Introduction

Unlike the well-studied western papers, Tibetan paper received less attention, although a number of significant Tibetan manuscript collections are held in heritage institutions in the UK (Dalton and Van Schaik 2006) and China (Li 2011). Understanding its material science is of great importance for better conservation and for understanding the Tibetan material culture. The two dominant designs of Tibetan manuscripts include the use of coloured inks against a blue or black background, and black ink on the natural paper surface without decoration (Fig. 1a) (Helman-Ważny 2014). Layers of indigo and carbon are commonly used to coat papers, while the inks include mineral pigments and carbonized plant material or soot mixed with glue (Helman-Ważny 2014, 2016; Luo et al. 2021b). The usual format is horizontally rectangular (‘pothi’ format), handwritten or printed. Diverse binding styles are used, ranging from loose-leaf books to more complex multi-section codices (Helman-Ważny 2014).

For the convenience of notation, in this study, all papers with Tibetan characters are denoted as ‘Tibetan paper’ in this study, meaning that the term denotes paper used to produce Tibetan books, as evidenced by the characters on the paper and its format. Although such paper might or might not have been produced using historically authentic Tibetan papermaking techniques or even produced in the Tibetan geographical region, it should still be seen as part of Tibetan cultural heritage. As an example, the paper produced in Europe, but finished, e.g. sized and polished in Islamic countries, as well as written with Islamic characters, has been denoted as Islamic paper in a recent study (Mahgoub et al. 2016).

The laminated structure is a special feature of Tibetan paper (Fig. 1b), where wheat flour is often employed for pasting two to eight or more layers together as well as for sizing (Helman-Ważny 2014). Sizing is a process to make the paper smooth, rigid and to prevent ink penetration. Sometimes, animal/yak glue or buttermilk was applied either by itself or added into the wheat flour glue (Helman-Ważny 2014).

Fig. 1
figure 1

a Two representative Tibetan paper samples from the historical material reference collection at UCL Institute for Sustainable Heritage. A manuscript written in silver ink on a black and blue background and a manuscript written in black ink on the untreated paper; b Laminated structure of a typical Tibetan paper in the collection

Full size image

Much recent research focused on early Tibetan books from an archaeological perspective, studying their provenance (Helman-Ważny 2014), raw materials (Helman-Ważny 2016), the papermaking process (Helman-Ważny and Van Schaik 2013), writing styles (Richardson 1985) and language (Beyer 1992), while scientific evidence on the material composition and durability of more contemporary papers used as writing support for Tibetan books (19th−20th century) is limited. In the studies by van Schaik (Van Schaik et al. 2015) and Helman-Wazny (Helman-Ważny 2014) on the manuscripts dated to the 9th century and originating from Dunhuang, China (Helman-Ważny and Van Schaik 2013; Helman-Ważny 2014, 2016), fibres from indigenous plants of the Thymelaeaceae family from the foot of the Himalayas, including Daphne/Edgeworthia species and Stellera chamaejasme, were identified, where root bast fibres were used for papermaking in the latter. Besides using fibres of local plants, the recently dated Tibetan manuscripts were found to be written on Russian papers and papers from other regions of China (Tsien 1973; Collings and Milner 1978). However, in-depth scientific surveys of paper properties, as available for western papers (Hunter 1978; Strlič et al. 2015), are still missing.

In terms of paper durability, it is generally accepted that acid-catalysed hydrolysis and metal-catalysed oxidation are the two primary ageing processes that lead to loss of mechanical strength of paper (Strlič and Kolar 2005). It has been reported that corrosive transition metal-based inks were rarely applied on Tibetan manuscripts (Helman-Ważny 2014; Van Schaik et al. 2015), indicating a low possibility of metal-mediated corrosion, unlike in iron gall ink-containing manuscripts (Liu et al. 2017). Since fitness-for-use is of great concern for both scientists and conservators, we investigated the durability of 19th− 20th -century papers used in Tibetan books specifically to explore whether its degradation behaviour differs from European paper, using accelerated thermal degradation as explored previously for other types of papers (Zou et al. 1996a; Strlič et al. 2015; Liu et al. 2017).

