Howell G. M.
a School, of Chemistry and Biosciences, Faculty of Life Sciences, University of Bradford, Bradford BD7 1DP, UK. E-mail: email@example.com
Research on ancient manuscripts has dealt with the identification and use of pigments in their historiation and the degradation of the vellums, but relatively little has been undertaken on the composition of the inks. This chapter traces the use of carbon-based inks, the so-called Indian inks, through to mediaeval iron-gall or gallotannate inks and, more rarely, coloured inks made from logwoods and cinnabar. Recent interest in ancient ink composition has arisen from the significant destructive effects noted particularly in the use of iron gallotannate inks on vellums and paper. Following an historical survey of ink through the ages and a discussion of ink preparation and composition, this chapter considers three case studies in which the analysis of inks has been instrumental in providing information for assessment by conservators and historians: these are the Vinland Map, the Voynich Manuscript and the Beato de Valcavado manuscript. In particular, the role of Raman microscopy in the discovery of anatase in the ink of the Vinland Map is highlighted in the ongoing controversy surrounding the importance of this manuscript in the discovery of North America, and the interesting Raman spectroscopic analyses of inks in the Beato de Valcavado are discussed.
Many of the earliest studies of manuscripts involving Raman spectroscopy focused upon the analysis of coloured mineral pigments in historiated manuscripts from which much novel information could be obtained about their composition and preparation technologies. However, complementary studies of these inks have generally been less frequent, yet as will be seen, this can often provide interesting and supportive information on the holistic preparation of manuscripts and printed works of art. The distinction between a paint and an ink is not easily defined as both contain pigments dissolved or suspended in a carrier liquid or solvent with added binders and drying agents to produce a coloured fluid: the major difference, therefore, is one of application rather than composition and in some cases printing inks can be used as paints and vice versa with an adjustment of pigment content. Generally, paint is applied more thickly than ink, and often the drying process is chemically different, depending upon the ingredients used in the ink or paint.
The adoption of fluid inks and reed pens for their application has been attributed to China around 2700 BCE; hence, ink can be accredited to almost 5000 years of human history. The name ink is believed to originate from the Latin incaustum and the later mediaeval French encre; as was experienced with pigment minerals, both have been found to apply to the earliest carbon-based inks and to the later iron gallotannate (iron gall) inks.1–3 The earliest ink, formulated in China from carbon black suspended in an aqueous solution of water and gum arabic, is now known as Indian ink because the best carbon supplies at that time for this purpose were sourced in India; the carbon was synthesised by burning wood such as pine, in which the partially combusted resin helped to bind the sooty deposits, in a limited amount of air under an upturned iron or ceramic dish. The gum arabic had a multifunctional role in keeping the carbon particles in suspension whilst also thickening the writing fluid for ease of application using quill or reed pens. Other carbons were sourced from the calcination of animal bone and ivory, which produced a deeper black colour – these contained residues of calcium phosphate derived from the hydroxyapatite component and provide a useful analytical Raman spectroscopic signature for the detection of bone black or ivory black, with the characteristic wavenumber of phosphate ion stretching at 960 cm−1, which can be used to discriminate between vegetable and animal origins for the source of the amorphous carbon used in the manufacture of ancient inks.4
The discovery of natural mineral oils and petroleum afforded another opportunity for the production of a deep black sooty residue upon combustion in a limited supply of oxygen. A hierarchical basis for recipes for the production of carbonaceous soot existed in which certain botanical materials were highly prized for combustion to form the blackest inks, such as peach stones, almond shells and vine twigs. There was much empiricism in the formulation of early inks as evidenced by the universal adoption of gum arabic as a binding agent; in addition to assisting the suspension of the insoluble carbon particles in the aqueous ink medium, the water-soluble gum arabic modified the viscosity of the ink, so assisting in the writing flow when applied with reed pens, quills and brushes and also improved the adhesion of the ink to the writing substrate. However, the addition of too much gum arabic resulted in a brittleness of the applied ink when dry and a tendency for the writing to flake off – hence, the debris found between the leaves of ancient manuscripts frequently contains particles of ink from the associated script that can provide a rich source of sampling to derive analytical information without involving the further destruction of the manuscript text. Different names have been recorded for carbon black inks through the ages, dependent upon their formulation or source, such as bistre,5 an extract from sooty fires that possessed a warm brown colour, and sepia, a dark, semi-transparent ink from cuttlefish, which was much used by the Roman scribes.
