The analysis of medieval European manuscripts
Mark Clarke
Abstract
The benefits of analysing the materials of manuscripts, particularly pigments, are outlined. The difficulty of analysing manuscripts, compared to that for other artefacts, is explained. The development of the analysis of manuscripts, the techniques used and the results reported, are reviewed. Suitable techniques are identified, and their strengths and weaknesses are assessed. Tables are given, which provide keys to the published analyses, indexing them by date and by the techniques used. The results of these published analyses are collated in a referenced table, showing those pigments that have been positively identified by reliable techniques, by century, for medieval European manuscripts.
Table 1. Abbreviations used in the text
3DF 3-Dimensional Fluorescence spectrometry
EDX Energy Dispersive X-Ray analysis
FTIR Fourier-transform Infra-Red spectrometry
HPLC High Performance Liquid Chromatography
IR Infra-Red spectrometry
IRFC Infra-Red False Colour photography
MS Mass Spectrometry
PIXE Particle Induced X-Ray Emission
TLC Thin Layer Chromatography
UV-Vis Ultraviolet-Visible spectrometry
XRD X-Ray Diffraction
XRF X-Ray Fluorescence
Introduction
This paper reviews a selection of published analyses of paint on manuscripts. It covers both the materials identified and the methods used to make these identifications. Publications that describe techniques particularly clearly, or which describe unusual techniques are given preference. The techniques described here may of course be used on any work of art on paper or parchment, but this review concentrates on the results of analyses of medieval European manuscripts. Results from further publications are included in summary tables.
Considering the large amount written about medieval manuscript decoration there have been surprisingly few analyses published. The study of the materials used for making manuscripts is, compared to the study of paintings on other supports, in its infancy. The main obstacle has been the aspiration not to take samples for analysis. The consequence of this has been that for a long time the only analysis possible was visual examination.
Issues in manuscript analysis
Sampling versus 'non-destructive' analysis
'Non-destructive' to a chemist usually means that the sample is not consumed or altered during analysis, and that it may be recovered afterwards in pristine condition. 'Non-destructive' to a curator or conservator usually means that the artefact is not altered. Removal from the artefact of a sample is therefore viewed as 'destructive', regardless of whether the sample survives the analysis.
For many painted artefacts sampling is commonly acceptable. It is almost always possible to take samples for microscopic examination (typically cross-sections to reveal layer structures), or for micro-chemical or instrumental analysis. A typical paint sample mechanically removed using a scalpel is tiny: a microgram or so. The gap it leaves in the painted surface is almost invisible to the naked eye, being a fraction of a millimetre wide. Because the paint layer is relatively thick, there is quite a lot of material present from which information can be extracted. However, the paint and ink layers on decorated manuscripts or other works of art on paper are typically very thin compared to those on other coloured items such as easel paintings, murals or polychrome sculpture. The painted decoration also usually covers only a relatively small area - perhaps a few square centimetres or even millimetres. Removing a sample sufficiently large to be analysed may result in an area of loss, which, while it would have been far from obvious on a large object, will often be comparatively conspicuous, usually to an unacceptable degree. In consequence, it is generally considered to be unacceptable to take samples from manuscripts, and a great many curators will not permit it. This preference may have developed in part from increasing interest and appreciation of the value of manuscripts, in part from increased curatorial awareness of conservation issues, particularly for works on paper, and in part from a growing awareness among curators, conservators and even art historians of the availability of analytical equipment of greatly improved sensitivity.
This is not to say that no analyses of manuscripts have been carried out using sampling. Where analysts used samples, this is indicated below. Sampling may be justified on several grounds, well summarized by Orna and Matthews in 1981 [1]. Despite being aware of in situ techniques, they performed sampling with a scalpel and a microscope. Where samples were taken '...the lacunae were hardly discernible under the microscope, much less by the naked eye' [1, p. 62]. They further justified their standpoint:
'Given the availability of samples from a manuscript, there is no substitute for the 'good old-fashioned' methods of analysis of microscopic particles. Given that [certain methods of in situ analysis] are non-destructive, it is the position [of Orna and Matthews] that taking pigment samples from a manuscript can be equally non- destructive. First, sample-taking does not require removal of the manuscript from its permanent location. Secondly, the sampling, if well-planned, is a one-time operation which need never be repeated. Thirdly, sample-taking involves much less handling of the manuscript than [in situ] methods' [1, p. 60].