In this study, ‘Tibetan’ paper properties were explored using both conventional destructive methods to explore paper properties, and modified methods based on the available colorimetric and spectroscopic methods to determine the contents of glues. We explored fibre morphology with optical microscopy, the laminated structure and traces of polishing using scanning electron microscopy (SEM), as well as determined pH and viscometric DP (Zou et al. 1996a; Strlič and Kolar 2005). Measurements of mechanical properties were discarded as they require large sample quantities, which could not be obtained, and are additionally characterised by significant measurement uncertainties (Gurnagul et al. 1993; Havlínová et al. 2009). Starch and protein contents were quantified as they were widely used as glues for pasting layers together. To explore the durability of the studied papers, we tested the applicability of the Ekenstam equation (Ekenstam 1936) and of the dose-response function, which models the degradation rate of paper with its pH and environmental variables (Strlič et al. 2015). This has already been proven to be applicable to various western papers as well as Xuan paper (Luo et al. 2021a).

This research presents a survey of material properties and investigates the durability of 19th−20th-century papers found in Tibetan books, with the aim of expanding our understanding of the Tibetan material culture, provide the evidence for conservation management for heritage institutions holding Tibetan book collections.

Methodology

Samples

The reference material collection of ‘Tibetan’ paper at UCL Institute for Sustainable Heritage (n = 92) was used for the study. The collection was purchased from antique stores, and unfortunately, there is no associated data about its provenance or previous storage environments, except that they were from the 19th−20th century. All the papers used as writing supports for Tibetan books were visually examined and documented before sampling and analysis. Paper areas with ink, mould, visible stains, as well as those within a 1-cm margin were not sampled to decrease the uncertainties. In this study, TP numbers represent individual samples from the reference collection.

Lignin spot test

The phloroglucinol spot test (TAPPI T 401 om-15 2015) was used, i.e. a saturated solution of phloroglucinol in 20% HCl. One drop of the phloroglucinol solution was dropped on three spots of each paper sheet. The presence of lignin resulted in reddish colouration, and no colour change indicated lignin-free paper. Colour change was assessed on a 3-point scale: (i) no colour change, (ii) < 1/4 of fibres showing colour change, and (iii) > 1/4 of fibres exhibiting colour change.

This preliminary test is useful to decide what samples can be used for viscometric determination of DP, as papers with high lignin content cannot be dissolved in the cupri-ethylenediamine (CED) solution.

Fibre furnish

Fibre furnish analysis using Herzberg stain test (ISO 9184-3 1990) was conducted to observe fibre morphology, and qualitative differentiation of raw material into chemical, mechanical and rag pulps. Given the time-consuming process, 10 papers were analysed in detail out of our paper sample set (n = 92), the selection covering various pH, DP, and optical features, which were then used for subsequent degradation experiments. The samples were examined with optical microscopy (Brunel Microscopes Ltd, Wiltshire) equipped with a DSLR camera (Canon EOS 1100D, Tokyo) using transmitted light.

pH

The standard cold extraction procedure (TAPPI T 509 om-11 2011) was applied to measure pH with a modification to allow for reduced sample quantity. 1.0 ± 0.1 mg sample was extracted in 100 µL of deionised water (Millipore) and soaked overnight. A Mettler Toledo SevenGo pro™ pH/Ion Meter (Columbus) with a micro-combined glass electrode (Mettler Toledo Inlab® Micro 51,343,160) were used. Triplicate measurements were carried out to evaluate pH homogeneity for the sample set, with measurement the uncertainty of 0.1–0.8 pH units. In addition, pH was also determined layer-by-layer for 8 multi-layered Tibetan paper sheets to check for pH differences between or among the layers.

Water and ash contents

Water content (%m/m) in the samples were determined by drying for 3 h at 105 °C. Following the standard (ISO 2144 2019), after ignition at 550 °C for 3 h, the residue weight represents the ash content (%). Triplicate determinations were conducted, with average contents of water and ash for the paper set (n = 92) at 5.0 ± 0.7% and 8.1 ± 8.1%, respectively.

