Monitoring of damage in historic tapestries

In contrast to the use of surface techniques such as SEM, SIMS and XPS, ICP-MS and ICP-OES allow the identification of metal ions in the bulk of an historical sample. However, the more-sensitive technique (ICP-MS) showed prohibitively large interference of ArO+, which meant that quantitative data for Fe were difficult to obtain. This was not a problem with ICP-OES, but, using current state-of-the-art instrumentation, quantities of sample required for analysis are prohibitively large for application to historical artefacts. These techniques are very sensitive to the presence of contaminants, such as those arising from dirt and dust particles, so that the detection of a given metal ion does not necessarily mean that it was used as a mordant. Using these techniques we have shown: - the presence of Al, Fe, Cu, Cr, Pb in model and/or historical samples (WP5); - a concentrating effect of Fe3+ if this species is present as a contaminant in alum when used as a mordant; these results support the historical records which specify "pure" sources of alum for optimum dyeing results (minimising the dulling effects of iron contaminants) (WP2/5)

Various corrosion morphologies have been observed on metal threads. The corrosion products were found to be mainly sulphur compounds with only relatively few and patchy chlorine containing products. The copper content in the corrosion products and in the surface regions of the metal filaments were always relatively higher than that measured in the bulk material, indicating copper ion surface migration. X-ray Photoelectron Spectroscopy (XPS) analysis of corroded metal threads showed the presence of chlorine and sulfide and to a lower extent sulfate. X-ray Diffraction (XRD) was employed for the first time to analyse the corrosion products on historic metal threads. Various corrosion crystals were identified, including digenite, chlorargyrite and acanthite. Accelerated tarnishing studies of copper and silver foils, based on the thermal and photo-degradation of associated silk and wool, produced a variety of corrosion morphologies. Degradation of wool was found to cause higher levels of visible corrosion than silk. A greater variety of corrosion levels were observed due to the photodegradation of differently dyed wool and silk, compared to thermal degradation of differently dyed fibres. XPS of the accelerated tarnished silver and copper indicated that the main corrosion products were sulfide with some sulfate observed on copper coupons.

The red dye, brazilwood (Caesalpinia spp.) and black dye, logwood (Haematoxylum campechianum) contains the neoflavonoids brasilein and haematein respectively. These have proved difficult to identify conclusively in historical yarn samples because the routine extraction method for preparing samples for PDA HPLC analysis, involving acid hydrolysis, alters the main colouring component. The chemical structure and properties of these altered components, detectable by PDA HPLC, was unknown. The effect of the extraction process on both brazilwood and logwood (included in the study because it contains a structurally-related neoflavonoid compound, haematein, and was available in the historical period for the tapestries being researched) had to be established before analysing model and historical samples. Studies with UV-visible spectrometry and PDA HPLC on pure standards and dye extracts confirmed that haematein and brasilein were both acid sensitive. Hematein was found to form reaction products, which gave reproducible chromatographic and spectral results. Following molecular structure elucidation studies by the UoE, the logwood marker compound was identified, giving confidence to PDA HPLC analysis results for the historical tapestry samples. The molecular structure for the brasilein-derived product was not possible to elucidate. However, the UoE was able to demonstrate that a second, uncoloured component in brazilwood dye extracts was consistent and reproducible, thus enabling it to be used as a reliable marker for this dye source in the PDA HPLC analysis of historical samples.

