Elemental compositions of papyrus removed from ancient cartonnage reveal technology and date papyrus

Papyrus was used as a writing support from 2500 BC to the 11 th century AD to record and transfer information both in Egypt (where it has predominantly survived) and the wider Mediterranean and Near Eastern world. During the Ptolemaic and early Roman periods in Egypt (c. 322 BC-14 AD), discarded written papyri were recycled and widely used as a support to make cartonnage, which was a glued encasement to protect mummiﬁed bodies. This research investigates the elemental characteristics of papyri known or believed to have come from cartonnage, in order to understand how that usage might have changed the elemental compositions of those papyri. In this research, the elemental compositions of ink, no-ink and “white” areas (where observed) were characterised using X-ray ﬂuorescence spectrometry in twenty papyri. Inked areas have signiﬁcantly greater concentrations of lead than areas of papyri with no ink, and X-ray diffractometry revealed that calcite (CaCO 3 ) formed the white compound of the preparation layer of the cartonnage. Changes in ink composition may have the potential to be used as a technique for dating papyri.


Introduction
Cartonnage is a protective casing which provides physical and magical protection for mummified bodies [1] .It offers protection from humans and animals, and the magic spells used on cartonnage decoration protect the deceased on their journey to the hereafter [2] .The first use of cartonnage, dating to the First Intermediate Period (ca 2160-2040 BC), is in the form of a mask that covered the head and chest of bandaged mummies, while in the Middle Kingdom (ca 2133-1786 BC), the cartonnage became one piece covering the whole body [3] .Subsequently, during the Late Period (664-332 BC) [4] and Ptolemaic Period (332-30 BC), the one-piece cartonnage was replaced with four to six separate sections; a mask covering the head and shoulders, a pectoral, an apron for the legs and a foot casing, and sometimes two pieces covering the rib cage and stomach [ 5 , 6 ].
Cartonnage is a composite material of linen, animal glue, Gum Arabic, calcite (CaCO 3 ), gypsum (CaSO 4 .2H 2 O), and organic and inorganic colourants [7][8][9] .The structure and composition of carton-nage varied throughout time, but was typically made from eight layers consisting (from the outer surface inwards) of polychrome paint, calcite, gypsum plaster, linen, a support layer made from thick unwoven fibres, linen, gypsum plaster and calcite [10] .Some cartonnage also had a layer of mud mixed with sawdust [11] ; a schematic sketch of cartonnage layers has been presented elsewhere [ 6 , 10 ].During the Graeco-Roman period (322 BC-313 AD), layers of glued papyri, with animal glue as the adhesive for these layers [ 12 , 13 ], came to be used to make cartonnage in place of linen.The source of the papyri was overwhelmingly documents discarded by the Ptolemaic administration [14] .Pieces of waste papyrus were cut and dampened with water to make them flexible to form the required shape of cartonnage, and layered plaster or glue were used as adhesives for those layers [13] .Cartonnage from the Ptolemaic and Roman periods represent a rich source of papyrus texts from these times [15] , yielding information about the governmental administration of Egypt [16] .Mummy cases made with cartonnage manufactured from papyri have been found in many sites in Egypt during excavations at Umm al-Baragat (ancient Tebtunis) on the southern edge of Fayum, Al-Hiba (ancient Ankyronpolis) in the southern part of the ancient Herakleopolite Nome, Bani Suwayf Governorate, and Abu Sir al-Malaq (ancient Busiris) [1] .
While the use of waste papyrus in mummy cartonnage is traditionally understood, based on the latest dated texts extracted from cartonnage, to have finished in the time of the Roman Emperor Augustus (30 BC-14 AD; [1] , a form of cartonnage was later employed to stiffen the covers of codices (the ancestor of the modern book).There has been limited analytical attention to the ways in which these book covers were produced, and it is unclear what sort of adhesive was used, or if substances such as calcite or gypsum were used.It has also been argued that these should not even be referred to as 'cartonnage', but rather as 'papyri from book covers' [1] .Further research is required on papyri both from book covers and from cartonnage.
Papyri used to make mummy cartonnage have significant value for papyrologists because of the variety of texts they contain [17] , so extracting papyri from cartonnage was a goal of many researchers.Papyrus can be extracted from cartonnage by dissolving the binders of the preparation layers above the papyrus support layers.Various materials were used to extract papyrus from cartonnage, including enzyme solutions [ 13 , 18 ], hydrochloric acid [18] , water at 60 • C [19] , and an aqueous solution of dipotassium hydrogen phosphate (K 2 HPO 4 ) and potassium dihydrogen phosphate (KH 2 PO 4 ) [20] .A solution of Pancreatic Trypsin NOVO, a commercial product that contains trypsin and chymotrypsin enzymes was used to soak cartonnage and extract the papyrus from the preparation layers and also to separate layers of papyrus [ 13 , 18 ].Microscopic examination of untreated papyrus, and papyrus treated by proteolytic enzymes, revealed no detrimental effects on the papyrus tissue after treatment [21] .Sigma protease P5005 type V (a commercial enzyme from Sigma-Aldrich) has also been used to remove papyrus from cartonnage by immersion of the cartonnage in the enzyme bath.Cartonnage painted surfaces were first faced with Fieux Japanese Contact Adhesive Tissue to protect the paint layer during papyrus extraction [22] .
It is unclear why papyrus was used in cartonnage manufacture; perhaps it was the cheap price of waste papyrus during Greek-Roman times [1] , the combination of flexibility and strength of papyrus that can readily form the required shape of cartonnage while giving support, or the religious value of papyrus plants.The papyrus fragments used to form cartonnage are therefore worthy of further study.