Starch content

Starch contents in the samples were determined following the total starch assay procedure (Megazyme 2019), with modifications to reduce sample size. The paper samples were cut into tiny pieces and dried in a muffle furnace at 105 oC for 3 h. Approx. 10 mg dry paper sample was weighed, and 2.55 mL working reagent was used, including 2.5 mL 100 mM sodium acetate buffer (pH 5), 25 µL of thermostable α-amylase and 25 µL amyl-glucosidases (AMG), after which 0.1 mL aliquot of the supernatant was used. The experimental temperature and time for enzymatic actions were followed by the standard procedure, as described in the assay (Megazyme 2019). The absorbance was measured against the reagent blank at 510 nm using the Agilent 8453 spectrophotometer (California, United States) after adding 3 mL glucose oxidase, where quinoneimine dye was generated. Starch content was determined as weight% of dry paper following the equation:

$$ {\text{W}}\left( {{\text{starch}}} \right){\text{ }}\left[ \% \right]{\text{ }} = {\text{ }}\Delta {\text{A }}*{\text{ F }}*{\text{ EV}}/0.{\text{1 }}*{\text{ D }}*{\text{1}}/{\text{1}}000{\text{ }}*{\text{ 1}}00/{\text{W }}*{\text{162}}/{\text{18}}0 $$

where ΔA is the absorbance at 510 nm of a sample solution minus the sample blank, F is the factor, calculated by dividing the amount of D-glucose analysed (100 µg ) by the absorbance obtained for this amount of D-glucose in the standard assay, EV is the extracted volume of the solution (2.55 mL), 0.1 means the volume of sample analysed, 1/1000 means conversion from µg to mg, 100/W is conversion to 100 mg sample, 162/180 means the factor to convert from free glucose, as determined, to anhydroglucose, D means further dilution of the incubation mixture (if performed).

Protein content

The Pierce™ Bicinchoninic acid (BCA) Protein Assay Kit, purchased from Thermo Scientific (Massachusetts, United States) was used to determine protein contents in paper samples, before which samples were pre-processed: dried 15–20 mg sample was weighed in duplicate, with 0.5 mL deionised water added. The mixture was mixed using vortex mixer (IKA TTS 2, Staufen, Germany) for 10 s and then placed in ultrasonic bath for 30 min, and vortexed. 0.1 mL of the supernatants were used for analysis following the assay (Scientific 2007).

Degree of polymerisation (DP)

On the basis of the standard ISO 5351 2010, ~ 20 mg paper was dissolved in 10 mL deionised MilliQ water mixed with 10 mL cupri-ethylenediamine solution (1 mol L/−1, Merck). The intrinsic viscosity of samples was determined using a modified viscometer, and at least three repeated measurements were performed to gain average efflux times. It is known that intrinsic inhomogeneity of historic papers leads to higher measurement uncertainty (Mahgoub et al. 2016; Brown et al. 2017; Liu et al. 2017); therefore, triplicate determinations were conducted on 51 papers, which are lignin-free or contain < 1/4 lignin, to assess paper homogeneity with measurement uncertainties determined to be 1–15%. Measurements were also performed for each layer at the same position for 8 laminated Tibetan papers to evaluate the DP differences.

The limiting viscosity number was calculated using Wetzel-Elliot-Martin’s equation, considering water, ash, starch, and protein contents in the paper samples at room conditions. The DP was then calculated using the Mark-Houwink-Sakurada equation: DP0.85 = 1.1[η] (Evans and Wallis 1987). The uncertainty of DP determination for Whatman No. 1 filter paper (Maidstone) was confirmed to be less than 1% (n = 5) before measuring the DP for our studied paper samples.

Scanning electron microscopy (SEM) with energy dispersive X-ray analysis (EDX)

Three polished paper samples were visually assessed to display traces of polishing and high surface gloss. Using a scanning electron microscope (Hitachi TM3030, Tokyo) in charge-up reduction mode, images of three layers of a polished sample (TP173) were observed.

For elemental analysis, a Zeiss Ultra Plus microscope equipped with EDX (Oxford X-Max SDD 50 mm2 detector and INCA 4.14 X-ray microanalysis software). Samples’ preparation included cross-Sect. polishing to obtain pristine cross sections of the selected papers (using Jeol IB-19510CP Cross Section Polisher), fixation of the selected paper samples on conductive C-tape and subsequent sputtering with Au/Pd without any additional polishing. The EDX detector was calibrated just before the analysis with Si-standard at the operating conditions. The EDX spectra were recorded on flat regions of the samples using process time 5 s, lifetime 120 s and 15 kV accelerating voltage, which is an acceptable compromise between the analysed volume and the overvoltage needed for excitation to produce X-rays. Using the Anderson–Halser (Friel and Lyman 2006) estimation the X-ray production depth was calculated to be approximately 4.9 μm.