All metal thread samples included in the MODHT project consist of metal filaments wound around a silk core. A previously unnoticed correlation of thread diameter and metal strip width with coils per unit length has been established. Rare double and triple wrapped metal threads have also been identified. Energy Dispersive X-ray microanalysis (EDX) showed 6 to 12 wt % copper in the silver bulk of most samples, suggesting European origin of manufacture. Out of 16 gilt metal threads analysed only one showed double-sided gilding of the metal filament. This was surprising as, according to the literature, the manufacturing dates of the tapestries lie within the transitional period of two manufacturing techniques. Namely, 'beaten and cut' and 'cast, drawn and rolled', the latter resulting in a double-sided gilding, the former leading to single sided gilding. New information on manufacturing techniques was obtained by Field Emission Gun Secondary Electron Microscopy (FEG-SEM) of single sided gilt metal filaments in longitudinal- and cross-section, which produced evidence of the metal being cut but rolled rather than hammered, suggesting a combination of manufacturing techniques. Gold layer thicknesses were measured to ~ 100-300nm. FEG-SEM images also revealed evidence for a leaf gilding technique.

Drawing on many years of archival and literature research, Dr Thomas P. Campbell (Curator of European Sculpture and Decorative Arts, Metropolitan Museum of Art, New York) produced a report for the MODHT project which collated information on the display history of every tapestry from the UK Royal Collection located at Hampton Court Palace (under the guardianship of HRP). The report details the curatorial history of the collection as a whole; acquisition, and use from Cardinal Wolsey to Henry VIII, through to the present day. The tapestry collection at Hampton Court Palace is of international significance, and is represented in detail in archival sources. The context of use of the tapestries is of great interest in relation to the current condition and state of degradation. The Tudor court was peripatetic: furnishings were transferred seasonally between palaces. Tapestries were displayed according to the taste of the monarch, and status of visitor permitted to the monarch's presence. After Hampton Court Palace was no longer occupied by the Royal Family, visitors were permitted and some tapestries have permanently displayed. This necessarily has an effect on the preservation of the textiles, and implications for the future care and conservation required for their long-term preservation.

A tapestry is usually hung at 90? to the angle at which it is woven on the loom. This means that the warp direction in a displayed tapestry is horizontal and the weft direction vertical. The silk and wool yarns of the weft direction carry the woven image, and also bear the weight of the enormous textile. In addition to the material preservation of a tapestry, the preservation of context of use is of great importance. A tapestry comprises a pictorial image, which were designed to be hunged. The physico-mechanical properties of the textile, such as the tensile strength, are therefore vital to the continued utility and understanding of the piece. To measure the tensile strength of a historic tapestry directly is problematic - large samples size and many replicates are required for reproducible results. Removal of large samples is incompatible with conservation, so the project aimed to develop innovative analytical techniques to provide chemical information about a fibre, which could be directly related to overall physical properties. This has been achieved with the size exclusion chromatography technique developed by HRP. A linear relationship between average molecular weight and elongation has been established through the analysis of model tapestries.

The accelerated ageing regime was carried out to produce artificially light-aged model tapestry samples for the development of novel experimental techniques by all project partners, and also to determine, under standardized conditions, the effect of historic manufacturing techniques and materials on subsequent vulnerability to environmental-induced degradation. Light ageing of model samples was carried out according to the conditions specified in the international standard BS EN ISO 105-B02, using conditions which are known to damage fibres. The precise length of accelerated ageing required for optimum utility was determined though trials between HRP and partner BBK. The accelerated ageing apparatus chosen for this work was a Xenotest 150S (Heraeus Inc.), because it incorporates a Xenon-arc bulb, which produces a light of extreme illuminance, but with a continuous spectral distribution which mimics, as closely as possible, that of sunlight through window-glass. This was determined by partners to be the most suitable likeness for the illumination conditions under which historic tapestries have been displayed. The relative humidity (RH) and temperature conditions were maintained at 65% and 20?C respectively, as required by the standardised method. The accelerated aged models have proven to be an invaluable resource for the advancement of the scientific analysis techniques.