In mediaeval times, iron gall ink replaced carbon black ink as the favoured medium of writing; although iron gall ink has recently been detected6 using X-ray spectrometry on the Codex Eusebii Evangelorum (the Vercelli Gospels), the oldest existing version of the Gospels written in Latin and dating from the 4th century CE. Iron gall ink, more correctly described as an iron gallotannate, was the first water-based ink to be made from a chemical reaction between aqueous solutions of iron(ii) sulfate and extracts of oak galls with the addition of gum arabic. Oak galls are spherical, nut-like protuberances resulting from the egg-laying of wasps on oak trees. The best galls were those fully developed from which the emerging wasp larvae had hatched. As encountered with the carbon-based inks, there are many empirical recipes in existence for the manufacture of iron gall inks; indeed, as we have noted above, the chronology for the first appearance of iron gall inks actually predates the mediaeval period and Pliny in the 1st century CE describes in detail the preparation of aqueous gall solutions that blacken in the presence of copperas, an iron sulfate ore. However, Pliny was specifically referring to the detection of the adulteration of verdigris by the addition of cheaper copperas through the formation of a black colouration on exposure to an infusion of nutgalls. Outside of the Vercelli Gospels, the first record of the use of iron gall ink as a writing medium seems to have occurred in the Dead Sea Scrolls, from the late 3rd century CE.
The preparation of iron gall ink was a rather complex alchemical procedure, as indicated by the following ancient recipe for the manufacture of the highest quality iron gall ink:
8 oz powdered Aleppo galls; 4 oz logwood chips; 4 oz iron sulfate; 3 oz powdered gum arabic; 1 oz copper sulfate or verdigris (basic copper acetate); 1 oz sugar; all heated and triturated in 12 pounds water followed by filtration and the addition of alum, ammonia, beer, lemon juice, oil of cloves, ground walnuts, lavender, wine, boiled oil, and extract of amber or shellac in brandy to minimise the growth of mould.
An understanding of the chemistry of the preparation of iron gall inks reveals the roles of the iron complex and its formation: the enzyme tannase from the fungus Aspergillus niger in oak galls releases gallic acid, a triphenol carboxylic acid, C6H2(OH)3COOH, and glucose through the catalytic hydrolysis of gallotannic acid ester. Iron(ii) ions from ferrous sulfate then form a dark grey 1 : 1 iron gallate complex, which releases hydrogen ions and is then oxidised by aerial oxygen on the manuscript to a ferric pyrogallate complex that is black in colour. An excess concentration of iron(ii) causes the ink to gradually fade, a problem experienced with the multifarious recipes in existence in the Middle Ages, and this also stimulates the release of hydroxyl ions and the formation of hydrogen peroxide through a Fenton reaction. It is this last property that causes the destructive damage effects noted on ancient manuscripts involving iron gall inks.
Iron gallotannate inks quickly became the medium of choice for mediaeval scribes because, unlike the carbon-based inks they replaced, they interacted physically and chemically with cellulose substrates, conferring better adhesion and permanence of the writing on the script.7,8 Even when used with parchments and vellum, the iron gall inks had a noteworthy adherence to their substrate and could only be removed by mechanical detachment and scraping, unlike the carbon-based inks which could be more easily erased and washed off. However, this improvement in the writing permanence of iron gall inks caused severe corrosion problems for paper manuscripts in particular. In some cases, this process resulted in the formation of holes in the manuscript (lacunae) in place of the writing; many manuscripts have suffered irreversibly in this way and pose problems for their conservation and the preservation of their integrity.9,10 It has been found that arresting the decay can be achieved by the application of calcium bicarbonate, lime, magnesite and calcium phytate, but generally, irreparable damage has been done to the original script and text.11–13
The corrosive effect of iron gallotannate inks upon cellulosic substrates can be related to the iron-catalysed breakdown of cellulose in an acidic environment.14 It will be seen below that the formation of the iron(ii) gallate complex releases hydrogen ions and decreases the pH significantly to about 2; in this process, excess Fe2+ ions then react with acidic decomposition products of the cellulose to form hydroxyl and oxygenyl radicals, from which the subsequent creation of hydrogen peroxide destroys the cellulose substrate and oxidises the iron15 to Fe3+. Hence, it has been suggested14 that measurement of the Fe2+/Fe3+ ratio in an ancient iron gall ink could provide a means of assessing its age, although clearly the actual rate of degradation of the ink would be dependent upon several environmental factors, not the least of which would be the recipe and formulation of the original ink, which was certainly not standardised in any way. The elemental migration of iron atoms into the substrate also can provide a measure of the age of the script, as determined from Auger spectroscopy, but pitfalls can be encountered particularly in the form of hair follicles in vellum substrates,16 which provided a route of penetration of the ink components into the lower regions of the substrate. It may perhaps be inferred from this discussion that iron gall inks used on vellum are not subject to the extensive corrosive material degradation and destruction noted for their cellulose analogues. This is not strictly the case since even the preparation of high quality vellums and parchments for scriptorial purposes resulted in the presence of chemical residues in the writing substrate, such as alum, lime and bicarbonate, which of course can react with the iron gallotannate fluids that are applied along with the ink. Another problem arising from the excess of iron(ii) ions in gall inks from preparations that utilised too high a ratio of copperas to oak galls was their oxidation to ferric ions, creating a brown halo surrounding the inked regions.8 Analytical Raman spectroscopy can identify the presence of exogenous chemical components and residues present in treated vellums and the effects of acid degeneration of the skin proteins resulting from the gall ink additives, which itself can be an early warning for conservators of manuscripts and ancient documents. Examples of the definitive Raman spectral signals from iron gall inks will be provided in several case studies outlined later.