Thermal degradation

Thermal degradation experiments were conducted using a Vötsch Climate chamber (VC0018, Balingen-Frommern) for 9 papers at 80 ˚C and 65% RH for 10, 15, 20, 40 days and for another set of 7 papers at 60 ˚C and 80% RH for 4, 8, and 12 weeks, where the selected samples have a series of initial pH and DP. The degraded samples were conditioned at room conditions (22 ˚C, 50% RH) in in the darkness for 1 day before any measurement and no further pre-treatment was carried out.

Results and discussion

Paper properties

Descriptive information

Fibre furnish

The choice of fibres is the fingerprint of paper-based materials. As indicated in the study of Helman-Wazny (2014), local plants from the Thymelaeaceae family are unique raw materials for Tibetan books, as the plants can survive in harsh high-altitude environments. Papers made of these plants are thought to be more stable and durable than those from other types of fibres, and resistant to biological damage due to plant toxicity (Helman-Ważny 2014, 2016); however, none of the examined 19th–20th-century papers contained fibres from these plants.

Fibres are dyed with the Herzberg reagent to induce colouration specific to fibre categories and papermaking processes (Helman-Ważny and Van Schaik 2013; Helman-Ważny 2016). Based on the basic fibre morphologies, fibre ends and the associated parenchyma cells, fibre identification was carried out for the 10 samples to be used for degradation experiments. Fibres of paper mulberry, wood pulp, hemp, ramie, straw and bamboo were identified, which is consistent with the findings that Tibetan manuscripts and books could be executed on papers imported from nearby papermaking regions, e.g. traditional Chinese fibres including ramie, paper mulberry, hemp, bamboo, rice straw etc. (Hunter 1978) or Asian fibres including flax, hemp, wood pulp etc. (Helman-Ważny 2014), and that did not necessarily come from the Tibetan region. It is worth noting that some of these plants can also grow in the Tibetan region at lower altitudes especially in southern Tibet (Helman-Ważny 2014; Brown et al. 2017). Therefore, only papers made from Thymelaeaceae-family plant fibres can be confirmed to be produced locally in Tibet (Helman-Ważny 2014), while manuscripts executed on papers containing more common fibre types probably come either from low-altitude Tibetan regions, other raw materials and modern paper available in Tibetan regions, or recycled fibres of local papers. Finally, finished papers could have been made to Tibetan requirements and transported from other regions to Tibet.

Polishing

Surface polishing is to make paper surface smooth, even and more impervious to ink and paint penetration. While polishing is hardly ever applied in Chinese papermaking (Brown et al. 2017), we found three examples in our reference collection (n = 92) with obvious traces and marks of polishing on the recto and verso (Helman-Ważny and Van Schaik 2013; Helman-Ważny 2014). In Fig. 2, SEM images of the three-layer TP173 show that the recto and verso surfaces are flat and compressed, while the middle layer is not, indicating that the outer surfaces were polished after sizing. It should be noted that during separation of layers, some fibres in the middle paper layer were unstuck, making the surface more uneven.

Fig. 2
figure 2

SEM images of three layers of TP173: a upper layer; b middle layer; c bottom layer

Full size image

Layered paper structure

It has been reported that in traditional Tibetan papermaking, a few layers of paper were glued together and the surface was finished with a layer of plaster (gypsum or lime) or chalk powder, indicating that calcium compounds can be present on the top and bottom layer of Tibetan paper (Helman-Ważny 2014). To observe the layered structure in our paper samples, cross-sectional micrographs of two samples (TP209 and TP211) were taken, as shown in Fig. 3. As evident from the SEM-EDX mapping, Ca was detected in each layer, with no Ca-rich surface layer, indicating that all layers have been treated with calcium compounds. Unfortunately, the layers are not easily distinguishable in the images as evidently the glue penetrates the layers very well.

