Non-destructive Dendrochronology: The Effect of Conservation Agents on Tree-ring Measurements in Archaeological Oak with Micro-computed Tomography

ABSTRACT The effect of different conservation methods on non-destructive dendrochronology with micro-computed tomography (μCT) of waterlogged archaeological wood (WAW) was assessed. To this end, oak samples were conserved with alcohol-ether-resin, Kauramin 800®, lactitol/trehalose, saccharose, silicone oil and different polyethylene glycol (PEG) treatments with subsequent freeze-drying. Tree-ring measurements were compared with respect to the number of rings and the mean ring width using (a) an analog linear measuring table and (b) the digital μCT. Overall, the measurements in the μCT data agreed very well with the corresponding microscopic measurements. It was possible in all cases to recognize the rings in the samples with μCT. A dendrochronological cross dating with regional absolutely dated oak ring width chronologies was successful. However, the varying influence of conservation agents on the quality of the μCT data were also evident. To quantify this influence, the contrast in the μCT data were calculated using gray-scale profiles. An influence on the contrast in the μCT data were detectable for all conservation agents. The contrast is especially reduced due to conservation methods using lactitol/trehalose, PEG, saccharose and silicone oil. Overall, the experiments confirm, though, that μCT is a powerful and accurate tool for dendrochronology of conserved archaeological oak objects.


