Raman microscopy and x-ray fluorescence analysis of pigments on medieval and Renaissance Italian manuscript cuttings

Abstract

Italian medieval and Renaissance manuscript cuttings and miniatures from the Victoria and Albert Museum were analyzed by Raman microscopy to compile a database of pigments used in different periods and different Italian regions. The palette identified in most manuscripts and cuttings was found to include lead white, gypsum, azurite, lazurite, indigo, malachite, vermilion, red lead, lead tin yellow (I), goethite, carbon, and iron gall ink. A few of the miniatures, such as the historiated capital “M” painted by Gerolamo da Cremona and the Petrarca manuscript by Bartolomeo Sanvito, are of exceptional quality and were analyzed extensively; some contained unusual materials. The widespread usage of iron oxides such as goethite and hematite as minor components of mixtures with azurite is particularly notable. The use of a needle-shaped form of iron gall ink as a pigment rather than a writing material was established by both Raman microscopy and x-ray fluorescence spectroscopy for the Madonna and Child by Franco de’ Russi.

By comparison with the huge advances made over the last 60 years in the technical analysis of panel paintings, that of the materials used and techniques adopted for painting illuminated manuscripts remains comparatively underdeveloped (1). This is largely on account of difficulties in obtaining pigment samples from manuscripts for analysis. The technique of Raman microscopy (RM) has opened up this field of enquiry since it has high specificity, sensitivity, reproducibility, spatial (approximately 1 μm), and spectral (approximately 1 cm^-1) resolution, and is both nondestructive and applicable in situ. It is thus highly appropriate to the study of materials for which sampling may be undesirable.

Application of RM to the study of manuscripts, paintings, ceramics, glass, icons, and archaeological artifacts etc., permits the identification of most of the pigments present and many dyes, facilitating the establishment of artists’ palettes at different periods and in different localities (2–17). It can also aid in the identification of previous restorations and in the detection of forgeries (18). Further, x-ray fluorescence (XRF) spectroscopy has also been used, this having previously been shown to be very effective for the identification of gold, silver and bismuth in illuminations (19 and 20).

The purpose of this study was to undertake RM analyses of an extensive collection of Italian manuscript cuttings as well as a series of high profile medieval and Renaissance manuscripts and miniatures, all held in the Victoria and Albert (V&A) Museum. All the studies pertain to materials in use between the 12th and 17th centuries, principally in Lombardy and Tuscany, together with significant comparative materials from the Veneto, Emilia-Romagna, and Umbria. The research involved the study of works attributable to many well-known illuminators such as Gerolamo da Cremona, Franco de’ Russi, and Bartolomeo Sanvito (see Fig. 1). The sizable database gathered has made it possible to estimate the frequency of use of individual colorants and how often pigments were used in admixture to achieve desired color effects. This information would augment our knowledge of any regional variations in the use of pigments and any developments to the typical palette in Italy during the later Middle Ages and Renaissance.

Fig. 1.

Some of the cuttings analyzed in this study: (A) St. Giustina of Padua disputing with the Emperor Massimian, by Gerolamo da Cremona (V&A accession number 817-1894); (B) Madonna and Child, by Franco de’ Russi (V&A accession number E.1275-1991); (C) two pages from the Sonnets and Triumphs by Francesco Petrarca, painted by Bartolomeo Sanvito (V&A accession number L.101-1947). Copyright Victoria and Albert Museum, London.

Results

Manuscript Cuttings.

Of the approximately 1,000 cuttings available, some were loose but most were already framed in large cardboard mounts, usually about 53 cm × 38 cm. These had been custom made in the late 19th or early 20th century, when numerous medieval manuscripts were being cut up to permit the extraction of their most interesting portions. As the miniatures could not be removed from their mounts for analysis, and no remote head for the spectrometer was available, only the miniatures nearest to the edges could be examined.

In total, 174 manuscript cuttings and miniatures were analyzed, all being Italian albeit from different schools (Lombardy, Tuscany, Veneto, Emilia-Romagna, and Umbria) and various workshops. Many groups of cuttings exhibited similar styles, characteristics, and pigments, and were obviously taken from the same manuscript; in such cases only one or two representative cuttings were analyzed.