Fig. 3
figure 3

SEM-EDX images of cross-sections and main elemental maps for two laminated Tibetan manuscripts: a TP209 and b TP211. In the mapping images, higher concentrations are indicated by more intense colours, and lower concentrations are indicated by less intense colouration or darker regions

Full size image

Alkaline reserve in the two papers is in the form of Ca and Mg compounds, as shown in Fig. 3, suggesting that acid-catalysed hydrolysis can be mitigated. No appreciable amounts of transition metals were found in the paper materials, like Fe and Cu, where Fe was only found in ink/pigments of TP209, indicating low possibility of metal-catalysed oxidation. Both available alkaline reserve and low amount of transition metals would ensure stability of paper.

In order to be able to treat the laminated samples as unique samples without separation into layers, we had to determine the overall measurement uncertainty for the properties averaged across the layers with those of the properties measured per layer. It was found that 70 paper samples are laminated, however, in only 8 papers the layers could be separated to allow for separate analyses. On visual observation the individual layers could not be distinguished. We therefore determined the thickness, pH and DP for each layer of the 8 multi-layered Tibetan papers to determine the uncertainties, and the results are presented in Table 1. The corresponding coefficients of variation between and within layers for thickness, pH and DP are low at < 8%, < 1% and < 5% (n = 8) respectively, indicating that the differences in thickness and chemical properties are lower in comparison to the differences observed from one paper to another. For example, the overall uncertainties of determination of pH and DP due to material inhomogeneity are in the ranges of 1–12% and 1–15% respectively, which is consistent with the uncertainties as determined for other types of historic paper samples, such as Chinese and Islamic paper (Mahgoub et al. 2016; Brown et al. 2017). Therefore, we decided to treat the composition of a sample as homogenous for the purpose of the analyses to follow.

Table 1 Eight laminated samples used for validation of layer similarity within the corresponding paper sheet, where thickness (mm), pH and DP values were determined, and corresponding uncertainties are presented
Full size table

Chemical properties

Once the uncertainty associated with material inhomogeneity was determined and it was established that averaging of a layered structure is justified, further analyses were performed to enable the determination of DP. The lignin spot test was conducted to establish the suitability of samples for viscometric DP determination. The amount of lignin in paper depends on the raw material used in the papermaking process, and on the process of lignin removal during pulping (Hunter 1978). Lignin can affect paper stability as its oxidation leads to the production of acidic compounds, which could promote cellulose hydrolysis (Schmidt et al. 1995; Barclay et al. 1997). Using the phloroglucinol test, 40 out of 92 papers were established not to contain lignin, i.e. no coloration was present. Further 11 papers contained small amount of lignin, meaning that < 1/4 of a sample exhibited some colour change. Only these 51 paper samples were used in the process of viscometric DP determination. The calculation of DP was corrected for water, ash, protein and starch content.

Figure 4 presents the frequency histograms with data on starch, protein, water, and ash contents, as well as pH and DP for the studied ‘Tibetan paper’ sample set. As discussed in the introduction, both starch and animal glue can be found as binding or sizing media in Tibetan paper (Helman-Ważny 2014), which is consistent with the findings of our study. A significant variation in starch content is observed, ranging from 0 to 31.8%, indicating that it is used as the main glue for lamination of paper sheets. Interestingly, proteins are also present in all samples (> 1%) except one (TP186, 0.01% protein content), which verifies the assumption that animal glue has been used for pasting/sizing as well. It has been reported that proteinaceous materials can cause paper discoloration and makes them more susceptible to insect attack (Helman-Ważny and Van Schaik 2013; Helman-Ważny 2014).

The average water content of was 5.0%, which is related to the environmental conditions during storage (May and Jones 2006). Only one paper (TP186, machine-made) was found to be hydrophobic, containing ~ 1% water content. The high content of ash must be related to the Ca, Mg and Si compounds used in the papermaking process, as evident from the elemental maps for the two papers examined in Fig. 3.