Introduction
Archaeological objects that were crafted from wood are preserved in the temperate zone mainly under waterlogged and anoxic conditions, where microbial decay is slowed down, and the wood structure is more or less completely filled with water.Depending on the degree of degradation, waterlogged archaeological wood (WAW) may collapse and shrink when it is allowed to dry in an uncontrolled manner after excavation, leading to a total loss of the object's original shape and its information.A variety of methods and conservation agents have been tested for the conservation of WAW and several comparative studies aimed to compare the efficiency of the treatment methods (Christensen, Kutzke, and Hansen 2012;Walsh et al. 2014Walsh et al. , 2017;;Broda and Hill 2021;Stelzner et al. 2022).
As wooden objects have a long history in human development and technological evolution, they bear essential information in archaeological research.As plants adapt to environmental conditions throughout their lives and store this information in their tissue, analysis of the wood materials allow reconstruction of the living conditions of the trees.Archaeological wood offers an archive for interdisciplinary research.
Ecological, economic, vegetation-historical, climatic and cultural-historical questions can be addressed.Furthermore, dendrochronology is a widespread, reliable, and very accurate method for dating archaeological wooden finds and, therefore, an essential tool for archaeological research (Schweingruber 2001;Baillie 2002;Hughes 2002;Speer 2010;Billamboz 2014).When the outermost tree ring under the bark (waney edge) has been preserved, both the felling date and the felling season can be determined (Sass-Klaassen, Vernimmen, and Baittinger 2008).The width of the tree rings is usually measured on the cross-section with an accuracy of 1/100 mm (Westphal and Heußner 2016).For conventional dendrochronological dating, the artifact has to be sawn or drilled with an increment borer, which is contrary to the aim of conservation to preserve the object unaltered and as a whole.Also, preparing the cross-sectional surface with a razor blade for a dendrochronological dating is an invasive procedure on an artifact (Sass and Eckstein 1994;Kuniholm 2001;Bräker 2002;Westphal and Heußner 2016).
Among the non-destructive methods currently available, X-ray computed tomography (CT) is an appropriate technique that allows inspection of the entire volume of the sample at different spatial resolutions (Withers et al. 2021).The technique provides for archaeological sciences very promising information in the field of dendroarchaeology, including conservation science (Stelzner et al. 2021(Stelzner et al. , 2022)).In addition to being noninvasive, it also requires almost no sample preparation and enables both qualitative and quantitative threedimensional (3D) tree-ring analysis (Van den Bulcke et al. 2014;Martinez-Garcia et al. 2021, 2022).For these reasons, CT has made great progress as a leading non-destructive microscopic technique in the field of dendrochronology.
First attempts in non-destructive dendrochronological dating based on CT images were pursued in the 1980s (Onoe et al. 1984).However, non-destructive analysis and successful dating of wood samples only became feasible with the development of industrial µCT with small focal spot size X-ray tubes and high-resolution detectors, which greatly improved the spatial resolution of X-ray CT systems (Okochi, Fujii, and Mitsutani 2007).For instance, the oak figures from Fellbach-Schmiden were dated to 127 ± 10 BCE (Keefer et al. 2006) and the oak figure from Eschenz to 9 ± 10 BCE (Belz et al. 2008).In both cases the sapwood was present on the oak.The felling date can then be narrowed down to 20 years which is written as the date ± 10 (Hollstein 1980).Grabner et al. (2007Grabner et al. ( , 2009) ) managed dating archaeological objects from the archaeological site Hallstatt made of fir (835 BCE, no waney edge), beech (1378 BCE, waney edge dating) and silver fir (1362 BCE, no information about the waney edge).Furthermore, the dendrochronology of wood in blocks of soil was explored with μCT (Stelzner and Million 2015), enabling the dating of fragments of oak (CE 500, no waney edge).Kastner et al. (2007), Grabner et al. (2007Grabner et al. ( , 2009)), Bill et al. (2012) and Stelzner and Million (2015) in the same way, reported on the limits of the measurability of ring widths, as the image resolution depends on the object size.
A large-scale study on more than 90 oak objects confirms the potential of µCT for non-destructive dendrochronological analysis, with only six objects remaining undated (Bill et al. 2012).However, an important factor affecting the validity of the results is the condition of the objects.The waterlogged finds examined by Bill et al. (2012) had been air-dried, because the measurements of the water-saturated timbers and of the samples impregnated with high-molecular polyethylene glycol (PEG) did not produce sufficient contrast for a determination of tree-ring widths.The problem of conserved WAW with regards to non-destructive dendrochronology was confirmed by Wiesner et al. (2016) in the case of a prehistoric wheel made of maple, which had been conserved with 40% PEG 2000 and freeze-dried.In contrast, Daly and Ebert (2021) successfully dated a lid of an oak barrel from Odense, Denmark, conserved with 38% PEG 2000 and subsequently freeze-drying, to 1332 CE (no waney edge) based on 217 measured tree rings.Apart from this case, the figures from Eschenz (alcohol-ether-resin method) (Belz et al. 2008) and Fellbach-Schmiden, which had been conserved with obsolete materials like Lyofix (Keefer et al. 2006), no conserved archaeological objects have been dated so far by CT and therefore also Bill et al. (2012) recommend that further experiments should be carried out with CT and conserved wood for dendrochronology.
The principal aim of conservation is to bring an artifact into a condition of long-term chemical stability and to achieve a homogeneous mechanical stabilization of the structure.However, in the case of archaeological wood, treatment may impact the future non-destructive dendrochronological analyses with CT.Due the difficulty of analyzing archaeological wood in a wet condition the question arises whether the structural information of the historical sources is still accessible and evaluable after conservation.Therefore, this study aimed to elucidate which common conservation method allows nondestructive dating with µCT, and what influence these methods have on the visibility of the annual rings in the µCT data.Since oak is the most important wood species for dendrochronology in Europe (Towner 2002;Haneca, Čufar, and Beeckman 2009), the conservation methods were carried out on oak samples.To check the results of the ring width measurements they were compared to analog conventional tree-ring measurements under the stereo microscope.

