heterogeneity of the sample material may be responsible for the low effect observed (Hajdas et al. Reference Hajdas, Guidobaldi, Haghipour and Wyss2024). All treatment protocols (#2-#8) failed to remove contamination, moreover, extremely high offset (up to 10–11 kyrs 14C) is observed for treatment intended to separate cellulose. In general, cellulose is poorly preserved in waterlogged wood and yield of separation of this compound is very low (Hajdas et al. Reference Hajdas, Hendriks, Fontana and Monegato2017, Reference Hajdas, Guidobaldi, Haghipour and Wyss2024). It is worth to note that all the samples in this study were waterlogged pieces of wood and all resulted in very low amount of cellulose. In fact, contamination is apparently amplified in cellulose separted from samples preserved by all chemicals.
The FTIR spectra of the original sample (Alnus with Arigal C) and on the rest material of this sample after method #6: Cellulose BABAB are compared in Figure 3. The transmission spectra of the original sample show lines corresponding to waterlogged wood (Abies cephalonica) treated with melamine-formaldehyde MF (Vlata et al. Reference Vlata, Rapti, Boyatzis, Bardet, Lucejko and Pournou2024). The characteristic MF lines (⁓2900 cm–1 and 2877 cm–1, 811 cm–1, 780 cm–1) (Kiliç and Kiliç Reference Kiliç and Kiliç2019) can be identified in the spectrum of the treated sample.
FTIR spectra of wood sample conserved with Arigal C: untreated sample (blue line) and sample after treatment with method #6 (Cellulose BABAB) (orange line).
No. 2 Sample 1991.004 H 34, Quercus, preserved with PEG (submerged)
Contamination introduced into the Quercus sample was considerable, with an RA offset of 11.5 kyrs measured on the original sample (Figure 1). However, as shown in Figure 2, most of this contamination has been removed by all treatment methods. The result for cellulose preparation (#6) is unknown because of the poor preservation. The best effect was achieved by treatment #5 (Soxhlet and cellulose preparation) on wood associated with the Roman house in Eschenz, which was dendrochronologically dated to 159 CE. However, this result gave a slightly older than expected result, as calibration of the measured RA = 2037 ± 27 BP yielded calendar ages between 147 BCE and 61 CE.
The efficiency of removal PEG is also visible in the FTIR spectra of the sample (Figure 4). The characteristic lines of PEG (Han et al. Reference Han, Guo, Tian, Jiang and Yin2022; Kılıç and Kılıç Reference Kılıç and Kılıç2024) detected in the original sample are not anymore prominent in the rest material of samples treated with Soxhlet+ABA (method #4) and Soxhlet+ABA_Strong (method #8).
FTIR spectra of wood sample conserved with PEG (submerged): untreated sample (blue line), sample after treatment with method #4 (Soxhlet+ABA) (orange line) and #8 (Soxhlet+ABA strong) (green line).
No. 3 Sample 1994.006 H 96 S, Fagus, preserved with Sucrose
Sucrose used to preserve wood was most likely produced around 1990, when atmospheric F14C was approximately 1.15. Sucrose is soluble in water but poorly soluble on ethanol and other solvents, therefore no treatment resulted in complete decontamination (Figure 2). The complexity of contamination is evident from a wide range of RA measured on this sample. Only measurements carried out after pretreatment #3 (Soxhlet) returned a young age, as expected in the case of modern contaminant. The presence of old contamination suggests that other conservation substances such as PVA has been added possibly to protect the surface of the impregnated wood.
No. 4 Sample 1995.001 H3006, Tilia, preserved with Ethanol-Ether
The 14C measured in the untreated sample (#1) shows a negative offset of nearly 6000 14C years (Figure 2) and indicates a contamination F14C > 1.8, corresponding to the early 1960s. The maximum of the bomb peak F14C values in the atmosphere was in 1963 (Hua et al. Reference Hua, Turnbull, Santos, Rakowski, Ancapichún, De Pol-Holz, Hammer, Lehman, Levin, Miller, Palmer and Turney2022). Wood from log ladder found in Arbon Bleiche 3 in 1995 was certainly preserved later than the excavations therefore, the components used appear to have been stored for three decades. Also, this conservation proved resistant to any of our seven treatment methods. However, Soxhlet and ABA treatments (methods #3 and #8) might be worth investigating, perhaps by a new choice of solvents and time of treatment.
No. 5 Sample 1995.006 H 3358, Quercus, PEG freeze-dried (FD_Q)
Treatment of Quercus wood with PEG and subsequent freeze-drying introduced contamination with old carbon, which is irreversible by any treatment (Figure 2).
No. 6 Sample 2002.051 H 2145, Alnus, PEG freeze-dried (FD_A)
Figure 2 shows that contamination of this sample appears to be nearly removed by the seven treatment methods. The most significant offset (ΔRA=1813 14C yrs; ΔF14C= -0.16) from the expected age is observed for sample treated with Soxhlet (#3) but other methods were not sufficient and ΔRA is between 43 and 178 14C yrs. The lowest ΔRA for method #4 (Soxhlet +ABA) and the measured RA when calibrated results in calendar age between 161 calBCE and 10 CE. This result is slightly older than the expected from the expected dating to the end of 1st/beginning of the 2nd century CE (Benguerel et al. Reference Benguerel, Brem, Ebneter, Ferrer, Hartmann, Leuzinger, Müller, Rast-Eicher, Rühling, Schweichel and Spangenberg2012).
No. 7 Sample 2010.006 H 27 S, Quercus, room dried (Air)
All methods yielded consistent radiocarbon ages for wood sample No. 7, which was not preserved with conservation materials but was air-dried. The combined RA =1918 ± 10 BP (χ2 Test: df=7, T=5.7 [5% 14.1]). Due to the wiggle on the calibration curve the calibrated age of this combined RA falls into 2 intervals 66–170 calCE and 184–204 calCE in agreement with the expected age of the Roman wooden bridge which was constructed and renovated from the beginning of 1st century to the 3rd century CE (Benguerel et al. Reference Benguerel, Brem, Fazer, Giger, Hartmann, Leuzinger, Mayer, Müller, Schnyder, Schoch, Schweichel and Steiner2011).
Conclusions
Conservation of wooden objects, which is essential in the protection of objects in cultural heritage, poses one of the most significant challenges in obtaining accurate radiocarbon ages. This study demonstrates that standard treatment methods applied in radiocarbon laboratories cannot entirely remove contamination introduced by Arigal C, Sucrose, Ethanol-Ether and PEG. The closest results have been obtained for samples consolidated by PEG but also this substance remained could not be removed entierly by the treatement, thus testing new approaches is essential to improve accuracy of radiocarbon dating. The FTIR analysis were only possible on the rest sample material and the documentation is not complete. Thus, currently, the best approach is to avoid analysis of contaminated material. Such strategy requires prescreening the samples with the help of ATR-FTIR or other analytical tools.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/RDC.2026.10250
Open access