Later developments in ink manufacture occurred in the mid-to-late 19th century, when a new range of organic dye-based coloured inks became available following upon Perkin's synthesis of mauveine. At the London Society of Arts in 1858 the advent of this new range of brightly coloured dyestuffs was advocated first for ink manufacture with the additional adoption of graphite to replace soot for carbon-based inks.
It is true that whereas most ancient manuscript texts and writing have been accomplished using black ink, a rather limited range of other colours have been identified; examples include red inks (utilising cinnabar, dragon's blood, Brazil wood extracts for pigmentation), purple (using folium, caput mortuum and purpurum extracts for pigmentation) and gold. The major problem that was identified early on with the use of alternatives to carbon black or iron gall based inks is that unless the pigment was a mineral, such as cinnabar or in some much rarer cases haematite, extracts from botanical or marine sources were dyestuffs, which were badly affected by light and handling – such was the case for folium (also known as turnsole) and purpurum. The ancient scribes and artists termed these dyes “fugitive” and thereby recognised their impermanence17,18 and unsatisfactory adoption for scriptorial work.
Outside of conservation, a major problem that faces analytical scientists and art historians is the dating of inked manuscripts. Unlike historiated manuscripts, many ancient texts and maps often do not have any colouring pigments associated with them so the identification of the chemical composition of the ink could be an important analytical source of information. Radiocarbon dating procedures are usually not practicable for the dating of ancient inks because of the quantities of a specimen required in the destructive analysis, although of course, some useful ancillary information can result from the radiocarbon dating of the manuscript substrates themselves. Traditional methods of characterisation of ancient inks3,16 involve the application of techniques such as proton-induced X-ray emission (PIXE) spectroscopy, thin layer chromatography, ultra-violet visible spectroscopy, infrared spectroscopy, Raman microscopy, micro X-ray fluorescence spectroscopy, capillary electrophoresis, Rutherford back-scattering spectroscopy, scanning electron microscopy, scanning Auger spectroscopy and mass spectrometry.19,20 The major problem facing analysts arises from unrecorded variations in the ink composition and empirical formulations used over the centuries. The central analytical discrimination for the differentiation between iron gall inks and carbon-based inks is the recognition of elemental carbon signals for the former versus those of metallic iron for the latter. Secondary issues relate to the analytical identification of potential additives such as gum arabic, shellac and copper; in this respect, infrared spectroscopy and mass spectrometry provide valuable molecular information that other techniques do not. Thin layer chromatography has not been found to be particularly useful for the compositional discrimination of ancient inks.20 In contrast, the identification of elemental iron by SEM/EDAXS, micro-XRF and PIXE techniques is often used to indicate the presence of an iron gall ink in preference to a carbon-based ink from the detection of iron elemental signals. Iron gall inks possess an ultraviolet absorption with a sharp, strong band at 215 nm and a shoulder at 269 nm, characteristic of an iron gallotannate complex, compared with corresponding bands at 218 and 274 nm for the uncomplexed gallotannic acid. However, a word of caution should be advanced in adoption of this either/or concept regarding carbon-based inks and iron gallotannate inks as recent analytical studies have revealed that in some cases mixtures were used and in one example,21 mixtures of carbon black, iron gall and logwood-derived inks were used in the same manuscript. During the restoration of the 18th century manuscript of F.M. da Ponticelli entitled Nova Rhetorica, changes in the colour of the inks were noted from well-defined black to reddish-brown smudged portions on the same page: Raman spectral signatures of iron gall inks and also logwood inks to which had been added metal salts were indicated. The addition of metal salts, especially iron, conferred a darkening of colour on the reddish logwood ink, but also assisted the dispersion of the ink into the surrounding substrate, giving rise to a smudging effect. It has been suggested that this unusual procedure occurred as a result of a shortage of iron gallotannate inks in the scriptorium. Studies have been reported of logwood inks complexed with metals such as iron and chromium and also with aluminium from alum residues in the vellum, which compromise the preservation and colour of the resultant script.22,23