Samples
For this study, conserved oak samples from the Roman period were available from the scientific reference collection of the Leibniz-Zentrum für Archäologie in Mainz, Germany.The collection was built up between 2008 and 2011 by collecting large archaeological samples from central European sites, of different wood species and degrees of degradation.These samples were then divided into several equal sized samples and impregnated with the most established conservation methods at that time (Wittköpper et al. 2016;Stelzner et al. 2022).The samples were conserved at different institutions with their standard methods: alcoholether-resin (Bräker et al. 1979, Schmidt-Ott, André, andBader 2022), melamine-formaldehyde (Kauramin 800) (Wittköpper 1998), lactitol/trehalose (Imazu and Morgós 2002), saccharose (Parrent 1985), silicone oil (Smith 2003) and polyethylene glycol (PEG) with subsequent freeze-drying.PEG treatment followed either a one-step process with PEG 2000 (Jensen, Petersen, and Straetkvern 2011), two-step with PEG 400 and PEG 4000 (Cook and Grattan 1990) or three-step with PEG 400, PEG 1500 and PEG 4000 (www.rgzm.de/kur).The degree of degradation was determined by the maximum water content (Umax) in percent and the basic density in g/cm 3 (Stelzner et al. 2022).A detailed overview of the conservation methods and the samples is given on the project homepage (www.rgzm.de/kur).For the analyses in this study, the three largest oak (Quercus robur, Q. petraea) test series of the collection were selected in order to cover a high variety of methods (Table 1).This systematic collection of samples provides a unique chance to compare the structural differences of the conserved wood (Wittköpper et al. 2016;Stelzner et al. 2021Stelzner et al. , 2022)).Each test series of the collection includes samples derived from the same object or context (same wood species and excavation site, similar state of preservation) that were conserved with the different treatments.Due to the different sizes of the objects, the number of samples varied.Therefore, not all test series contain all conservation methods.
The sample names in Table 1 indicates from which oak object and test series (Oa1, Oa2 and Oa3) the sample originates and with which conservation agent it was conserved.In addition, the KUR number is given in order to identify the sample in the database on the project homepage (www.rgzm.de/kur).

Micro-computed tomography
The tomographic analysis was performed on the inhouse laboratory X-ray µCT system (Diondo d2, Germany) at the Lucerne University of Applied Sciences and Arts.An optimized setup and acquisition protocol for the µCT measurements was developed for conserved wood.The measurements were conducted by setting the X-ray source (XWT-225 TCHE + from X-ray works, Garbsen, Germany) in high power mode and choosing an operation voltage of 120 kV and a filament current of 167 µA with a 1 mm aluminum filter.The wood samples were mounted in a sample holder and placed in the sample chamber.The sample was rotated 360°a nd measurements were performed in the continuous mode during the acquisition.The radiographical projections were recorded with a 4343 DX-I X-ray detector (Varex, Salt Lake City, U.S.A.), with a pixel size of 139 µm.The distances between the X-ray source and the sample varied between 160 and 250 mm and between the X-ray source and the detector was 860 mm, giving a magnification between 3.4 and 5.4 and a nominal voxel size between 27 and 44 µm.A total of 3000 projection images were acquired during the sample rotation of 360°.The resulting projections were converted into a 3D image stack of approx.3000 × 3000 × 3000 voxels using the CERA reconstruction software based on the filtered back projection Feldkamp algorithm (Feldkamp, Davis, and Kress 1984) from Siemens.The achieved resolution of the µCT measurements depends on the size of the samples (Stelzner and Million 2015).Table 1 provides an overview of the samples, their size and the achieved resolution of the µCT measurements.

Tree-ring width measurements
Ring width measurements in the image cross-sections of the µCT data were performed with VGStudio Max 3.4©.For each sample two radii were measured.
After µCT measurement, the samples were prepared for the conventional analog manual measurements.For this purpose, the surface was leveled and polished with the aid of a geared eccentric sander (Festool, RO150FEQ ).An ascending grit of 40, 80, 180, 400 and finally 800 was used for sanding the sample surfaces.Only in a few cases a preparation of the surface with a conventional razor blade was necessary.Subsequently, the annual ring widths were measured with a stereo microscope on a linear measuring table along two radii (Rinntech Lintab).For the validation of both measurements, the tree ring curves of each radius per sample of the analog and the µCT measurement were compared to each other (Figures 1-3).
In the next step the number of rings of the analog and µCT measurements were compared.In this way, it was checked whether all rings could be detected in the µCT cross-sections.Additionally, the arithmetic mean of the tree-ring widths was calculated.With the purpose to get a ratio easy to compare and to interpret, the number of tree rings and the arithmetic mean of the ring width, the individual value for the digital measurement was subtracted from the analog, by achieving a calculated difference as a measure for the detection of divergences.

