|Home | About | Journals | Submit | Contact Us | Français|
Non-destructive pigment analysis by Raman microscopy (RM) and portable X-ray fluorescence (pXRF) has been carried out on some Bolognese illuminations and cuttings chosen to represent the beginnings, evolution and height of Bolognese illuminated manuscript production. Dating to the thirteenth and fourteenth centuries and held in a private collection, the study provides evidence for the pigments generally used in this period. The results, which are compared with those obtained for other north Italian artwork, show the developments in usage of artistic materials and technique. Also addressed in this study is an examination of the respective roles of RM and pXRF analysis in this area of technical art history.
This article is part of the themed issue ‘Raman spectroscopy in art and archaeology’.
Technical art history studies of illuminated manuscripts and cuttings ranging in date from the Anglo-Saxon period (eighth century) to late medieval times are numerous. This situation owes much to the role of Raman microscopy (RM) which has been notably successful in achieving high-resolution identifications of pigments. As a result, the evolution of developments in artistic materials from the Roman period onwards is becoming better understood. The Roman palette used in wall paintings was dominated by locally available minerals, earths and plant extracts with a limited range of synthetic materials such as lead white, verdigris, red lead and Egyptian blue [1–3]. It was later supplemented by other pigments as painting media and techniques diversified in the Byzantine and neighbouring worlds. One such pigment was lazurite (extracted from lapis lazuli found in Afghanistan) which was identified as a blue pigment in the earliest (i.e. tenth to thirteenth century) Armenian and Byzantine manuscripts reviewed by Orna . By the fourteenth century, if not earlier, communications by land or sea had enabled painters working in many parts of Europe to access a wide variety of pigments. Besides technical treatises, expertise was probably passed on orally and through practical experience . But underlying the basic commonality of the palette at that time was the way that individual artists could improvise or experiment, according to the status or scale of the manuscript, by mixing colours that were already part of that palette. One of the issues considered in this paper is the variability in artistic materials achieved through improvisation or experiment at the local, intra-school level and within a well-defined period.
In the earlier ‘Middle Ages’, most books were produced in monasteries and other major church establishments . The first RM study of an Anglo-Saxon manuscript was made on the Lindisfarne Gospels, which revealed (contrary to perceived wisdom of the day) no trace of lazurite on the manuscript; the blue colour was achieved solely by the use of indigo extracted from the woad plant indigenous in northeast England and Scotland . By contrast and dating a century later (ca AD 800), the painters of the Book of Kells in Ireland made the same use of indigo but also employed lazurite . Differential access to lazurite was surely an issue at the time as this pigment was not found on any of the other eighth, ninth and tenth century manuscripts studied in the British Library  until it was identified as a ca AD 920 pigmentary addition made as part of the restoration of a much earlier manuscript .
With the growth of cities and the evolution of the ‘Gothic’ styles from the ‘Romanesque’ between 1100 and 1200, professional resident bookmakers and mendicant clergy took over from monks. Notable sites were the universities of Paris and Bologna, and later university cities. One of the earliest manuscript studies by RM was of a thirteenth century Paris bible, in which the following pigments were identified: azurite, orpiment, lazurite, realgar, lead white, red lead, malachite and vermilion . A second early study was of the Skard manuscript, a finely illuminated fourteenth century Icelandic manuscript (ca 1360). The pigments identified were vermilion, orpiment, realgar, red ochre, azurite, bone white, verdigris and possibly green earth . Notable was the absence of lead-containing pigments, but whether this was due to stylistic preferences or non-availability of lead pigments in Iceland is not known.