Research aim
This research addresses four questions; i. Do papyrus compositions vary between inked, uninked and white uninked surfaces?ii.Do elemental compositions of papyrus from cartonnage vary during three time periods?iii.Does papyrus from cartonnage have distinct elemental compositions from papyri not used in cartonnage?iv.Can the elemental compositions of undated papyrus fragments be used to allocate these fragments to known age groups?

Samples
The Macquarie University History Museum collection contains more than 700 papyrus fragments.Most come from cartonnage and date to within the last three centuries BC during the Ptolemaic Period [23] .Twenty papyri extracted from cartonnage belonging to this collection were investigated and analysed.The papyri have Greek texts written in black ink ( Table 1 ).The dates are largely derived from palaeography, that is, an assessment of the date of their handwriting based on comparable dated examples at the time they were accessioned into the Macquarie Papyrus collection.

Micro-XRF spectrometry
A Bruker M4 Tornado μ-XRF spectrometer was used for nondestructive elemental analyses of the papyri.Papyri were placed onto the measurement platform of the instrument prior to analyses.The X-ray tube had a rhodium anode and was operated at 50 kV and 40 mA, in air.A polycapillary optic focused a beam spot size 20 μm diameter, and a 30 mm 2 X'Flash detector was used with a dwell time of 60 s per point.Multi-point mode was used to measure 30 points from each papyrus; 15 points on the un-inked papyrus and 15 points on the ink.Ten measurements were also made of residues of white material where it existed ( Fig. 1 ).The 17 elements quantified were summed, normalized to 100 %, and the elemental data expressed as percentages.The bundled calibration of the instrument was used to calculate elemental concentrations.Means and standard deviations of the concentrations of aluminum (Al), silicon (Si), phosphorus (P), sulfur (S), chlorine (Cl), potassium (K), calcium (Ca), titanium (Ti), manganese (Mn), iron (Fe), cobalt (Co), nickel (Ni), copper (Cu), zinc (Zn), arsenic (As), strontium (Sr) and lead (Pb) are reported for ink, no-ink, and white regions in the twenty studied papyrus fragments.

Data analyses
Means and standard deviations of the elemental data were reported for the 20 fragments, for the three regions (ink, no-ink, and white).Elemental data were analysed to explore possible differences between ink, no-ink, and white regions in the tested papyri, and differences within these groups through different ages 199-100 BC, 299-200 BC, and 244-243 BC.One-way analysis of variance (ANOVA) was used, and the significance of the differences was tested at 95 % confidence.ANOVA was followed by a Bonferroni post hoc analysis test using IBM SPSS Statistics Release 28.0 for Windows.
Boxplots were also used to compare the concentrations of calcium and sulfur of the 20 papyri for ink, no-ink, and white spots from each papyrus.This was conducted to help identify if calcium and sulfur are characteristic elements for papyrus from cartonnage, and to check whether the white material might be calcite or gypsum.
Cluster analysis was used to create dendrograms; twodimensional figures used to visualize similarity between variables, whereby one axis represents the measured samples and the other axis is a similarity measure [ 24 , 25 ].Dendrograms have been employed to address questions related to provenance, age of creation, and technology according to the chemical compositions of archaeological materials including pottery and other ceramics [26][27][28] , coins [29] , and papyrus, including age estimation [ 30 , 31 ].