Irregularly shaped tree-ring width measurements
For samples exhibiting irregularly shaped rings (i.e., irregular annual growth), an alternative tree-ring width measurement approach based on image processing algorithms is used (Martinez-Garcia et al. 2021, 2022).From the µCT image cross-section, the approach enables to detect first the pixels located along each annual ring (i.e., the ring pixels).Once recognized these pixels, their coordinates, (x,y), are extracted from the µCT image cross-section.This data is then used to compute the distance between two consecutive rings, along the radial direction (i.e., along the line running from the pitch to the bark).Since these coordinates are points lying along the entire annual ring, the distance between two consecutive rings is calculated along a large number of radial directions (usually more than thousand directions are considered).Finally, the ring width distances obtained along all radial directions are averaged and its mean value is taken as the measure of the tree-ring width.In this way, the approach provides a more realistic tree-ring width estimate than considering only few radii in the measurement.All algorithms used were implemented in Python 3.7, supported by tools provided by the open-source image processing libraries scikit-image and OpenCV.This measuring method is applied here as an example to two samples of the branches with irregularly shaped tree ring of the third series of samples (Oa3): the sample conserved with alcohol-ether-resin (Oa3-AlEt) and the sample conserved with silicone oil (Oa3-Sil).

Gray value analysis
To get information about the influence of the conserving agent on the visibility of the rings in the µCT images the quality of the data were evaluated by calculating the contrast.For this purpose, the data were calibrated during import into VGStudio MAX 3.4 by  defining the background (air) and the most strongly absorbing part of the wood (e.g., medullary ray) to specific gray values.The distribution of the gray values possible with 16-bit data sets is 65535.Within the possible 65535 gray values, the background (air) was set to the gray value 10000 and the material (wood) to the gray value 50000 when calibrating the measurements.In µCT data, gray value profiles were taken along the radii of the ring width measurements.The distances were equal for each test series and the profiles were taken without gaps or inclusions in the wood that might interfere with the result (Figure 4).The values were exported and the image contrast was calculated by dividing the difference of the maximum and   minimum by the mean value of the gray value profiles (Peli 1990).

Tree-ring width measurements
In the first test series, the sample without conservation (Oa1-Air) illustrates the need for conservation.Figure 5 clearly shows how the heavily degraded areas of the sample have completely collapsed in the outer region.In contrast to the conserved samples, tree-ring widths can no longer be measured here.The inner areas of the air-dried specimen are very well preserved.Though, no strong shrinkage and no collapse of the wood can be observed here.
The results of the ring width measurements show that in all conserved samples the rings could be captured on the basis of µCT data.The resolution of 19-44 µm achieved in the various measurements made it possible to detect ring widths down to the smallest ring width of 0.18 mm, regardless of the conservation method used.Moreover, in some cases it was possible to detect some rings at the end of the radii better with µCT than with the conventional measurement method (Table 2).The reason for this is that the  heavily degraded and partially erupted edge areas were in some cases better visible in the µCT data than on the surface (Figure 6).The surface was also partially smeared by the conservation agent, as illustrated by the smooth area at the edge of the sample in Figure 7.In the µCT data, the rings in these areas were more clearly visible in the axially slightly lower planes.
Here, the µCT data also allowed a broader inspection of ring widths in both horizontal and vertical dimension, in a 3D way, while the conventional measurement method is restricted to the cross-section which is cut physically only once.
To get an impression of the accuracy of the ring width measurements in the µCT data the results were compared with the conventional measurements in term of the mean ring widths.Overall, the ring width measurements in the µCT data agreed well with the microscopic measurements (Figure 8).The deviation for all samples in the first two test series (Oa1 and Oa2) showed only minimal overall deviations of less than 0.1 mm.In terms of the tree-ring widths, this corresponds to a maximum deviation of approximately 5%.The larger deviations over 0.1 mm in the third test series (Oa3) can probably be explained by the more eccentric growth of the tree rings in the examined samples of an oak branch.Such irregular growth can be compensated for by the application of algorithms, which enable the analysis and measurement of entire rings as a result.Figure 9 shows a comparison of tree-ring widths measurements using a recently proposed entire-profile automated approach (Martinez-Garcia et al. 2021, 2022) and considering only two orthogonal radii using the VGStudio Max 3.4© software.The measurements were performed on image cross-sections of µCT data obtained for samples conserved with alcohol-ether-resin (Oa3-AlEt) and silicon  oil (Oa3-Sil) of the third test series.The results show that in both samples the agreement between manual and automated measurements are not good enough at some annual rings.For the sample Oa3-AlEt (Figure 9(a)), a deviation of 0.4 mm between both ring-width measurements was obtained at the third ring-width point (i.e., ring width measured between the rings No. 3 and 4 counted from the pith to the bark), whereas for the sample Oa3-Sil (Figure 9(b)), deviations between 0.4 and 0.6 mm were observed for ring-widths No. 2, 7, and 8.These results indicate that for irregularly shaped rings, ring width measurements along only two orthogonal radii are not always accurate enough and averaging over the entire ring profiles can provide a more realistic measurement.
On the basis of the measured annual rings the samples from the first test series (Oa1) could be dated to 86 CE (with waney edge) and the samples of the second test series (Oa2) to 115 BCE.
Due to the small number of rings, no dendrochronological dating of the third series of samples (Oa3) was possible.

