An ideal solution? Optimising pretreatment methods for artificially mummified ancient Egyptian tissues

Rationale Although the analysis of skeletal remains dominates the study of ancient dietary stable isotopes, mummified bodies also allow short‐term diet to be studied through the analysis of soft tissues. The application of resins, waxes and oils during mummification can affect the results obtained. This study assesses a range of methods for removing such substances from mummified tissue. Methods An experimental mummification model following ancient Egyptian methods was created using a modern pig leg. Sub‐samples of skin, muscle and bone were removed and coated with a range of substances used in Egyptian mummification. Four methods were used to clean these samples before the measurement of the carbon and nitrogen stable isotope ratios of their gelatinised collagen content using a ThermoFinnigan Flash Elemental analyser coupled to a DeltaPlus XL isotope ratio mass spectrometer via a ConFlo III interface. Results The results showed that embalming materials can significantly affect dietary stable isotope ratios, and that these substances are most effectively removed using a mixture of polar and non‐polar solvents. Results indicate that bone samples demineralised with HCl and skin samples produce more accurate results than bone samples demineralised with EDTA or muscle samples. Conclusions The choice of tissue and the preparation methods used can have a significant effect on the accuracy of stable isotope data obtained from mummified tissue, particularly when embalming materials are also present. A mixture of solvents appears to be a more effective cleaning agent than a single solvent. Demineralisation with HCl is preferable for well‐preserved bone, as used in this study, but whether this is the case for more fragile, less well‐preserved bone requires further study. Skin samples produce more consistent data than muscle, but visually distinguishing between these tissues is not simple on ancient mummies.

remains that were not subject to the complex artificial mummification process employed in ancient Egypt throughout most of the pharaonic period. The range of tissues frequently preserved in mummies (bone, tooth, skin, muscle, hair and nail) makes them a valuable resource for the study of both long-and short-term dietary isotopes that is not currently being used to its full potential.
The survival of mummified human remains in ancient Egypt is due to two factors: (1) the arid climate of the region, and (2) the range of physical and chemical techniques employed by the ancient Egyptians to artificially preserve their dead. There are a significant number of bodies from the pharaonic period and later that demonstrate varying degrees of preservation due to the simple placement of the body in the hot sand after death. Similar instances of natural mummification are also known from neighbouring Nubia, where preserved soft tissues and hair have been subject to some stable isotope analyses. 5,6 It is, however, the second method of preservation which has produced the large number of mummified bodies for which Egypt is well known. 7 Although the anthropogenic mummification process used by embalmers has been well studied, the implications of this process for a range of different methods of modern scientific analysis have not. This paper attempts to address this question with regard to carbon and nitrogen stable isotope analysis and to present some suggestions for ensuring future production of reliable analyses of mummified tissues.

| Ancient Egyptian mummification
The methods used to artificially preserve the body in ancient Egypt are well understood. [8][9][10] The main features of mummification were the removal of the internal organs, desiccation of the body using natron (a naturally occurring mixture of sodium salts found in dried lake beds) and the application of various unguents (salves) and resins.
Some of these resins are thought to have aided the mummification process by prohibiting the action of bacteria and/or by creating a physical barrier over the mummies' skin and internal cavities to prevent insect ingress. 11 Others such as myrrh were perfumed and are likely to have helped mask the smell of decay. The application of resinous materials to a mummified body is known to vary according to the quality of the embalming method employed and the historical age of the mummy. 7 Considerable analysis has been conducted to identify the constituents of embalming resins and their geographical origins using gas chromatography linked to mass spectrometry (GC/ MS). [12][13][14][15] As a result, several key resins and waxes have been identified as recurring throughout the periods when resin application was favoured (from the 18 th Dynasty onwards). These include beeswax, animal fats, pine resin and bitumen (often from the Dead Sea area 13 ), alongside more infrequently used substances such as pistacia resin and myrrh. 14 Many of the coatings found on mummified bodies have also been identified as mixtures of a series of these substances.
As the majority of ancient Egyptian mummies available for analysis come from the later periods of Egyptian history when the use of various resins was more common, it is important to determine whether these resins can be adequately removed from a sample prior to stable isotope analysis. As residual resin within a sample is organic in origin it has the potential to contaminate the stable isotope signature.