The use of size exclusion chromatography for the analysis of historic silk textiles was pioneered at HRP in the 1990s. The existing methodology suffered from co-elution of the silk and solvent (lithium thiocyanate) peaks, particularly at the low molecular weight region of silk. This was of concern, because the low molecular weight region is of importance in the study of intensely degraded silk, such as that found in historic tapestries of the Renaissance. The rationale for the use of the existing detection wavelength was verified, so a study into the possibility of altering the sample dissolution characteristics was undertaken. From a literature search of dissolution of fresh Bombyx silk, alternative solvents (based on alcoholic calcium salts solutions) were trialled. None of these could match the existing properties of lithium thiocyanate for complete dissolution at room temperature without associative polymer damage. An alteration to the eluent system was tried - the use of a highly concentrated urea solution caused denaturation of the silk protein, which increased the hydrodynamic volume for the sample in solution. The solvent was unaffected, providing the means for successful peak resolution. The existing technique was significantly improved, permitting increased sensitivity and thus the use of smaller sample sizes.

In the search for potential micro-markers of damage, the research partners trialled and evaluated many state-of-the-art analytical science techniques, some of which had not been applied to historic tapestry material before. This generated large amounts of data, which needed to be examined and cross-referenced between the partners in great detail. HRP designed and maintained two "master" spreadsheets to enable this important process. Proposed markers of change on all materials studied (silk and wool, metal threads, mordants and dyes) were pooled into one file for model tapestries, and another file for historic tapestries. The latter was used to correlate numerical data with conservators' visual condition assessments. The spreadsheets enabled partners to compare and establish a link between model studies and historic samples. They were an invaluable tool in the process of establishing a set of parameters for damage assessment.

Historical tapestries have a particularly dense weft faced weave structure where the warp yarns are typically undyed wool and the image forming weft yarns are wool and silk dyed with a range of natural dyes. Great care was taken in the sourcing of wool and silk yarns of similar provenance, Tex value and ply as those found in historical tapestries. The yarns were dyed according to historically accurate dyeing recipes which were transcribed and converted from manuscripts and dyeing books from the 16th and 17th centuries. The correct biological source was verified for all natural dyestuffs used. To achieve the dense tapestry weave structure warp and weft were reversed during weaving; the resulting structure was assessed by tapestry conservators and found to be a very close resemblance of Renaissance tapestries. 24 wool fabrics and 14 silk fabrics were woven on a Northrop Shuttle loom and subsequently accelerated light aged in a Xenotest 150S light ageing apparatus to 60 Megalux hours which is the approximate equivalent of 400 years museum lighting of sunlight behind window glass. The unaged and aged model fabrics were used by all project partners to optimise analytical methodologies, establish parameters of damage and ascertain dye degradation profiles.

Previous studies from these groups had shown that the typical acid extraction procedures used for the removal of textile dyestuffs from wool and silk sources may alter the products of photooxidation of flavonol dyestuffs, increasing the complexity of the analytical data as a result. In this study, we have used a variety of 2D NMR techniques in combination with LC-MS to provide unambiguous identification of the "marker" compound, which is seen after the extraction of logwood-dyed fibres (WP5). In addition, the LC-MS analysis of a "marker" compound in brazil-wood-dyed fibres has provided important clues as to the source of this compound. This study has allowed our colleagues at the National Museums of Scotland and KIK greater confidence in the assignment of: - 3 samples from the current historical dataset as being dyed with logwood (and hence suspected to be restoration threads); - 16 samples from the current historical dataset as being dyed with brazil-wood.

Refinement of the PDA HPLC method to distinguish between the flavonoid dyes weld and dyer's greenweed (DGW) was essential prior to analysis of historical and model samples. This was successfully achieved by improving the chromatographic separation of three characterising compounds - luteolin and apigenin (flavonoids present in both dyes) and genistein (present only in DGW). With this improved PDA HPLC method, component ratios for five DGW model samples prepared by different dyeing processes were measured quantitatively. The respective ratios for genistein: luteolin: apigenin were found to vary between 9%:86%:5% (alum mordant, three overdyeings) and 48%:42%:9% (unmordanted, single dyeing). This indicated that the dyeing process varied the relative amounts of the three characteristic flavonoids for DGW even before ageing. This had been speculated, but not investigated, before the project. On the basis of this result and that from the dye's accelerated ageing study, it was concluded that identification of DGW is more reliable by qualitative PDA HPLC analysis than by quantitative analysis. The dye classification of historical tapestry samples by PDA HPLC was continuously reviewed in response to information from the PDA HPLC results from model samples and accelerated ageing studies.