Visibility of rings in µCT data
Overall, the annual rings were visible in the µCT data of all samples.Differences were nevertheless obvious for the sample series themselves and for the different conservation agents.This was confirmed by the gray value profiles and the contrast calculated from them.Evident was the poorer contrast in the µCT data of the third series of samples (Oa3).These differences in the test objects themselves can be explained to some extent with the condition of the wood.The more degraded Figure 9. Average tree-ring widths computed over thousands of radii along the entire ring profile (automated approach) and over two arbitrarily chosen orthogonal radii (manual method) for wood samples conserved with (a) alcohol-ether-resin (Oa3-AlEt) and (b) silicone oil (Oa3-Sil).The detected ring positions used in the automated approach are highlighted in red.Annual rings are numbered starting from the center.
samples from the third test series (Oa3) show the lowest contrast.The trend of poorer contrast for more degraded areas in the sample is also seen in the other two series of samples (Oa1 and Oa2).Another explanation for the contrast reduction is that samples of the third series (Oa3) are taken from branches.Here, the pores of the early wood are less pronounced than those of the other two series of samples (Oa1 and Oa2), due to the tapering effect that is triggered by the distance to the soil floor (Koçillari et al. 2021).
When considering the different conservation methods, it is first noticeable that all of them have an influence on the contrast in the µCT data.Thus, a decrease in contrast can be clearly determined for all conservation agents compared to the non-conserved samples.The highest contrast values after the measurements without any conservation agents are observed for the samples conserved with alcohol-ether-resin (Figure 10).
Lactitol/trehalose treated samples and especially the ones treated with silicone oil had the lowest contrast values.The influence of conservation is most evident in the third series of samples (Oa3), where there is for example a clear difference between the sample conserved with alcohol-ether-resin (Oa3-AlEt) and the sample conserved with silicone oil (Oa3-Sil) (Figure 9).One explanation again could be the less pronounced pores of the early wood of the branches in this test series, which could be filled by the silicone oil, thus reducing the contrast of the measurement.The hypothesis that the conservation methods that tend to fill the pores of the wood lead to reduced contrast µCT data seems to be plausible.This hypothesis is supported by the observations that full impregnation of oak with PEG as described by Bill et al. (2012) prevents the visualization of the rings using µCT, whereas conservation with PEG followed by freeze-drying still allows the ring analysis as described here and by Daly and Ebert (2021).In contrast, the rings of the maple wheel conserved with the same method by Wiesner et al. (2016) could not be visualized, which can be explained by the different wood species and the smaller pore size (Grabner, Salaberger, and Okochi 2009).