Fig. 4
figure 4

Frequency plots presenting distributions of pH (n = 92), corrected DP (n = 51), water (n = 92), ash (n = 92), starch (n = 92), protein (n = 92) contents for the ‘Tibetan’ paper set, in which DP was calculated using sample weights with deduction of water, ash, starch and protein contents

Full size image

It is widely acknowledged that acidity is one of the most important factors defining the durability of paper-based materials (Zou et al. 1994, 1996a) and that DP serves as an indicator of fitness-for-use for cellulosic materials (Zou et al. 1996a). In general, the paper samples examined here could be considered as stable, given that the average pH is 7.0 and DP 1820 (after correction of sample weight). The distribution of pH does not exhibit the usual binary distribution typical of western collections representing alkaline and acidic papermaking practices (Strlič et al. 2020). It could be hypothesized that this represents a steady trend towards acidification that could be a consequence of the environment in which the documents were used. Many documents namely exhibited heavy soot staining. Using the available dose-response function for western paper (Strlič et al. 2015), a paper with pH 7 and the starting DP of ~ 2000 could take more than 1200 years to degrade to DP 300 (i.e. the threshold at which paper loses its fitness-for-use) at room conditions (20 °C, RH 50%).

Comparing the properties of paper used in Tibetan books with mainland Chinese paper (Brown et al. 2017), Islamic paper (Mahgoub et al. 2016) and European paper (Stephens et al. 2008) from the same period of time, most of Islamic paper presents pH and DP in ranges of 5–7.5 and 500–1750, and European paper presents pH and DP at ~ 7 and > 2000. Paper used in Tibetan books tends to have similar chemical properties to modern mainland Chinese paper, given that the majority of both types of papers present pH and DP in the ranges of 6.5–8 and 1000–2500, respectively (Brown et al. 2017). These properties can be affected by raw materials, papermaking process and precious storage conditions (Strlič and Kolar 2005). This is also in line with the results of fibre furnish discussed in Sect. 3.1.1 and the archaeological research by Helman-Ważny (2014), indicating that Tibetan books and manuscripts could be executed on typical Chinese paper consisting of ramie and bamboo fibres.

Degradation of 19th–20th-century papers used in tibetan books

The degradation behaviour in terms of kinetics and mechanism could help to set preventive conservation strategies for libraries and archives. We selected 16 samples as representative of the studied sample set to perform degradation experiments, based on layer features, as well as range of pH and DP (Table 2).

Table 2 Properties of the selected 16 papers (TP) for accelerated degradation experiments. The asterisk denoted the sample where < 1/4 of the fibres exhibit a colour change according to the phloroglucinol spot test. TP186 is the only machine-made paper, and the others are all handmade, as assessed visually. The text in TP184 and TP186 is printed, the rest are manuscripts. The DP was corrected using contents of ash, water, starch, and protein
Full size table

On the basis of the Ekenstam equation (Zou et al. 1996a; Strlič et al. 2015), the linearity of change of 1/DP over time was established for western papers regardless of whether the mode of degradation was oxidation or acid-catalysed hydrolysis (Liu et al. 2017). The applicability of this relationship to ‘Tibetan’ papers made of divers fibres using special paper preparation techniques and high amounts of starch (Helman-Ważny and Van Schaik 2013; Helman-Ważny 2014, 2016) had to be established. The 14 samples were therefore degraded to various extents at two sets of conditions: 80 °C and 65% RH and 60 °C and 80% RH, the results being shown in Table 3. Good linear regressions of Δ(1/DP) against time were established, which allows us to calculate the rate constants (k) for the degradation process, which were then used for comparison with the values modelled by the dose-response function. Due to the limited sample availability, the DP of TP190 and TP200 was determined only after degradation for 10 and 20 days respectively under 80 °C and 65% RH.

Table 3 The rate constants (k) for the accelerated degradation of the studied samples (at 80 °C, 65% RH and 60 °C, 80% RH) and R2 values for the linear fits of the Ekenstam equation, for 14 ‘Tibetan’ papers. Due to the restricted sample quantity, k of TP190 and TP200 was calculated using the Ekenstam on the basis of two points only (before and after degradation at 80 °C, 65% RH), which is 2.8537 × 10−2 and 3.7652 × 10−2k (year −1). SE stands for standard error
Full size table

To establish whether the degradation of 19th−20th -century paper found in Tibetan books follows the same principles as western paper, the available dose-response function was applied to predict the rates of degradation, and compare the predictions with the observed rates. As shown in Fig. 5, both modelled and actual k increase with the temperature under degradation, indicating the devastating role of high temperature to paper materials. Also, in case of agreement between the modelled and actual k, and in case of overlap with the previously observed data for western paper (Strlič et al. 2015), we could establish that despite the layered structure and high content of starch, ‘Tibetan’ papers follow the same principles of degradation as western paper. It should be noted that k was calculated using the corrected DP data, i.e. contents of water, ash, starch and protein were excluded from the calculation of cellulose concentration. Such an agreement was in fact observed, as presented in Fig. 5. This proves that rates of degradation of the complex ‘Tibetan’ paper principally depends on pH, T and RH of the storage environment. This may not be particularly surprising given that the function was found to describe well the ageing behaviour of materials as different as painting canvases (Oriola et al. 2015) and recycled paper (Coppola et al. 2018).