| The implications of artificial mummification for stable isotope analysis
The presence of resin within a soft tissue or bone sample has previously caused issues, in particular with material submitted for radiocarbon analysis. Both pine resin and bitumen have inherent radiocarbon ages and so must be removed prior to dating. The acidbase-acid pretreatment methods used commonly in AMS (Accelerator Mass Spectrometry) radiocarbon dating have been found not to successfully remove these contaminants, resulting in the production of anomalous radiocarbon ages. 16 A study from the University of Minnesota 17 did, however, demonstrate that benzene would successfully remove resin from samples submitted for radiocarbon analysis without affecting the radiocarbon age or stable isotope ratios of ancient material (liver, muscle, hair and linen). One of the two published stable isotope papers dealing with artificially mummified remains also reports attempts to remove contaminating resin, in this case from ancient Egyptian soft tissue and hair. 18 Different cleaning methods, based largely on sequences of chloroform and methanol, were applied to the samples with success judged by the degree of change in the δ 13 C values. In the absence of true baseline values for the ancient tissues, however, it is impossible to fully determine the success of these methods.

| Wider significance
The ability to reliably remove embalming materials from soft tissue is of fundamental importance for palaeodietary research. Bone samples are predominantly used in palaeodietary stable isotope analysis due to the preponderance of skeletal remains in the archaeological record compared with mummified remains. Using a combination of different tissue types, where available, can offer significantly greater insights into the diet of an individual or population, as the different tissue turnover rates allow dietary inference over different time periods.
Analysis of collagen from bone can provide information on diet over decades, whereas collagen from skin and muscle can provide information of the order of weeks to months. 19 Analysis of hair in 1-2 cm sections can provide information on a similar scale to muscle collagen but, by analysing sections from root to tip, it is also possible to gain insight into the seasonality of diet. 20 As such, the development of a successful preparatory method for mummified Egyptian tissues will potentially enlarge the array of material suitable for analysis and, in doing so, will greatly expand understanding of ancient Egyptian diet, health and lifestyle.
Despite being perhaps most commonly associated with ancient Egyptian mummies, the application of animal-and plant-based coatings is also seen in several other cultures . Anthropogenically   created mummies from Peru and Chile, the Canary Islands and   Australia have all shown evidence for the use of such embalming   materials, 21-23 making them potentially subject to the same inherent pre-treatment problems. A preparatory method that can remove the range of embalming substances used in Egyptian mummies would provide at least a starting point for the development of similar methods for mummies from these other cultures.

| Aim
For this pilot study, our approach is based exclusively on the use of a modern animal model rather than ancient tissues. This allows for the assessment of a range of cleaning methods and the provision of untreated and non-decomposed samples of skin, muscle and bone as controls that can be used to provide a reliable baseline for the interpretation of the other samples. Such a baseline cannot be provided for ancient tissues, as they are necessarily affected by a complex array of chemical changes caused by time, embalming methods, and degradation. Their use in preliminary testing of experimental methods such as that described herein is also inappropriate from an ethical standpoint.
Pig tissue was selected as the generally accepted closest analogue to human tissue and a leg from an organically reared pig was selected.
Organically reared pigs in the UK have a diet similar to the ancient Egyptian diet, i.e., mainly C3 plants, with a small quantity of C4 plants and fish protein. The leg was artificially mummified using a synthetic natron mix and, once desiccated, sub-samples were coated with various resins, oils and other substances, both alone and in mixtures, with uncoated tissues used as further controls. Hereon in, the different methods of preparing the tissues will be referred to as 'embalming treatments' and the different coatings used as 'embalming materials'. Five pretreatments (detailed in section 2) were then used in an attempt to remove these substances using common stable isotope and radiocarbon pretreatment methods, as well as other methods based on these protocols. Hereon in, these pretreatment methods will be referred to as 'cleaning methods'. We aim to identify a cleaning method that will remove contaminants effectively without causing undue damage to the tissue itself, or affect carbon or nitrogen stable isotope ratios.