Conclusion
The results of this study confirm that all annual rings of the oak samples could be measured in the µCT data.The tree-ring series of two test series (Oa1 and Oa2) could also be successfully cross dated to reference chronologies.The achieved resolution in the µCT data were sufficient to detect all annual rings down to the smallest tree-ring width of 0.18 mm.The comparison with the conventional measurement confirmed the accuracy of the method.Furthermore, in some cases it was possible to obtain more rings at the outer areas of the samples using µCT than conventional microscopy due to the three-dimensionality of the data.Moreover, annual rings in the strongly degraded and compressed areas of the samples could be better identified in the µCT data and additional ring widths could be measured.Furthermore, the exemplary applied automatic measurement of tree-ring widths in the µCT data provides more accurate results for irregularly shaped tree rings.The varying influence of the reviewed conservation agents on the quality of the µCT data were also clearly evident.The best results could be achieved on the samples conserved with the alcohol-ether-resin method.However, an influence on the contrast in the µCT data were detectable for all conservation agents compared to the non-conserved samples.This influence seems to be especially present for the conservation methods using lactitol/trehalose, PEG, saccharose, and silicone oil, which have the property of filling the pores of the woods (Stelzner et al. 2023).Due to the large earlywood pore size of oak, the influence of the conservation methods investigated here seems to be small.So far, only conserved objects made of oak have been successfully dated using µCT.Therefore, further research is needed on other wood species whose ring boundaries are less easy to recognize.
In conclusion, the experiments confirm that µCT is a powerful and accurate tool for dendrochronology of conserved WAW.Successful ring width measurements are dependent on the quality of the µCT data.Besides resolution, the contrast of the measurement is decisive for the visibility of the annual rings.To obtain sufficiently good µCT data, however, certain factors must be considered: In addition to the equipment used (Bill et al. 2012;Křivánková, Nasswettrová, and Šmíra 2018;Mori, Kuhara, and Suzuki 2020), the quality determining factors also include the type of wood and the diameter of the object to be examined (Grabner et al. 2007(Grabner et al. , 2009;;Stelzner and Million 2015).As this study shows, for conserved WAW finds, in addition to the above criteria, the condition of the wood and the conservation method used are relevant for successful dating using µCT.lishment of the reference collection for the conservation of waterlogged wood was funded by the Federal Cultural Foundation and the Cultural Foundation of German States (KUR Programme).