This study demonstrates that the two models, i.e. the Ekenstam equation and the dose-response function, are applicable to the degradation of 19th− 20th -century Tibetan manuscripts. Based on this, prediction of lifetimes for entire collections can be carried out, offering curators and collection managers the data to support sustainable preservation. Some caution is advised given that our study included only two sets of environmental conditions and 16 paper samples; however, this was the consequence of limited availability of representative sacrificial historic material. Further studies would be needed to establish the applicability of the dose-response function to the highly valuable archaeological collections; however, these would need to be based on model papers.

Fig. 5
figure 5

Comparison of the modelled degradation rates and the actual degradation rates for 16 paper samples (solid circles and squares) with the historic western papers as reported in the literature (hollow circles). The modelled rates were calculated using the dose-response function (Zou et al. 1996a, b; Sedlbauer 2002; Strlič and Kolar 2005)

Full size image

Conclusion

Systematic characterisation and accelerated degradation of a sacrificial 19th− 20th century papers of Tibetan manuscript collection was carried out to explore material properties and durability using conventional analytical methods (e.g. determination of pH, DP, water and ash content), modified methods (determination of starch and protein contents), in conjunction with SEM-EDS and accelerated degradation. The following conclusions can be made:

  • Fibre analysis shows that 19th− 20th century papers in Tibetan books was made from a wide range of diverse and mixed fibre types, such as straw, ramie, bamboo, mulberry, hemp, and softwood or hardwood pulp, which is similar to mainland Chinese paper of this period. This may suggest that these Tibetan books were written or printed on paper produced outside of Tibet, or otherwise that traditional Tibetan craft has been replaced with modern industry. Layering of paper sheets was observed in most samples, while polishing was only occasionally applied in our samples. This suggests that even though paper used for Tibetan books may not necessarily follow Tibetan papermaking tradition, thought its processing involved the traditional Tibetan book-making techniques.

  • For the first time, starch content was successfully measured in paper materials based on a method modified from food industry. Both starch and animal glue have been used to glue and size paper sheets, with starch being the main component. The thin layers within one sheet appear to be the same as evidenced by thickness, pH and DP measurements.

  • The majority of studied paper samples are nearly neutral with pH ranging between 6 and 8, with 11% of the sample set presenting pH less than 6. The cellulose degree of polymerisation is generally good, with the average DP being 1820. Such paper is estimated to degrade at similar rates as contemporary mainland Chinese paper. Together with the fibre furnish results, the papers in loose-leaf Tibetan books used in this study might be made using typical mainland Chinese fibres, or other modern paper available in Tibetan regions, or papers made from plants in Tibetan regions.

  • The Ekenstam equation has been verified to be applicable to 19th− 20th century paper samples used inTibetan books. The rate constants of degradation could be calculated for two sets of conditions and compared to the predicted rate constants estimated using the dose-response functions available for historical western paper. It was found that the studied papers follows the same pathways of degradation as western paper, i.e. the rate constants of degradation depend on paper pH, temperature, and relative humidity of the environment.

The life expectancy of heritage collections is of great concern to curators, collection managers, material scientists and archaeologists. Having systematically studied material properties and the degradation behaviour of 19th− 20th century papers in Tibetan books, preventive conservation guidelines can be developed to support the practice of collection management. However, for such guidelines to be applicable to older Tibetan paper, which is often represented in valuable heritage and archaeological collections, a comparison between model older Tibetan paper and model paper representing 19th−20th century papers used in Tibetan books would be necessary. This would confirm, or not, the general usefulness of the otherwise widely applicable dose-response function for historic paper, even to the most complex paper types as represented in Tibetan paper collections.