Raman spectroscopy has been used to detect amorphous and graphitic carbon and has been applied to the analysis of ancient carbons and carbon-based inks: a typical Raman spectrum of an ancient carbon derived from a botanical source gives two features (Figure 1.1), a broad band at 1320 cm−1 ascribed to sp3-hybridised carbon atoms (the D band) and a sharper band at 1580 cm−1 attributed to sp2-hybridised carbon atoms (the G band). It has already been mentioned above that an additional feature can occasionally be observed at 960 cm−1, which is the symmetric P–O stretching band of orthophosphate ions, characteristic of the presence of calcium hydroxyapatite, a basic calcium orthophosphate, which is indicative of a source for carbon that has been derived from the thermal treatment of animal bones (usually cow or ox bones) or ivory. During the low temperature combustion process in a furnace or kiln with a limited supply of air, the organic matter is converted to amorphous carbon, leaving the inorganic phosphatic matrix: at higher temperatures the carbon is burnt off to leave a white deposit of predominantly calcium triphosphate, Ca3(PO4)2, which is used as “bone ash” after fine grinding in the production of bone china and phosphatic porcelains.24 Several scribes and artists professed favour towards ivory black or bone black, as these products were called, because of their greater depth and intensity of colour. Examples can be found of carbon from animal and vegetable sources that have been identified by Raman spectroscopy in ancient manuscripts, whether in the inks or in the black pigments in historiated texts.
In contrast with the well-established Raman spectroscopic signatures that have been defined for carbon-based inks, those of iron-gall inks have been correspondingly much more difficult to attain, it is assumed because of their complexity of composition and the presence of potentially fluorescent additives, which render the detection of the weaker intensity bands definitive of the Fe-gallotannate complexation to be recognised.25,26 Hence, a survey of the previous literature reveals that the presence of iron gall inks is inferred by the absence of carbon signatures, as defined above, and vice versa. Generally, too, unless special care is taken with the choice of laser excitation, the onset of background spectral emission from the inked areas on a potentially degraded manuscript with visible laser wavelengths can result in rather broad ill-defined carbon signatures situated on a larger than expected broadband emission background, so making these carbon signatures rather difficult to observe anyway! It is dangerous to assume that because carbon signals are noted from a manuscript ink, the absence of an iron gall component may be inferred. As will be seen in an example given below, the spectrum of an ancient manuscript demonstrates the presence of a mixture of carbon and iron gall, and iron gall and logwood extracts in the same inked script!
As an illustration of the information that can be deduced from the ink analysis of ancient manuscripts and works of art, three case studies will now be presented here: The Vinland Map. The Voynich Manuscript. The Beato de Valcavado manuscript.
The Vinland Map and the Voynich Manuscript are two important and highly controversial manuscripts purporting to be from the 15th century. One of these, the Vinland Map, possesses only black ink on two sheets of vellum,5,27 whereas the other, the Voynich Manuscript, is polychrome and comprises 240 sheets of vellum, but again, with only black ink script.28,29 Both manuscripts were discovered in the early-to-mid 20th century with limited historical provenances and have since been the subject of some intense analytical, eschatological and historical research, from which diverse conclusions have been forthcoming. The Beato de Valcavado manuscript is of a Visigothic 10th century origin and, unlike the previous two examples, has an impeccable provenance but still has some surprises in store for the pigments analyst and ink specialist.30 These three examples will now be discussed in more detail.