Optimum conditions for the light-ageing of dyestuffs using a light-box from Complete Lighting Systems were determined (WP1), with ageing study samples taken at intervals up to 4,500h at 8kLux (ambient temperature/humidity):"Model" samples of wool and silk yarns, dyed according to historical recipes were aged under this regime (WP2) and dye degradation profiles have been established which confirm the relative rates of ageing of a number of historically important dyestuffs (WP5). These dye degradation profiles have allowed us to establish the following important variations in component ranges for historical dyestuffs, which have in turn allowed the identification of these dyestuffs by our colleagues at the National Museums of Scotland and KIK to be made with greater confidence (WP5): - the ageing of the quercetin/kaempferol (flavonol) component is around 6 times faster than that of the luteolin (flavone) component of sawwort; - there is no variation in the rate of ageing of flavone dyestuffs (luteolin) in samples which were dyed firstly with weld, followed by indigo, or alternatively with indigo followed by weld; - in silk dyed with dyers greenweed, the initial ratio of genestein and luteolin varies according to the dyeing procedure so this parameter should be used with caution in the analysis of aged historical samples; - degradation of the red dyestuff madder is much slower than that of the flavonoid yellows and the two main components of madder degrade at significantly different rates; carminic acid (from the red dyestuff cochineal) also degrades at a much slower rate than the flavonoids.

Building upon previous results reported by these groups, LC-MS has been developed to allow the successful chromatographic separation and mass spectrometric identification of individual components of complex mixtures of dyestuffs as found in extracts from wool and silk dyed with historical textile dyes. Negative ion electrospray ionisation has proved to be particularly useful for flavonoids and related species. Trapping of the [M-1]- species followed by MSn fragmentation has been shown to provide additional valuable structural information. This has: - allowed the unambiguous identification of dye sources such as young fustic, and hence allowed the preparation of fully authenticated tapestry models (WP2); - highlighted the possibility of other dye sources, such as sawwort, being used in historical dyeing (WP2) [although, as yet, no evidence for the use of sawwort in the current tapestry dataset has been found (WP5)]; - provided evidence for the chemical characterisation of species-specific "marker" compounds such as those used for the identification of weld (luteolin methyl ether), madder (dc II), hematein (elimination compound) and brazilein (non-degraded minor component) (WP5).

Traditional dyeing methods for preparing dyed silk model tapestries were taken from a mid-sixteenth century Italian treatise (the Plictho). Instructions for mordanting silk with alum and iron, and dyeing with madder, brazilwood, weld, dyer's greenweed and young fustic were interpreted and converted to metric units. An early seventeenth century European treatise (De Nie) was used for the dyeing of silk red with cochineal. Expertise was sought from SPINDIGO, another EC-funded project, for dyeing with wood/indigo. To ensure that the correct biological source was used for the models, authentification was undertaken by analysing test samples and dyeings with high performance liquid chromatography with photodiode array detection (PDA HPLC). The yellow and redwood dye sources used were confirmed as Reseda luteola L. (weld), Genista tinctoria L. (dyers' greenweed), Pterocarpus santalinus (sanderswood), Caesalpinia sappan L. (brazilwood) and Curcuma longa L. (turmeric). Commercial sources of Cotinus coggygria L. (young fustic) were not botanically correct, but supplied instead by the Association de Garance in Lauris, Provence, France. These well-characterised references were essential for identifying the dye sources of the historic samples as well as for modelling ageing behaviour of tapestry materials. Their value as standards extends beyond MODHT, for example in the exchange of expertise to develop a dye analysis strategy for heritage textiles though EU-Artech, a COST G8 network.