| MATERIALS AND METHODS
A full rear leg weighing 9.85 kg with skin and foot intact from a freerange reared pig was obtained from a butcher. Samples of muscle, skin and bone were taken immediately to act as fresh tissue controls (embalming treatment no. 1) and stored at −80 C. As previous experiments have shown 24,25 that a ratio of around 5:1 (w/w) natron/ tissue is sufficient for mummification, 50 kg of synthetic natron was prepared from 25 kg anhydrous sodium carbonate (Extra Pure grade; Sigma, Gillingham, UK), 15 kg sodium hydrogen carbonate (VWR RECTAPUR, Lutterworth, UK), 5 kg sodium sulphate (VWR RECTAPUR) and 5 kg sodium chloride (99.5+%; Fisher, Loughborough, UK). This ratio is similar to that of naturally occurring deposits of natron from Egypt that have been shown to effectively promote mummification in previous experiments. 24 The leg was placed on a 5-7 cm thick layer of natron weighing approx. 15 kg, with the remaining approximately 35 kg covering it completely.
After 40 days, samples of skin, muscle and bone were taken from the lower leg, as the upper leg showed signs of significant decomposition. All samples were taken from sites immediately adjacent to each other to reduce the possible impact of sampling site on the stable isotope ratios obtained. These samples were further sub-divided into four parts, each receiving a different embalming treatment as detailed in Table 1.
Following the embalming treatments, the post-mummification samples were placed for 6 weeks in a food dehydrator set to 35 C.
Each piece of skin and muscle from each embalming treatment was then divided into five sub-sections, and each piece of bone into 10 (5 to be demineralised with HCl, 5 to be demineralised with EDTA).
One sub-sample of each was then either solubilised (soft tissues) or demineralised and solubilised without any cleaning method as follows:

| Solubilisation
Approximately 250 mg tissue was placed in 7.5 mL water (pH 3) for 48 h at 70 C. The samples were then filtered through an 8-μm Ezee filter, with the filtrate frozen at −20 C, then freeze-dried.

| Demineralisation with HCl
Bone samples were placed in 7.5 mL 0.5 M HCl for 5 days at 4 C. The samples were then rinsed three times in distilled water, then solubilised as above.

| Demineralisation with EDTA 26
Bone samples were placed in 30 mL 0.5 M EDTA (pH 8.0) for 5 days at 4 C, with the EDTA solution changed daily. The samples were then rinsed 15 times in distilled water, then solubilised as above.
The four cleaning methods being tested were then applied to the remaining subsections. Two of these cleaning methods (chloroform/ methanol and methanol/chloroform/water) have been previously reported in the literature for stable isotope analysis of mummified remains. Another (dichloromethane) is commonly used to extract resinous materials from mummified remains for analysis by gas chromatography/mass spectrometry (GC/MS). 27 This was included to test the possibility of using a single sample for both stable isotope and resin analysis. The last, toluene, was included as a similar but lesstoxic alternative to the use of benzene as recommended by Aufderheide et al. 17

| Chloroform/methanol 19
Samples were ultrasonicated in 10 mL distilled water for 1 h, then soaked in 2:1 chloroform/methanol for 2 × 48 h, with sonication for 1 h after each change of solvent. They were then washed in 3 × 10 mL distilled water, and demineralised or solubilised as above.

| Toluene
Samples were ultrasonicated in 10 mL distilled water for 1 h, then soaked in toluene for 2 × 48 h, with sonication for 1 h after each change of solvent. They were then washed in 3 × 10 mL distilled water, and demineralised or solubilised as above.

| Methanol/chloroform/water 18
Samples were soaked in 30 mL 10:5:4 methanol/chloroform/water at room temperature for 3 × 30 min. The samples were then soaked in 30 mL chloroform for 2 × 30 min, and 30 mL methanol for 2 × 30 minutes at room temperature. They were then air dried on filter paper at room temperature, and demineralised or solubilised as above.

| Dichloromethane
Samples were soaked in 10 mL dichloromethane for 2 × 48 h, with sonication for 1 h after each change of solvent. They were then washed in 3 × 10 mL water, and demineralised or solubilised as above.

| RESULTS
For this pilot study, resources were limited to the use of stable isotope analysis. Chromatographic or spectroscopic methods that could identify residual embalming materials following cleaning were not available. As such, the efficacy of each cleaning method is judged according to two criteria. The first is the ratio of carbon to nitrogen in the gelatinised collagen under analysis. The commonly accepted range for C/N ratios is 2.9-3.6, 28 with ratios outside this range indicating contamination. The second criterion is the comparison of stable carbon and nitrogen isotope ratios from samples subject to embalming treatments 2-5 with the fresh control samples (embalming treatment 1). Table 3 shows the results of the stable isotope analyses of the materials used in embalming treatments 3-5. Despite a limit of detection of 0.1% for nitrogen, no nitrogen content was detected in any of these samples. As such, across the embalming materials and other reagents used, the only source of exogenous nitrogen is the EDTA used in demineralising one set of bone samples. The inclusion of residual natron within these samples is certainly a possibility, either from encrustations that were not removed when the samples were brushed clean, or from natron that had been absorbed by the tissue itself. That the depletions in 13 C are most evident in those samples cleaned using the methods with the least amount of rinsing and soaking in water is certainly supportive of natron being at least a contributory factor. Unfortunately, the natron itself was not subject to stable isotope analysis at the same time as the tissue samples and embalming materials, so we cannot be sure of its likely impact if not fully removed.