The palette on most of the cuttings was consistent and showed little variation regardless of the provenance of the fragments. Table 1 shows the pigments identified on the cuttings, and Fig. 2 shows representative Raman spectra for most of the compounds identified (21–47). The occurrence percentages shown in Table 1 are calculated by taking the number of individual manuscript cuttings and miniatures found to contain the indicated pigment divided by the total number of pieces examined. This calculation is based on raw data and no attempt was made to correct for the fact that some of the cuttings probably came from the same original manuscript or that some pigments may have been present but were unreachable on a given item due to the intervening frame.

Table 1.

Characteristic Raman bands and % occurrence of pigments in the Italian miniatures

Pigment	Raman bands (cm-1)	Occurrence in cuttings (%)
Blues
Azurite, Cu3(CO3)2(OH)2 (2, 24, 25)	1,095, 400, 247	58
Lazurite, (Na,Ca)8[(Al,Si)12O24]Sn (2, 25–29)	1,096, 548	30
Indigo, C16H10N2O2 (14–16, 22, 23, 30–33)	1,575, 599, 546, 252	17
Reds and oranges
Vermilion, HgS (5, 6, 17, 25)	343, 253	61
Red lead, 2PbO·PbO2 (34–36)	548, 390, 223	22
Hematite, Fe2O3 (6, 8, 22, 37, 38)	613, 411, 292, 226	3
Goethite, FeO(OH) (22)	553, 390, 302	4
Realgar, α-As4S4 (21, 23, 39, 40)	358, 225, 196	2
Yellows
Lead tin yellow type I, Pb2SnO4 (3, 4, 25, 41)	458, 294, 275, 198	14
Orpiment, As2S3 (21, 23, 39, 40)	355, 312, 294	1
Mosaic gold, SnS2 (42 and 43)	314	9
Greens
Malachite, CuCO3·Cu(OH)2 (25)	1,491, 433, 269	20
Brochantite, Cu4SO4(OH)6 (44)	974, 443	6
White
Lead white, 2Pb(CO3)2·Pb(OH)2 (25, 35, 41)	1,052, 1,048	36
Blacks
Carbon black, C (45)	∼1,590, ∼1,345,	24
Iron gall ink, iron gallotannate (46 and 47)	∼1,580, ∼1,480, ∼1,340	39

Fig. 2.

Synopsis of typical Raman spectra of pigments on the manuscripts studied; mixtures of realgar and related photo-degraded compounds are shown for two manuscript cuttings (V&A accession numbers 247.1 and 277.4).

A significant number of the manuscript cuttings (∼25%) contained red, pink, yellow, and green (blue indigo or azurite plus yellow) areas which gave either no spectrum or high fluorescence and were probably painted with unidentified organic colorants. Calcite (CaCO₃, identified by its characteristic carbonate band at 1,088 cm^-1) (7), gypsum (CaSO₄·2H₂O, with a band corresponding to sulfate at 1,008 cm^-1) (7) and quartz (SiO₂, with a distinct Raman band at 465 cm^-1) (11) occasionally were found mixed with the pigments described in Table 1.

Orange realgar (α-As₄S₄), its yellow photo-induced degradation product pararealgar (β-As₄S₄), their intermediate phase (χ-As₄S₄) and mixtures thereof were encountered only rarely (see Fig. 2) (21). The single occurrence of the use of rutile, a TiO₂ polymorph with Raman bands at 607 and 447 cm^-1, as a pale orange pigment was noted (22).

In one case (manuscript cuttings 243.4 and 243.5) the unambiguous identification of Prussian blue, Fe₄[Fe(CN)₆]₃·14-16H₂O, with its strong Raman band at 2,154 cm^-1 attributable to ν(CN), prompted us to revise the date for a set of four cuttings from the 17th to the mid-18th century (23). Fragment 2,879, a capital E, also showed Prussian blue, but examination established that the latter had been applied as a substitute for the failing original pigment, azurite.

Only a few of the over 1,000 Raman spectra obtained in this project remain unidentified, including a blue (1,635; 1,489; 1,417; 1,181; 619; and 519 cm^-1) and a purple (1,645; 1,486; 659; 592; 508 cm^-1) organic dye or pigment.