We report here the results of a programme of non-destructive analysis of five illuminations and cuttings of the Bologna school in northern Italy using RM and portable X-ray fluorescence (pXRF). They were chosen to represent the beginnings, evolution and height of Bolognese illuminated manuscript production from examples in a private collection readily available for such analysis; they date to the thirteenth and fourteenth centuries and provide at least a partial indication of the pigments generally used in this period, though those used in many legal and other academic manuscripts are under-represented, as are the finest pigments of the most luxurious Bolognese productions from the Palaeologan Renaissance ‘Second Style’. They represent an extended period in which the university city became probably the most productive centre of illumination in Italy incorporating or influencing most artistic trends current in central and northern Italy between Naples and Venice. The pigments used are broadly typical of this area, though individual artists and local productions may occasionally make use of distinctive colours perhaps arising from unusual sources of supply. The first four Bolognese cuttings represent the evolution of a descriptive pictorial style in which French Gothic techniques fuse with Romanesque formalism; an increasing illusionism inspired by Byzantine models was rapidly replaced by a new synthesis of Roman and Gothic traditions, commonly but often inaccurately associated with the work of Italian artist Giotto (1266–1337). The fifth fragment represents Bolognese art of perhaps two full stages later, after the generation represented by the immediate followers of the 1328 Master and the Urb. Lat. 163 artist, the Illustratore (considered to be Tomaso Galvani) and the 1346 Master, and the arrival of Niccolò da Bologna, the most prolific and perhaps the most distinguished of all Bolognese illuminators, who dominated Bolognese illumination in the second half of the century .1 By the time of Niccolò da Bologna the ‘Giottesque’ transformation of art was fully realized in Bologna, and new concerns for emotional emphasis and descriptive detail were becoming dominant.
The results are discussed within the context of data obtained from other investigations for comparable productions in northern Italy and beyond. While this study is primarily an investigation in technical art history, it also has a methodological aim, which is to demonstrate further the combined use of different non-destructive analytical methods for illuminated manuscript analysis which is an established practice in the field [14–17]. Well established though pXRF is in providing the elemental characterization of pigments, it may be insufficient for analysing small areas or those areas containing light elements as found by Jones  in his study of illuminated manuscripts held in the University of Glasgow Library's collection . They included: manuscript (MS) 6 (S.1.6) Bartolo de Sassoferrato, Treatise on Infortiatum Bologna (ca 1400), MS 370 (V.1.7) Titus, Livius Patavinus, Livy History Bks XXI-XXX folio 153 Milan (ca 1450), MS 374 (V.1.11) Boethius, On the consolation of philosophy ?NE Italy, Genoa (1375), MS 425 (V.4.9) Lucius Lactantius Bologna (mid-fifteenth century) and MS General 1060 Michele da Genova (ca 1490; 11,12). The outcome of this study, in which these limitations of pXRF have been circumvented using RM in tandem, can now be used to direct further analysis of medieval and Renaissance illuminated manuscripts.
The Raman spectroscopic systems used for the present investigation were a Renishaw System 1000 Raman spectrometer (laser wavelength of 632.8 nm) and a Renishaw InVia Raman spectrometer (laser wavelength of 514.5 nm). Each was attached to a Leica microscope and equipped with a 1800 lines mm−1 grating, a holographic notch filter, and a thermoelectrically cooled charge coupled device (CCD) detector operating at a temperature of −70°C. Spectra were recorded in the range 2500–100 cm−1 by collecting 10–30 accumulations each with a duration of 10 s and an estimated spectral resolution of 1 cm−1; spectra were calibrated using the 520.5 cm−1 band of a silicon wafer and compared with those reported in published libraries of spectra obtained from reference materials [2,20–23].
Non-destructive elemental analysis was carried out with a Thermo Niton XL3t 900 SHE GOLDD Alloy Analyser, with a 50 kV Ag X-ray tube. The instrument, which was attached to a stand, had a video camera within the analyser head allowing the analysis area (8 mm diameter) on the manuscript to be defined. As wide a range of differently coloured areas as possible was selected for analysis with the proviso that the analysis spots were sufficiently large to encompass fully the diameter of the X-ray beam. Analysis time was 70 s. Niton's TestAllGeo and Mining calibration modes yielded semi-quantitative determinations of the content of thirty elements, of which Ca, Cu, Fe, Pb, Hg, Sn, Au, Si and S appeared to be the most informative in terms of determining the likely identity of the pigment under analysis. It is noted, however, that recent work has shown that two-dimensional scans by macro(MA)-XRF may give superior results  in this respect than the single point analyses presented in this study. In the case of fragments 1, 2 and 5 a helium flow through the analyser head helped to improve detection of the light elements present, but an instrument upgrade in advance of analysis of the other two manuscripts obviated its use. In the tables that follow, the element(s) in greatest concentration appear in italics.