Results
ANOVA tests between inked, uninked and white uninked surfaces revealed significant differences in Si, P, S, Cl, K, Ca, Ti, Mn, Fe, Ni, Zn, As, Sr, Pb (P value < 0.05), while Al, Co, and Cu did not show significant differences at the 5 % level ( Table 2 ).To identify which groups were different from other groups, Bonferroni post hoc tests were conducted (Supplementary Table S1).Ink and noink groups were significantly different in Ca, Ti, As and Pb (Supplementary Table S1); the ink had greater concentration of Pb and Ti, and lower concentrations of Ca than adjacent non-inked areas.Ink and white surfaces are significantly different in Si, P, S, Cl, K, Ca, Ti, V, Mn, Fe, Ni, Zn, As, Br, Sr, Pb, while Al, Co, and Cu are not significantly different (Supplementary Table S1).Si, P, S, Cl, K, Ti, Mn, Fe, Ni, As, Sr, Pb had lower concentrations in the white surfaces than the inked surfaces, however the white surfaces had significantly higher concentrations of Ca.The no-ink and white groups are significantly different in Si, P, S, Cl, K, Ca, Ti, Mn, Fe, Ni, Zn, As, Sr, while Al, Co, Cu, As, and Pb are not significantly different (Supplementary Table S1).Ca concentrations in the white surfaces are higher than those in no-ink areas.
To check for differences in the elemental compositions of papyrus from cartonnage during three selected time periods (299-200 BC, 244-243 BC, and 199-100 BC), the three different regions in papyrus surfaces were tested through the three age groups.ANOVA of inked areas between 299-200 BC, 244-243 BC, and 199-100 BC results showed that the age groups are significantly different in Al, Si, S, Cl, K, Ti, Mn, Fe, Ni, Cu, As, Sr, Pb, while P, Ca, Co, and Zn are not significantly different ( Table 2 ).Elemental concentrations were larger for Al, As, Cu, K and Cl in ink from 199-100 BC compared with the other age groups.The concentrations of Co, P, Ca and Zn were similar through the tested groups.
Elemental compositions of white material from 199-100 BC are significantly different from white papyri surfaces from 299-200 BC and 244-243 BC.Elemental compositions of the white papyrus surfaces were not significantly different in Si and Co, while the remaining 15 elements were significantly different through the three age groups ( Table 2 ).The presence of Fe, Mn, and Ti increased in white surfaces dating from 299-200 BC.
ANOVA results show that one of the no-ink areas on the papyri from the three age groups was significantly different for Al, Si, P, S, Cl, K, Ca, Ti, Mn, Fe, Ni, Cu, Zn, As, Sr, Pb, with no significant differences between any of the age groups for Co or Fe ( Table 2 ).Significantly larger concentrations of Cl and K are found on non-inked regions on papyrus from the youngest age group.The concentration of Fe is high in inked and non-inked areas through the three age groups, and also is higher than that in white material.
Characterizing the elemental compositions of papyrus from cartonnage included in Supplementary Tables S2, S3 and S4 showed high concentrations of calcium in the studied papyri, however, these high contents of calcium were not found in papyrus not used in making cartonnage [32][33][34] .Boxplots were used to visualize and compare the degree of spread of values of Ca in the 20 papyri in ink, no-ink, and white spots for each papyrus.Comparing the concentrations of Ca between ink, no ink, and white ( Fig. 3 ) revealed a high concentration of Ca in all samples studied.The main compound of the preparation layer of cartonnage might have been calcite or gypsum.
XRD was used to identify the mineralogical compositions of the white material of six samples that have remains of white material at the papyrus surfaces ( Fig. 4 a).Calcium carbonate was identified in all tested samples, the characteristic peaks of the mineral calcite (CaCO 3 ) are detected in samples 541 and 393 ( Fig. 4 b and c) and for samples 449, 390, 378 and 520 (Supplementary Figure S2); as seven sharp reflections can be seen clearly.Peaks associated with cellulose were also observed.The presence of calcium oxalate dihydrate (weddellite; CaC 2 O 4 •2H 2 O) in some samples is consistent with findings by Autran et al. [35] .
To test the hypothesis that changes in ink composition may be used to constrain the age of papyri, cluster analysis was employed.The tested samples include two papyrus fragments, inventory numbers 449 and 596/1, which have no date listed in the museum catalogue.Cluster analysis was used to indicate the age of these papyrus samples through identifying which groups the undated papyrus fragments likely to belong to.The data of the 17 measured elements (Al, Si, P, S, Cl, K, Ca, Ti, Mn, Fe, Co, Ni, Cu, Zn, As, Sr, Pb) of the tested ink were used to draw a dendrogram ( Fig. 5 ).According to the dendrogram, the vertical line at distance 20 forms two groups.The first cluster has samples 390, 476, 397, 398, 449, 393, 392, 394, 439, 399, 391, 395, 378, 519, 384, while the second cluster samples are 520, 518, 529, 596, 541.Papyrus 449 is between 398 and 393 while papyrus 596 is surrounded by 529 and 541.