| Skin
However, the results for the skin samples that underwent embalming treatment 3 offer further support for the suggestion that residual natron is present. The chloroform/methanol and methanol/ chloroform/water cleaning methods again produce samples with C/N ratios and δ 13 C and δ 15 N values very similar to those of the baseline values obtained from the fresh controls. Toluene cleaning resulted in a sample with a C/N ratio above the acceptable range, and depleted in 13 C, while dichloromethane results in an acceptable C/N ratio, but even greater depletion in 13 C than was seen for sample S3T. The root cause of these results is again difficult to explain, particularly that for S3DCM where the C/N ratio does not suggest any contamination, and no material with exogenous nitrogen was introduced. The results for S3N are strongly suggestive of residual natron. The C/N ratio is far above the acceptable range (9.1), and the sample is very depleted in 13 C. The δ 13 C value of −27.4‰ is only slightly higher than that of pure pine resin (−27.5‰). That the sample had a nitrogen content of 7%, whereas embalming material 3 contained no measurable nitrogen content, suggests that the depletion seen is due to another carbonbased contaminant even more depleted in 13 C than the embalming material. For this experimental procedure, the only plausible source of this contaminant is the natron, which would appear to be highly depleted in 13 C.
Sample S4N shows a similar but less extreme set of results.

| Muscle
The results of the stable isotope analyses of muscle samples are shown in Table 5. Unlike the skin samples there is considerable variation in the results seen for embalming treatment 1, i.e. the fresh controls. Those samples undergoing no cleaning method (M1N) and cleaning with toluene (M1T) produced C/N ratios beyond the acceptable range. As no exogenous materials were added to these samples this indicates both that some cleaning is required for muscle samples, and that toluene is not a sufficient solvent using the method applied here. This is somewhat surprising, as the triacylglycerols and sterols that make up the great majority of endogenous lipids in animals are generally very soluble in non-polar solvents such as toluene. The sonication stages in the cleaning method also noticeably warmed the solutions, which would have further increased the solubility of the lipids.
The remaining three samples, those cleaned with chloroform/ methanol, with methanol/chloroform/water, and with dichloromethane, all had C/N ratios within the acceptable range. Their isotope ratios varied substantially, however, as can be seen in Table 5.
With no additional data to inform which of these represents a more accurate baseline result than the others, a simple mean of the results has been used to provide this baseline, with the results from the other embalming treatments interpreted accordingly. further analysis, it is not possible to say for sure whether this is due to endogenous lipids, embalming materials, or residual natron. The extreme variation seen in some of the skin samples is not seen for muscle. If that was indeed, as hypothesised, due to natron contamination, this may be because only the skin was in direct contact with the natron. This will certainly have prevented clumps of natron from adhering to the muscle tissue and being carried forward to analysis. The samples that underwent no cleaning method or that were cleaned in dichloromethane are not notably more depleted in 13 C than those that had been cleaned with other methods, as was seen in skin. This suggests that natron is not a major source of contamination, as the lack of washing in water in the skin samples was associated with much lower δ 13 C values. Whether the contamination seen is due to endogenous lipids or exogenous embalming materials is not clear, and further analysis would be required to establish this.
Only two mummified samples had acceptable C/N ratios: M3T and M5T. These produced similar stable isotope ratios, as can be seen in Table 5. Although the nitrogen results are similar to the mean baseline identified from the control samples, the carbon isotope ratios are lower, at around −25‰ compared with just under −24‰ for the controls. Even with the variation seen in the control samples, the mummified samples appear to be quite depleted in 13 C. Interestingly, the only samples that provided results similar to the calculated baseline results were those cleaned with chloroform/methanol and with methanol/chloroform/water. This is true across all the embalming treatments, other than sample M5CM, which appears enriched in 15 N. Although this could be taken as suggesting that these cleaning methods are more reliable than the others, the combination of high C/N ratios for these samples and the lack of a clear baseline value against which to judge the results means that this is far from conclusive.
A possible explanation for the heterogeneity of the muscle isotope ratios observed is linked to tissue turnover rates. Muscle turns over on average every 40-60 days, 29 quicker than dermal collagen, which turns over approximately every 2-4 months. 30 It is thus possible that the shorter turnover time could result in differences in the muscle samples taken in terms of the period of formation that they represent. The increased consistency of isotope ratios seen in the skin and bone samples may be a result of the increased chance of selecting material that represents equivalent periods of formation.
T A B L E 6 Results of the stable isotope analyses of bone from the mummified pig leg, demineralised using HCl