The results obtained for each manuscript/cutting are shown in tables tables11–5 and figures figures66 and and77 and can be summarized as follows. Vermilion (mercury sulfide) has been used as the red in all fragments (figure 6a) and mixed with lead carbonate hydroxide (‘lead white’) to provide a fleshtone, as in Fragments 2 and 5. Lead white consistently was used for white (figure 6b). Orange-red is depicted in the form of lead(II,IV) oxide (‘red lead’) in one example only (Fragment 1, sites 3–5; figure 6c). Yellow is also uncommon: there is a single example of mosaic gold (tin sulfide) used for this colour in Fragment 2 (site 15; figure 6d) and yellow ochre in Fragment 1 (site 7; figure 6e). Certainly, the use of mosaic gold is not uncommon in manuscript illumination . The yellow-brown in Fragment 3 is uncertainly identified; it has a calcium carbonate (chalk) base (site 5; figure 6f) which may simply be the gesso layer which could, according to Cennini , also contain lead white; indeed XRF frequently detected lead together with calcium in this layer.
There is sophistication in the use of green in two fragments: Fragment 4 features in effect three separate pigment mixtures, one of which contains an organic (indigo; figure 7a); the dark green appears to be composed of azurite (figure 7b) mixed with yellow iron oxide. Yellow iron oxide (cf. figure 6e) was also identified as a component in the green in Fragment 2. In Fragment 3, the green paint used for the leaves is typically composed of lead tin yellow (type II) mixed with lazurite (figure 7c,d); azurite was also found as a component in other areas of green. It is remarkable, although not unknown [26,27], that the expensive pigment lazurite should be used in this way, a privileging of the materials of which the artist would be aware but not necessarily either the patron or any other viewer.
As regards the blues, both lazurite and azurite are represented: the former in Fragments 1 and 5, and the latter in Fragments 2, 3, 4 and 5, typically for the larger background areas. The relatively large scale and high quality of the original volume from which Fragment 1 is derived is perhaps reflected by the use of a luminous lazurite background (table 1) where more modest academic manuscripts would use the azurite backgrounds as found in the cuttings examined. The different colours and tones produced by the combination of lazurite with white lead is striking, though the difference between the pilgrim's mantle and James's tunic is presumably due to the predominance of azurite as the local colour of the latter, while lazurite was added to give luminous shadows to the tunic (table 4). A remarkable feature of Fragment 5 is that Niccolò's kneeling figure has a delicate pale blue with much of the luminosity of pure lazurite, while lazurite and azurite were layered or combined to create a very different green-blue colour for St James's tunic: the precious blue fits its wearer, but the visual result is far closer to what one would expect of an azurite–malachite blend. These pigments are all costly and therefore privileged in status wherever used in the art of this period: their presence imparts dignity to a notable degree to the individuals represented and to the manuscript which adopts them.
In Fragment 2, azurite was used with an organic red lake and other pigments to achieve a purple-brown (vermilion and carbon-based black; figure 7e) and a pale red (chalk). The ink in Fragment 1 has not been characterized, but is likely to be iron gall .
Comparison of the palettes of Fragment 1 and the Paris bible analysed by Best et al.  shows some similarities as expected, but the latter makes use of a wider palette (table 6). On the other hand, there are significant differences between the contemporaneous Fragment 5 and the Skard manuscript  which may be expected from geographical considerations; the latter features orpiment, realgar, verdigris/green earth and bone white, which are absent from Fragment 5 (table 6).