Historiated “M”, St. Giustina of Padua, Illuminated by Gerolamo da Cremona.

This historiated initial “M” (21cm × 19 cm) had been cut from a choir book (traces of staff lines and notes are still visible at the edges of the miniature—see Fig. 1A). It was painted by Gerolamo da Cremona probably in Lombardy, Italy, approximately 1470 and depicts St. Giustina (or St. Justina) of Padua, who is said to have been martyred in 304 A.D., disputing with the Roman Emperor Maximian (250–310 A.D.). Da Cremona was a painter and illuminator active in Ferrara, Mantua, Siena, Florence, and Venice in the latter half of the 15th century. This illumination is a very good example of the innovations brought in by Italian Renaissance artists from the beginning of the 15th century. The renewed interest in the ancient classical world is expressed by the style of the architectural space chosen to set the scene, characterized by marble columns with brazen capitals. The architectural space is fully analyzed in three dimensions and much emphasis is put on the perspective of the scene.

The miniature is very colorful, but the palette is remarkably simple. The artist obtained subtle variations of color by mixing only a few pigments (Table 2). The Raman analysis of this object allowed the identification of the following materials: lazurite, azurite, indigo, malachite, lead tin yellow type I, vermilion, lead white, carbon, and iron gall ink. An organic purple is also present but it only yielded a highly fluorescent background and could not be identified. When viewed under a microscope, the darkened shading on the blue column shaped as an “M” and the gray armor on the soldiers in the background reveal some sparkling particles which did not yield a Raman spectrum. These areas were analyzed by XRF, which showed that a large amount of silver is present. Gold was also detected in all the golden-looking areas, such as the golden embroidery on the Emperor’s dress. These results suggest that the miniature was originally decorated with various shades of gold and silver and so must have looked much brighter then than now.

Table 2.

Summary of pigments found on three miniatures

Pigment	St. Giustina Gerolamo da Cremona	Madonna and Child Franco de’ Russi	Petrarca manuscript Bartolomeo Sanvito
Lead white	✓	✓	✓
Gypsum			✓
Azurite	✓	✓	✓
Lazurite	✓	✓	✓
Indigo	✓	✓	✓
Malachite	✓	✓	✓
Vermilion	✓	✓
Red lead	✓
Lead tin yellow (I)	✓	✓	✓
Goethite		✓
Carbon	✓	✓	✓
Iron gall ink	✓	✓	✓
Mosaic gold		Impurity	✓
Shell gold		✓	✓
Shell silver	✓		✓

Illuminated “B”, Virgin and Child, Illuminated by Franco de’ Russi.

This initial letter “B” (70 mm × 63 mm, see Fig. 1B) had been cut from an illuminated manuscript on vellum. It shows the Virgin and Child seated among rocks and, at the top left, a putto holding a red shield signed by the artist, “M. Francho”. Franco di Giovanni de’ Russi from Ferrara was active in the second half of the 15th century.

This is one of only two signed works by this artist, who studied in Ferrara and worked there and in Urbino, Padua, and Venice. This miniature was probably made between 1460 and 1480. At first sight, the miniature looks very traditional and its palette unremarkably normal (see Table 2). However, a closer look at the various areas of the miniature reveals in all of the shaded areas a large number of needle-like crystals (see Fig. 3A–C). Their Raman spectra show features typical of the pigment in the background, plus the bands characteristic of iron gall ink (see Fig. 3D) (46). As far as we are aware, crystalline iron gall ink has not previously been encountered and documented, as iron gallotannate normally shows amorphous character when found attached to parchment.

Fig. 3.

Microscopy images of areas rich in needle-shaped crystals of iron gall ink: dark blue shadow area on the Virgin’s mantle, × 100 magnification (A) and × 400 magnification (B); proper right side of mouth of the Virgin, × 100 magnification (C); Raman spectra from the needle-shaped crystals of iron gall ink (signature band at 1,480 cm^-1) in the Child’s flesh tones. The asterisks mark the peaks due to vermilion and lead white in the carnation mixture (D).