| Bone
The results of the analysis of bone samples demineralised with HCl are shown in somewhat depleted in 13 C compared with the bone samples demineralised in HCl.
Insufficient demineralisation of bone is one possible cause for this variation, but apatite is enriched in 13 C over the collagen fraction. 31 Lipids are usually 13 C-depleted relative to collagen, 32 as are the embalming materials used herein; so these in isolation or combination are the more probable cause. This is borne out by the generally higher δ 13 C values obtained for the samples in embalming treatment 2. The extent of contamination must be quite low, however, because the C/ N ratios are within the acceptable range.
The extensive washing step at the end of the EDTA demineralisation stage makes it unlikely that residual EDTA contamination is the cause of this variation. Rather, the slightly higher, although still acceptable, C/N ratios of EDTA-demineralised samples over the HCl-demineralised samples suggests that there may be some lipid contamination. The results for skin and bone samples do show some clear trends.

| Evaluation of cleaning methods
In general, cleaning methods using a combination of polar and nonpolar solvents rather than a single solvent are more effective at removing both endogenous lipids and embalming materials. Being naturally derived materials, the embalming materials used in this study (and in ancient Egypt) would comprise a wide range of different substances. Some are similar to those found in the bodies themselves, such as palm oil, which is primarily made up of triacylglycerols, 33 whereas others are very different. Beeswax, for example, includes a variety of esters, diesters and hydroxyesters. 34 Resins contain large quantities of terpenes, with pine resin having large quantities of mono-and di-terpenes, 35 and pistacia resin being mainly triterpenes. 36 The range of solubilities displayed is substantial. Wax esters, triacylglycerols and some terpenes will be generally more soluble in non-polar solvents such as toluene or chloroform, for example, whereas other terpenes are more soluble in polar solvents.
Fatty acids are generally insoluble in water at lower pH when they are fully protonated, whereas at higher pH they may be somewhat soluble.
As the single-solvent cleaning methods used herein are based on non-polar solvents, there is certainly a risk that some contaminants will not be removed. A combination of polar and non-polar solvents allows for a wider range of substances to be removed, especially when the samples are repeatedly washed with water to remove residual natron. The apparent improvement in cleaning seen in those bone samples demineralised with HCl over those demineralised with EDTA may be due to the hydrolysis of esters and triacylglycerols to more polar molecules that are either more soluble in a particular solvent, or are soluble in a wider range of solvents, facilitating their removal. The EDTA solution used would not have allowed for the same level of hydrolysis in these substances, which may help to explain the greater variance in results seen in EDTA-demineralised bone.

| CONCLUSIONS
The results from this model show that stable isotope ratios can be significantly affected by mummification and the application of resins, waxes and oils. If the substances responsible for this effect are not removed, the results obtained could lead to the collection of misleading data. The experimental model presented here shows that tissue type, demineralisation method and cleaning method can also affect the data obtained.
Bone, when demineralised with hydrochloric acid, produces the most consistent results, probably because the demineralisation stage allows for improved removal of natron residues, and may increase the solubility of embalming materials and endogenous lipids. Greater variance is seen in bone samples demineralised with EDTA, suggesting that this demineralisation method is less effective at providing an initial cleaning stage than HCl demineralisation. Skin also produces generally consistent results when cleaned with a mixture of polar and non-polar solvents. The single-solvent methods tested are not, however, able to reliably remove embalming mixtures or natron residues.
Muscle, however, could not produce reliable results in this experimental model. Significant variation is seen across the results obtained for this sample, with possible causes including lipid contamination, ineffective removal of embalming materials or natron residues and poor preservation. The pretreatments used are not consistently successful in removing these contaminants, making it impossible to identify a reliable method for cleaning this tissue. As it can be difficult to distinguish between ancient mummified soft tissues without histological analysis, the refractory nature of muscle in comparison with skin may be particularly problematic.
In summary, skin and bone appear to represent the most reliable tissues for stable isotope analysis of mummified remains.
Demineralisation with HCl for bone and pretreatment with a mixture of polar and non-polar solvents produce accurate data from samples prepared using all the methods presented herein. However, muscle and bone that have been demineralised with EDTA have not consistently produced accurate data.