Comparison of the data with other broadly contemporaneous European illuminated manuscripts (Bensi's review ; see table 6) shows there is striking similarity in the use of pigments which, furthermore, appears to extend to most of the identifications made by pXRF of the University of Glasgow Library's illuminated manuscripts . But equally notable are the instances, which are few, of differences. The example of the red insect-based kermes pigment noted on one or other of the Latin manuscripts analysed  seems to stand alone, as does the unusual mixture of red lead and cinnabar in a Bolognese manuscript painted in a Paris workshop . There may also be other instances of ‘outliers’ among the Glasgow manuscripts; the pinks, for example, were rendered differently: an iron, calcium and mercury-rich mixture and a lead–mercury mix in MS370 and MS6, respectively .
The contents of table 6 point on the one hand to the existence of an artist's palette that had wide chronological and geographical currency. On the other, there is variability, albeit at a low level, expressing itself in the form of local adaptation or experimentation on a very local, even personal level. For instance, there appears to be as much variability among the five Bolognese examples as there is between these examples and ones executed in northern Europe in the thirteenth to fifteenth centuries. The question then arises as to whether a more general corollary of these observations is that artists' materials and techniques were more static during this time than their styles. In any case, this issue of variability should be open to review as more analytical data become available and perhaps more significantly as more data points are taken from each manuscript. The concurrence of lazurite and azurite can be used as an example to illustrate the point: Clark's  finding of lazurite layered over azurite in a thirteenth century north Italian choir book might have appeared anomalous until Ricciardi & Delaney  encountered it in a work by Niccolò da Bologna, thereby apparently supporting the view that this was not uncommon at the time ; on the other hand, this was not found in the cutting attributed to the same painter in this study (Fragment 5), although the layer structure was not explicitly examined.
Methodologically, the high spatial resolution of RM gives it a distinct advantage over pXRF in analysing fragments such as the ones here, in which the size of many individual painted areas is small. A further advantage of RM is its ability to identify organic pigments and ones, such as the crucial lazurite, that are based on light elements; pXRF cannot detect the former and the latter only by indirect inference. pXRF systems, however, are readily portable, thereby circumventing the need to transport the object to the laboratory, coupled with the associated high insurance costs . The role of pXRF is clear: it can provide a rapid preliminary and in situ assessment of the range of pigments and identify particular locations on the manuscript requiring subsequent more detailed (non-destructive) analysis by RM. Furthermore, the trace element content of pigments, determined by pXRF, has the potential of resolving different sources or even grades of an individual pigment (for example, Mn in a red ochre); this is a topic that could be usefully explored further. But equally evident is that instrumentation in technical art history is constantly evolving (for example, smaller and less expensive Raman systems are more widely available), technical advances are responding to the need for finer-grained information of the kind exposed in this study, complementarity of techniques and the availability of more refined reference collections of medieval pigments  will become increasingly important. As well as the role of MA-XRF  referred to previously, mention can be made of the developments of other techniques for examination of manuscripts used such as fibre optic reflectance spectroscopy for point analysis of pigments  and binders, and the broader view of illuminations and watercolour paintings provided by hyperspectral imaging [43–45].
1As most medieval illuminators are anonymous, it is customary to identify them by reference to their best known or most useful work. In the case of Bolognese artists, two of the most important can be related to dated statutes.
T.D.C., R.J.H.C. and R.J. designed the research, T.D.C. performed the Raman analyses; R.J. carried out the p-XRF analyses; R.G. selected the manuscripts for study and contributed to the project's art historical aspects. All authors contributed to the writing of the article and gave final approval for publication.
The authors have no competing interests.
R.J. acknowledges the Leverhulme Trust, which funded the project Towards non-destructive analysis in archaeological and conservation science which included some of the pXRF analyses mentioned in this paper. The spectrometer used in this study was funded through an award to R.J.H.C. by the Engineering and Physical Sciences Research Council.