Discussion
Four questions were addressed in this research.The first regarded whether or not papyrus compositions vary between inked, uninked and white uninked surfaces.ANOVA discriminates differences in the elemental compositions between ink, no-ink and white regions on the studied samples as visible on Table .2 .Ca, Ti, As, and Pb show significant differences in their concentrations between ink and no-ink regions in tested samples.The elements Ti, As, and Pb are in the measurements of the inked areas.The difference in Pb appears only when inked areas are compared with no-ink and white.However, comparing no-ink with white did not show significant difference in Pb (p = 1.0 0 0) > 0.05, which indicates that lead (Pb) is present in the ink's chemical composition.
The concentration of Pb in ink agrees with the results obtained previously [36] , where Pb was probably added deliberately possibly in the form of powdered galena (PbS) in preparing the black ink, as it was found in the elemental analysis of papyrus dating back to New Kingdom, ca 1350 BC [37] .Alternatively it could have been a drying agent [38] , or have arisen from accidental contamination of the water used in preparing the ink [34] or from the container of storing the ink [39] .The existence of Pb in blank regions of the papyrus is probably due to contamination of the water used in making papyrus sheets [40] .The tested samples have abundant Fe; however the source of the Fe is not clear as it appears in ink, no-ink and white groups.Identifying the source of the iron needs further study to know if it was used to prepare the ink or preparing the white layer in cartonnage or even as a result of contamination from the surrounding storage of papyrus at the museum.
The second question addressed whether the elemental compositions of papyrus from cartonnage vary during three time periods.A comparison of inks between the three age groups shows significant differences in the elemental composition of most of the identified elements.A difference in elemental composition between the age groups of papyri is also clear in the no-ink regions ( Table 2 and Fig. 2 ), and also observed when comparing white material from 199-10 0, 299-20 0, and 244-243 BC ( Fig. 2 ).The high concentration of K on ink and no ink surfaces of papyrus samples from 199-100 BC are probably due to using Arabic gum [ 41 ] as a binder in preparing the ink during this time.Differences in the chemical composition of the inks and white material could be indicative of different local recipes for the production of these materials, while differences in the elemental composition of the 'no ink' sections during the three age groups is suggestive of different materials and additives used in making papyrus sheets [ 42 ].Many additives could be involved in preparing papyrus sheets for writing, including natron salt, milk, clay, and starch that can cause changes in the chemical composition of no ink areas depending on presence or absence of these materials.Overall, the change in the chemical composition in the three age groups probably is due to changes in manufacturing techniques, raw materials or environmental factors and could be deliberate.
The third question considered whether or not papyri extracted from cartonnage have distinct elemental compositions compared to papyri not used in cartonnage.Identifying differences between papyri from cartonnage and those are not can be based mainly on the Ca concentration.Fig. 3 shows the high content of Ca in papyrus extracted from cartonnage, while this concentration cannot be found in other studies conducted on papyrus samples not from cartonnage [32][33][34] .Calcium, used in the preparation layer of cartonnage, remains despite papyrus conservation treatment and cleaning.Outliers in the data are due to the very high concentra- tions of calcium on these points.The preparation layer of cartonnage might typically contain calcium carbonate in the form of the mineral calcite (CaCO 3 ), or calcium sulfate in the form of gypsum (CaSO 4 .2H 2 O).To determine the mineralogy of the white material, white regions on the papyri were analysed using X-ray diffractometry ( Fig. 4 ).The diffractograms revealed that the preparation layer of cartonnage consists of calcite, rather than gypsum.Cellulose is the main mineral in the papyrus support.The calcium oxalate mineral weddellite (CaC 2 O 4 •2H 2 O), tentatively identified in samples 393 ( Fig. 4 c), 390, 449, and 520 (Fig. S1), could have formed in the parenchyma cells of the papyrus plant [17] , however in some cases these cells do not contain calcium oxalate.The presence of calcium oxalate was attributed to the biodegradation of carbonates by fungi [ 43 , 44 ], but no stains were observed in the tested samples to confirm this suggestion.
The fourth question addressed whether the elemental compositions of the ink of undated papyrus fragments can be used to suggest these fragments might belong to known dated age groups.The way papyri and their inks are made can affect their chemical compositions.Comparing compositions of unknown papyri and their inks with those of known provenance or age may help understand their origins.A further advantage is that the analytical meth- ods used here are non-destructive, in contrast with some chemical analysis methods, or for dating, radiocarbon analyses.
Cluster analysis was used to group the tested papyri according to the similarities in the chemical compositions of their ink.The twenty papyri were classified into two groups, one of which contains the samples with the date of 199-100 BC and the other contains papyri dated to 299-200 and 244-243 BC ( Fig. 5 ).The similarity observed between the inks from papyri dated from 299-200 and 244-243 BC indicates that techniques for making ink did not change appreciably during the third century BC, though there are measurable differences when comparing this group to the 199-100 BC age group.
The two samples of unknown date, 449 and 596, have different places in the dendrogram far from each other, that relate to the differences in the chemical composition of their inks, and they consequently group into different age groups.The chemical composition of the ink of papyrus 449 is close to that of 398, which dates to 244-243 BC, and that of 393, which is dated to 299-200 BC.The distance linkage for the ink in papyrus 449 (unknown date) and ink of papyrus 398 (244 BC) is shorter than the distance linkage between 449 and 393, which may indicate that papyrus 449 dates to the mid-third century, as does papyrus 398, as a smaller distance linkage indicates a closer similarity in chemical composition compared to others which cluster later [ 45 ].Papyrus 596 meanwhile is close to 529 and 541, both of which have been dated to 199-100 BC, so 596 may also be assigned to this period.While further analyses are required to confirm the reliability of the technique, the data presented here suggests the hypothesis that changes in the chemical composition of ink may be used as a dating technique.