Notes on contributors
Jörg Stelzner is a scientist in the research project CuTAWAY (conservation and wood analyses) at the Leibniz-Zentrum für Archäologie (LEIZA) and a freelance objects conservator.He was trained as an objects conservator and holds a PhD from the State Academy of Art and Design in Stuttgart, Germany.His research focuses on the application of computed tomography techniques in archaeology and conservation science.Address: Leibniz-Zentrum für Archäologie, Ludwig-Lindenschmit-Forum 1, 55116 Mainz, Germany.Email: joerg.stelzner@leiza.de.
Ingrid Stelzner leads the research project CuTAWAY (conservation and wood analyses) at the Leibniz-Zentrum für Archäologie (LEIZA).Ingrid was trained as an objects conservator and holds a PhD from the State Academy of Art and Design in Stuttgart, Germany.Her special interest focuses on the conservation of archaeological organic objects.She is currently assistant coordinator of the ICOM-CC Group on Wet Organic Archaeological Materials.Address: Same as Jörg Stelzner.Email: ingrid.stelzner@leiza.de.
Jorge Martinez-Garcia is a senior scientist at the Competence Centre of Thermal Energy Storage (TES) of the Lucerne University of Applied Science and Arts.He graduated in physics at the University of Havana and has a PhD in materials science from the University of Trento.His research focuses on the quantitative analysis of the microstructure and transport properties of materials using computed tomography imaging and X-ray powder diffraction technologies.He has a track record of using and developing advanced imaging and diffraction methods as well as mathematical and numerical models.Address: Lucerne University of Applied Sciences and Arts, Technikumstrasse 21, 6048 Horw, Switzerland.Email: jorge.martinezgarcia@hslu.ch.
Sebastian Million is a scientist at the tree-ring laboratory of the Landesamt für Denkmalpflege (State Office for Cultural Heritage) Baden-Württemberg in Hemmenhofen.He is analysing archaeological wood and bark from excavations in the state of Baden-Württemberg.He has a diploma in forestry from the University of Freiburg, Germany.Adress: Landesamt fuer Denkmalpflege, Fischersteig 9, 78343 Gaienhofen-Hemmenhofen, Germany.Email: sebastian.million@rps.bwl.de.
Damian Gwerder is a scientist at the Competence Center of Thermal Energy Storage (TES) of the Lucerne University of Applied Science and Arts.He graduated in mechanical engineering from the same university before joining the competence center as a scientist.In his research work, he develops novel protocols for the optimum acquisition of X-ray computed tomography experiments as well as sample holder to enable a reliable measurement of the sample under various environmental conditions including elevated temperatures, high/low humidity, and pressure loads.Address: Same as Jorge Martinez-Garcia.Email: damian.gwerder.01@hslu.ch.
Oliver Nelle heads the Tree-ring laboratory of the Landesamt für Denkmalpflege (State office for Cultural Heritage) Baden-Württemberg in Hemmenhofen, where he and his team analyze the archaeological wood finds of SW-Germany, with a focus on prehistoric lake dwellings.He has a PhD from the University of Regensburg and research questions of humanenvironment-climate interactions since the last glacial period.Address: Landesamt für Denkmalpflege, Fischersteig 9, 78343 Gaienhofen-Hemmenhofen, Germany.Email: oliver.nelle@rps.bwl.de.
Philipp Schuetz is a lecturer and research group leader at the Competence Center of Thermal Energy Storage (TES) of the Lucerne University of Applied Science and Art.He graduated in mathematical physics from ETH Zurich and has a PhD in theoretical biophysics from University of Zurich.Together with his team he develops novel methods for X-ray imaging with a particular focus on samples with low density contrast and the quantitative analysis of resulting image stacks in the fields of conservation and thermal storage materials.A second branch of his group is focused on energy data science and the assessment of thermal energy storage systems.Address: Address: Same as Jorge Martinez-Garcia.Email: philipp.schuetz@hslu.ch.

Figure 1 .
Figure 1.Measurement radii of tree-ring widths, analog on the prepared surface of the sample conserved with saccharose from the first test series (Oa1-Sac).

Figure 2 .
Figure 2. Measurement radii of tree-ring widths in the µCT data of the sample conserved with saccharose from the first test series (Oa1-Sac).

Figure 3 .
Figure 3. Tree ring curves of the measured radii from the different methods of the sample conserved with saccharose from the first test series in synchronous position (Oa1-Sac).

Figure 4 .
Figure 4. Gray value profile (50 mm) along one tree-ring widths measurement radius in the calibrated µCT data of the sample conserved with saccharose from the first test series (Oa1-Sac).

Figure 5 .
Figure 5. µCT cross-sections of the non-conserved sample from the first test series (Oa1-Air).

Figure 6 .
Figure 6.Breaks and voids of the outer areas in the surface of the three-step PEG conserved sample of the first test series (Oa1-PEG3) (left) and µCT cross-section of this part (right).

Figure 7 .
Figure 7. Degraded parts and smeared surface in the outer areas of the two-step PEG conserved sample of the first test series (Oa1-PEG2) (left) and µCT cross-section of this part (right).

Figure 8 .
Figure 8. Difference between the arithmetic mean tree-ring widths from µCT to analog measurement.

Figure 10 .
Figure10.Deviation of the mean contrast in the µCT data calculated from the gray value profiles of the two radii of the tree-ring measurements.

Table 1 .
Information on the conservation, condition, dimensions and µCT resolution of the oak objects and samples.