Their fourth point applies to certain techniques such as auto-radiography, for which '...only the samples, and not the entire manuscript, are subjected to high-energy irradiation...' [1, p. 61].
If carried out with sufficient care even samples large enough for polarizing light microscopy may be taken without visible damage. Increasingly sensitive methods of analysis need ever-smaller samples, and if sampling is done carefully these are undetectable to the naked eye. More recently micro-samples have been obtained by gently rubbing the surface of a miniature with a small cotton swab, which results in no visible alteration to the decoration. [2-5]. An alternative approach is to confine sampling to areas already damaged [6, 7].
Fingerprinting versus analysis
From a person's fingerprint you cannot deduce that person's appearance, but you could compare an unknown fingerprint with one of a specific individual and so know if it was left by that individual. Many analytical techniques are only to a limited extent truly analytic, i.e. only to a limited extent can they indicate what is there if you do not already have some idea. What they do allow is comparison of unknown samples with known ones. In other words, having once prepared samples of known composition (for example, by following medieval recipes), analyses and spectra can be prepared from them. Then, when you have an unknown sample taken from an artefact, you may compare the results and thus determine what was present. This approach to analysis may be described as 'fingerprinting'. Fingerprinting techniques require a short-list of candidate materials to be drawn up, using educated guesses based on documentary sources, previous analyses, etc. Fingerprinting techniques include visual examination, and IR, Raman and UV-Vis spectrometry. While it is possible to assign certain spectral features to certain molecular structures, and so deduce part of the structure of an unknown sample, in practice, for pigment analysis, spectra are used as fingerprints. Non-fingerprinting analytical techniques include the elemental techniques, such as XRF, PIXE and EDX, where information as to exactly what component elements are present is obtained.
Mixtures
When samples are taken, it is often possible to separate out components of a paint, i.e. the binding medium or media, mixtures of pigments or dyes, and any other additives. These can then be identified individually. Using non-sampling techniques, this is not possible, as the components of the mixtures will interfere with each other. Spectral peaks due to binding media can swamp low-intensity peaks from pigments in such techniques as Raman, IR or fluorescence techniques. IR and UV-Vis spectra frequently become un-interpretable when there are mixtures, as the multiple peaks combine to form broad meaningless plateaux. Even elemental analysis is complicated by several pigments having elements in common; for example a red paint showing peaks for lead may contain red lead, or red lead and white lead, but it may alternatively contain white lead plus some organic red. Raman microscopy is proving a powerful tool in the analysis of mixtures, as the optics can allow individual pigment grains to be selected for analysis, although for this to work, the parchment must not be moving too much (e.g. through vibration, inadequate support, or fluctuating relative humidity), and the specimen must be able to lie on the microscope stage.
Comparisons of analytical results with those from reconstructed samples
It has become common to proceed by reviewing contemporary written technical sources, or 'recipe books', followed by preparing reconstructed reference samples based on these sources, and then by comparing analyses of these standards with analyses of artefacts. Examples of literature comparing sources with the analyses of manuscripts sharing a date and place of manufacture include 8, 9 and 10. Of course, for such comparative techniques to succeed fully, firstly there must be samples made up of all possible pigment and medium combinations, secondly the pigments on manuscripts to be examined must be a subset of these samples, thirdly modern raw materials must produce similar results to medieval ones, and fourthly ageing over 1000 years must have no significant effect on analysable properties. As there is no guarantee that all or any of these criteria will hold, any practical work can only hope to approximate to these ideal conditions.
Medieval written technical sources
These are too many, varied, and problematic to discuss in detail here. Clarke, 2001, catalogues and describes over 400 European manuscripts from before c. AD 1500, which contain such texts [11]. Convenient introductions to the subject are Clarke, for an overview of the period, a guide for further reading and a catalogue of manuscripts, their editions and translations [11], Roosen-Runge, for the earliest treatises [9], Hawthorne and Smith, for Theophilus [12], Merrifield, for a wide range of material from the twelfth to eighteenth centuries [13], and Thompson, for the fourteenth-century treatise of Cennino Cennini [14].