Conclusions
This research is dedicated to analysing papyrus fragments removed from cartonnage.Elemental analyses of twenty papyri belonging to Macquarie University History Museum, Australia revealed that calcium is the predominant element of those reported, in all tested papyrus fragments.The elemental compositions allow detection of compositional differences and similarities among the grouped papyri.Dendrograms can allow effective classification of the inks on papyri according to similarities of their chemical compositions.It is possible to determine whether or not the papyri have been extracted from cartonnage by the presence or absence of high concentrations of calcium on the papyrus, and by the presence of the mineral calcite.

Fig. 2 .
Fig. 2. Mean of the elemental concentrations for ink, no-ink and white regions of the studied papyri.Means were of 15 measurements of the tested areas in each papyrus sheet.

Fig. 3 .
Fig. 3. Boxplots of Ca concentrations of the papyri studied for (A) Ink.(B) Blank.(C) White regions from different age groups; each boxplot represents fifteen measurements.Papyri identifications refer to Macquarie University History Museum inventory numbers.

Fig. 4 .
Fig. 4. XRD patterns of white material for (a) Stack of six samples, (b) Papyrus number 541, and (c) Papyrus number 393 showing that the calcium carbonate mineral calcite formed the white compound of all tested samples.

Fig. 5 .
Fig. 5. Dendrogram for the ink of the papyri studied using Euclidean distances; the oval shapes refer to the two papyrus sheets of unknown age.The figure was obtained from 300 XRF measurements from 20 papyrus pieces on the studied papyri ink.

Table 1
Details of the 20 papyri selected for testing.
* Fayum is a city in Middle Egypt located 100 km southwest of Cairo.?refers to unknown date or place of origin.

Table 2
Analysis of variance of the elemental compositions of ink, no-ink, and white groups, and analysis of variance of the ink, no-ink, and white within three age groups (199-100, 299-200, and 244-243 BC).
X-ray diffractometry (XRD) on white parts of the papyri was conducted using a Panalytical Empyrean X-ray diffractometer with a Cu anode tube (wavelength ∼1.54 Å) operated at 45 kV and 40 mA.Diffractograms were collected from 5 to 60 degrees two theta.Phase identification was performed by using Panalytical's Highscore plus software and ICDD PDF 4/ICSD databases at the Solid State and Elemental Analysis Unit, University of New South Wales, Sydney.