XRF analysis was carried out on a dark brown/black letter on the right border and also on areas in the faces of Mary and the Christ Child that contain an abundance of the needles. In the former case, iron and zinc are the predominant elements with small amounts of arsenic and mercury (Fig. 4, top). Medieval recipes for iron gall ink utilised iron(II) sulfate (green vitriol) and sometimes also copper(II) sulfate (blue vitriol) along with gallotannic acid (from galls) and a gum Arabic binder (48). These sulfate salts were probably not pure and may have been mixed with other metal salts including zinc(II) sulfate (or white vitriol) as it is known that some pre18th century historical inks contained more zinc than iron (47, 49–51). In the carnation areas lead (from lead white) is the predominant element, but small amounts of iron and zinc can also be detected (Fig. 4, bottom). It seems likely that the zinc is associated with the ink itself rather than another pigment or a material involved with some treatment of the vellum substrate. It is also uncertain as to whether the significant amount of zinc present is related to the morphology of the iron gall ink crystals.

Fig. 4.

Carnation area on Jesus’ face containing iron gall ink needles, and XRF spectrum obtained from the center of the crosshairs (top); iron gall ink letter at right edge of the de’ Russi cutting, and XRF spectrum obtained from the center of the crosshairs (bottom).

Several possibilities for the origin of iron gall ink in the painted areas were considered. Although unlikely, the artist could have painted over an area which was already covered with words or notes executed in iron gall ink and the crystals formed sometime afterwards, much like lead soaps are known to form in oil paintings (52). Alternatively, de’ Russi may have deliberately added the iron gall ink as a colorant along with other pigments. The evidence suggests that iron gall ink in crystalline form was used on purpose in the shaded areas to give modelling to the miniature; in fact, it is used in the folds of the Virgin’s robe, in the flesh areas, and several other sites.

Several observations support this suggestion. The crystals appear to erupt from the surface but seem to be an integral part of the paint layer. Some of the crystals exhibit the same cracks as the surrounding paint layer. The crystals were occasionally found with long (up to 400 μm), adjacent needles aligned in parallel, possibly as a result of brushwork. In other areas the needles have the appearance of being cracked off at the end as if the crystals have been subjected to grinding rather than having been formed via a crystallization process subsequent to the application of the paint layer.

Sonnets and Triumphs by Francesco Petrarca, Probably Illuminated by Bartolomeo Sanvito.

This manuscript is one of the few to be studied which is complete, though very few pages contain any illuminations (Fig. 1C). The 188-page codex is written in humanistic script on vellum and measures 23 cm × 14 cm. Bartolomeo Sanvito (1435–1518), was active as a scribe and perhaps as an illuminator in Rome and Padua in the second half of the 15th century. This illuminated work was made between 1463 and 1464, probably in Padua, and contains poetry by Francesco Petrarca (or Petrarch, 1304–1374). The three frontispieces with colored drawings, two illuminated pages, and several of the historiated initials were analyzed and the palette identified as being typical of its period (Table 2). In addition, several yellow and red areas, which failed to give a Raman spectrum due to high background fluorescence, were noted; these are probably organic dyes of vegetable origin.

The gray and black areas of the images in Fig. 1D (e.g., the shading in the face of the woman and the modelling of the columns) gave no Raman spectrum but XRF analyses verified that these areas contain a large amount of silver. This observation and the appearance of these areas under a microscope suggest that, as in the painting by da Cremona, shell silver was originally applied but underwent subsequent oxidation. Shell gold was used in the letters in these paintings as well as for initials at the beginning of stanzas.

Discussion

The spectroscopic studies on the cuttings have shown that the 12th–17th century palette of a traditional manuscript tends to be constant, regardless of date and school, at least within Italy. Very few cases were identified in which rare pigments or dyes were used (such as realgar/pararealgar species). Investigation of the occurrence of pigments in these manuscript cuttings and miniatures, which has heretofore not been reported in such a large study, provides a measure of quantitative evidence for certain trends. For example, it was anticipated that azurite would be found more often than lazurite, given the high cost of lapis lazuli. This study shows that, of the 140 cuttings containing one or both of these blues, azurite was used twice as often as lazurite, with 10% of these pieces having both pigments, often in admixture.