General works on pigment history
Still often cited are Thompson 1936 [15], Gettens and Stout 1942 [16], Forbes 1965 [17] and Singer et al. 1954-84 [18]. These provide overviews of pigment use over long periods, but treat manuscript painting cursorily. They tend to generalize, and treat medieval Europe as if it were a homogeneous whole. Given the limitations in techniques available for non-destructive analysis at that time, much of the information in these works was derived from documentary sources, supplemented by visual assessment. A weakness is the often inadequate detailing of the sources of their information, often not even indicating whether it is derived from textual sources or analyses. Nevertheless, their conclusions often come close to current thinking.
More specific, recent and useful are the ongoing Artists' Figments series, at time of writing in three volumes edited by Feller, 1986, [19], FitzHugh, 1997, [20] and Roy, 1993 [21]. Each consists of a series of monographs on individual pigments. They summarize and review a wide range of well-attributed material, analytical, archaeological and textual. They also include suggested methods and criteria for identification for each pigment. They give histories of use, but with insufficient fine detail for our purposes here. They largely neglect the period AD c. 300-c. 1300. The examples of analytical identifications of pigment use, 'notable occurrences', are derived from good analytical practice, but are primarily concerned with painting on canvas and wood, and treat manuscript art perfunctorily. Unfortunately, many of the analytical techniques and identification criteria given in this series require samples, so are not always appropriate for manuscripts.
The history and development of pigment analyses
The earliest pigment analyses: visual examination
Attempts have been made to identify the pigments on manuscripts since the late eighteenth century, but for a long time were hindered by the undesirability of removing samples from miniatures that were large enough to analyse using the 'wet' chemistry methods of the day. The earliest examinations of manuscript pigments therefore used visual assessment alone, comparing paint with a model, mental or painted out, of how various paints look. As Hartley [22] prefaced his examination of the Book of Kells in 1885: 'For obvious reasons the pigments in question could not be submitted to any process of chemical manipulation; hence conjecture, judgment, and comparison were exercised in deciding upon an answer' [22, p. 485]. In 1914 Laurie [23, chapter V] described and tabulated his visual examination of 60 manuscripts from AD 700-1500, including the Lindisfarne Gospels, to which he added Kells in 1935 [24]. Laurie, 1914, apparently reports results obtained solely by microscopic examination of paint in situ [23]. By 1949 he also had a few samples from manuscripts available for micro-chemical analysis, but does not state which [24, p. 51].
Heinz Roosen-Runge and bis followers
The systematic analysis of manuscript paint may be said to start with the publication by Roosen-Runge and Werner in 1960 [8] and Roosen-Runge, 1967, [9] of their examinations of the pigments and media of Insular and Anglo-Saxon manuscripts. Their method, involving comparison of paint on manuscripts with reconstructed samples based on contemporary medieval artists' written recipes, has become standard, although more sophisticated analytical techniques are now employed than the simple visual comparisons they used. In order to 'evaluate the microscopical appearance of the pigments', specimens of pigments for comparison were prepared by following approximately contemporary medieval artists' recipes, and painted out on parchment using egg white (glair), fish glue (ichthyocollon) and gum.
Roosen-Runge and Werner 1960 [8] examined the pigments and medium of the Lindisfarne Gospels (c. AD 700) and four contemporary manuscripts. They used visual examination of the paint, in situ, under a microscope, up to xl 00 magnification. Reflected and transmitted light was used, as was ultraviolet illumination. Features used for comparison include colour, grain size and shape, shine, craquelure, transparency and 'fluorescence'. They appear to use the term 'fluorescence' to describe the appearance of a pigment under UV, regardless of whether actual fluorescence is taking place. They drew further conclusions based on the characteristic degradation
discolouration of lead pigments. Regarding the media, they state that, lacking micro-samples for chromatographic analysis, it is 'rather difficult to speak with any degree of certainty'. Roosen-Runge 1967 [9] is a two-volume study of paint materials and techniques in early medieval manuscripts, developing the techniques described in [8], with special reference to Anglo-Saxon and early English manuscripts. As before, he made reconstructions, which are the basis for his comparisons. He hardly discusses the problems inherent in this technique, not acknowledging that as paints age they will alter their appearance. The preparation of the samples is unfortunately not always adequately documented. He includes a large number of colour photo-micrographs of pigment reference samples and sections of manuscripts. He proposed that readers could use these as a tool to make their own identifications. However, one limitation he did not recognize was that his own identifications had been made by comparing manuscript paintings with his samples, but those using the book could only compare manuscript paintings with his photographs.1