The flavonoid dye, Serratula tinctoria L. (sawwort) was included in the degradation study because it is documented as a significant yellow dye alongside weld and DGW during the historical period for the tapestries studied. There were no reports of sawwort having been identified by PDA HPLC in historical textiles, so the effect of ageing by light exposure on its analysis warranted investigation. Sawwort and weld both contain luteolin and apigenein. Because unaged sawwort contains quercetin, kaempferol and a minor component, 3-O--methylquercetin, it can be distinguished from weld. Also, sawwort does not contain luteolin methyl ether, a minor component of weld. After accelerated light ageing of dyed wools for approximately 4000 hours by the UoE, quercetin and 3-O--methylquercetin were no longer detectable by PDA HPLC in the sawwort samples. This resulted in a chromatographic and spectral profile for light-aged sawwort, which was very similar to those for light-aged weld. Luteolin methyl ether was still detectable in the aged weld; therefore this minor compound became a significant marker for dye in the PDA HPLC analysis of the historical samples. The minor components in weld - apigenin and luteolin methyl ether - may be undetectable when sample size is limited or extreme light ageing damage has occurred. Thus, dye sources for a significant number of historical samples were unable to be conclusively identified.

X-ray Photoelectron Spectroscopy (XPS) of the wool fibre surfaces has established trends in the atomic ratios and chemical states of elements due to accelerated and natural ageing. A decrease in carbon is typically accompanied by an increase in oxygen, which can be attributed to the oxidative loss of surface lipids. This was confirmed by ToF-SIMS analysis where a decrease in the signal intensity (m/z 341-) for the covalently bound surface fatty acid (18-MEA) was observed in aged samples. Covalent cystine disulphide bonds cleave to form cysteic acid due to oxidative ageing; this is a major cause for strength loss in wool. For unaged wool samples it was found that >60% of the total surface sulphur content was unoxidised (cystine disulphide, cysteine or thioester protein-lipid linkages), whereas in accelerated aged and historic samples >60% of surface sulphur was oxidised to cysteic acid. In general the relative % of oxidised sulphur and the C/O ratio at the fibre surface can be related to the reduction in tensile strength when comparing the unaged to the accelerated aged samples. XPS analysis also indicated that the majority of tested historic fibres have degraded further than the accelerated aged fibres.

From the outcomes of 2 and 3 above, the PDA HPLC analysis of 245 yellow, orange and green historic tapestry yarns revealed weld and DGW in 115 samples of wool and 59 samples of silk. Weld was found to be the principle dye for wool as a single colour source and in combination with other dyes. Re-evaluation of the weld identification after the accelerated ageing study of weld and sawwort indicated that sawwort had not been used. By contrast, DGW was predominant as a dye for silk. Dye sources for the remaining 71 samples could not be identified for several reasons: limited sample size; unknown flavonoid profile; or simply because the yarn was undyed. Logwood was identified in three black historical yarns and brazilwood in sixteen yarns by PDA HPLC using the marker compounds resulting from the acid hydolysis extraction of the neoflavonoids hematein and brasilein respectively. PDA HPLC analysis of the silk cores from nine historical metal threads revealed that young fustic was the dye of choice, used with and without weld and brazilwood. An important and necessary aspect of the project was re-evaluation of the historical sample results when new information emerged from parallel UoE research for minor dye component characterisation, and light ageing and analysis effects on the chemical compositions of the dyes. Results from other workpackages (deliverables 9 and 11) indicate that the dyeing process for flavonoid yellow dye sources influences the strength properties of aged fibres. Because the dyeing process is typically dependent on the dye, accurate identification of the original source becomes an important aspect of tapestry damage assessment. Dye identification also relates to materials selection, e.g. choice of dye source, which are certain to reflect availability as well as regional and period preferences. The results for the well-dated and provenanced tapestries studied provide context, which is relevant not just for tapestries but a wider range of historical textiles.

Exploitable results

Diese Seite teilen

Herunterladen