Several important observations were made about manuscript cuttings during this survey. Thus, when analysing cuttings from musical manuscripts of various dates and provenance, it was noticed that the color and morphology of the vermilion particles in illuminated or simple capital letters is different from that used to trace the staff lines. Invariably, these lines were painted using the darkest, most intense shade of the pigment which, viewed under a microscope, always consisted of very small, deep red particles, very uniform in size (1–2 μm at most) and hue. On the other hand, the crystals of the vermilion specimens used for capital and illuminated letters were always less vibrant in hue, slightly more orange or paler than those found on the staff lines, although they are still considered to be of good quality, showing that more binding medium seemed to have been used than on the staff lines. In addition, red lead was found mixed with the vermilion in other colored areas approximately 30% of the time but never in the staff lines. This finding shows another common practice among the various scriptoria, workshops, and artists, suggesting that it was an accepted, regular procedure to use a specific grade of vermilion for tracing the staff lines on choir books.

One of the unexpected results of this survey relates to the composition of all areas painted with azurite. Crystals obtained from the natural mineral are almost always mixed with a minority of green, orange, reddish and dark brown, or black crystals (see Fig. 5). The green ones are almost always malachite (we have verified this on most manuscript cuttings where azurite is present), and it had usually been assumed that the orange to black crystals were particles of the two copper oxides, the orange-red cuprite Cu₂O and the dark brown or black tenorite CuO, which are typically present as weathering or degradation products in most mineral specimens of azurite (53). However, when these particles were analyzed during this study, the only spectra obtained belonged to the iron oxides hematite, Fe₂O₃, and/or goethite, FeO(OH) (see Fig. 5), never cuprite or tenorite which do yield reasonably intense Raman spectra using the correct experimental conditions (22).

Fig. 5.

Microphotograph (× 200) of the blue area on manuscript cutting 4006 and Raman spectra from the blue, the orange and the dark brown crystals. The asterisks mark the goethite peaks in the hematite spectrum.

As far as we know, there are no medieval recipes which recommend the use of iron oxides in admixture with azurite when painting miniatures. However, this seems to be a regular occurrence on medieval manuscripts, regardless of the date and provenance of the miniatures analyzed, as if the combination of these compounds was a regular and accepted practice in scriptoria and workshops to achieve the best possible appearance for blue surfaces containing azurite.

In summary, this survey of some of the V&A medieval and Renaissance manuscript cuttings allowed the gathering of data on the use of pigments and dyes in various regions of Italy from the 12th–17th century. The occurrence of unusual materials, such as a crystalline form of iron gall ink used as a pigment, was also documented. No significant regional variations in the palette found were identified, but a regular pattern in the usage of specific pigment mixtures (azurite with iron oxides) or pigment grades (high quality vermilion used in specific circumstances) was recognized. The development of a sufficiently representative database of artists’ materials on Italian medieval and Renaissance manuscripts will help to define area-specific patterns of pigment usage. In some fortunate cases it might be possible to get an idea as to the scriptorium or workshop from which the manuscript came, and even by whose hand. It will also help in refining the known chronology of usage of various materials, and possibly allow the tracing back of mineral sources and the history of known manufacturing processes and trade routes.

Materials and Methods

The V&A manuscript cuttings and miniatures were analyzed by RM and also by XRF when deemed necessary. RM was carried out using a Renishaw 1000 Ramascope spectrometer, equipped with a Renishaw HeNe laser operating at 632.8 nm. The system was calibrated to better than 1 cm^-1 using a neon lamp before measurement. Spectra were recorded as an extended scan. The laser beam was focused with a 50× objective lens to realize a spatial resolution of approximately 2 μm. The laser power at the surface of the sample was held to ≤ 1 mW. The low wavenumber limit of operation for this instrument was approximately 170 cm^-1.

XRF measurements were performed with an ArtTAX micro-XRF spectrometer equipped with an air-cooled, low-power molybdenum tube and new generation polycapillary x-ray optics. The experimental conditions were 50 kV, 600 μA, and 100 s livetime. The spatial resolution of the micro-XRF spectrometer was approximately 200 μm.