The Vinland Map is a vellum map of the Old World (Figure 1.2), measuring approximately 28 × 40 cm, that identifies an area in the Western Atlantic, Vinilanda Insula, the so-called Vinland of Scandinavian Viking folklore, showing an area to the north-east of North America, described thereon as “a new and fertile land to the west”. It first appeared in 1957 bound together with a manuscript known as the Tartar Relation (the Historia Tartarorum), which had an unusual composition of vellum and paper, two leaves of vellum alternating with six of paper, the latter exhibiting bulls' head watermarks, which were identifiable with the Basle Council of the 1430 s. The cartographic importance of the Vinland Map centred upon its pre-dating the voyage of Christopher Columbus and his epic discovery of the New World in 1492 by approximately 60 years and fed the rumour that Columbus could actually have used such a map based on earlier Viking seafaring exploits, notably those of Leif Ericsson.5
In 1957, the Vinland Map and Tartar Relation were examined by British Museum experts in ancient maps and incunabula.27 The philanthropist Paul Mellon donated them to the Beinecke Rare Book and Manuscript Library in Yale University and in 1972 the Yale University Library commissioned an analysis of the Vinland Map, thereby initiating a controversy which has raged for over 45 years and is still ongoing. The scientific analysis of the Vinland Map opened with a detailed polarised microscopic and X-ray diffraction (XRD) analysis of the inked areas,31 from which Walter McCrone concluded that the presence of anatase in the ink, a polymorphic form of titanium(ii) oxide, dated the map firmly to the 20th century32 indicating the Vinland Map could be a fake.33 Further analyses using scanning electron microscopy and energy-dispersive X-ray (EDX) analysis34,35 determined that the ink was not an iron gallotannate as found on the associated Tartar Relation but was carbon black. The ink and parchment of the Vinland Map were re-analysed36 using proton-induced X-ray emission (PIXE) spectroscopy and yielded results that challenged the conclusion of McCrone.13 Radiocarbon dating of the vellum37 gave a date of 1432 ± 11 ACE, which agreed to within one standard deviation with the watermarked date on the associated paper in the Tartar Relation.
In 2002, a definitive Raman microspectroscopic study by Brown and Clark38 of the Vinland Map was undertaken with 632.8 nm laser excitation. The inked areas were apparently composed of two parts: a yellow line that was strongly adherent to the parchment substrate and an overlaid black line, which showed evidence of severe loss in parts due to the black pigment “flaking off”. Analysis of the black ink gave the characteristic D and G bands for amorphous carbon at 1325 and 1580 cm−1, respectively. The presence of anatase was evident, with characteristic bands at 142 and 398 cm−1 and seems to be indicative of a clever forgery; however, care must be taken as anatase has actually been identified in genuine archaeological artefacts significantly pre-dating its established synthesis in the 20th century.39,40
In 1912, Wilfrid Voynich revealed the discovery of his eponymous manuscript which apparently dates to the early 15th century: the 240 vellum pages with black script and historiated polychrome figures, now labelled the “World's most mysterious manuscript”, resembles a herbal compilation of botanical medicine and alchemy. An example is shown in Figure 1.3. Believed to be written either in a secret code or in a lost language, the major problem with this manuscript is that it has thus far defied all attempts at translation. The historical provenance of the Voynich Manuscript, like that of the Vinland Map, is steeped in international intrigue.28,29 A recent comprehensive and elegant analysis of the iron gall ink on the Voynich manuscript has been undertaken41 using polarised light microscopy, SEM/EDAXS spectrometry, micro XRD and IR spectroscopy.
Unlike the Vinland Map, the Voynich Manuscript is historiated and has the advantage over other inked documents in having the presence of supporting analytical information resulting from the interrogation of the associated pigments to assist in the chronology,42–48 which seems to be scientifically well established and supportive of an assignment of the manuscript to the mediaeval period.
Towards the end of the 8th century, in 776 CE, a monk called Liebana prepared an illuminated manuscript on the Apocalypse of St. John in the monastery of Santo Martin. Although this original is now lost, it generated several commentary texts over the next two or three centuries addressing heretical disputes and the coming of the first Millennium. Some 32 of these so-called Beato manuscripts are known, and the most complete is the Visigothic Beato de Valcavado (the Codex de Valladolid) now in the Biblioteca del Colegio de Santa Cruz, Valladolid, in Castile y Leon, Spain, which was completed in 970 and comprises 230 sheets of vellum with 87 historiated miniatures.49 A further five sheets reside in the National Library of Spain, detailing the genealogy of Christ. This Beato is unique among the contemporary Beato copies in that it contains the signature of its author and the dates of its commencement and completion, namely the 8th June and the 9th of September, 970; in addition to its very high quality, it is unique eschatologically in that it is identifiable as the work of the single scribe, Obeco. A Raman spectroscopic study of this Beato de Valcavado was undertaken30 and several important pigments were identified from the historiated scenes: of particular relevance here, is that iron gall ink has been used not only for the script but as a black pigment in the historiated scenes; not carbon black as suspected. In another specimen from a comparative 10th century Beato from the Najera monastery, now in Santo Domingo de Silos, iron gall ink was found in the admixture with the bright red mineral cinnabar, mercury(ii) sulfide, to produce a dark red colour. Later manuscripts from the collection50 in the Santo Domingo Monastery in Silos dating from 17th century exhibited mixtures of iron gall ink and carbon black in their black pigmentation.
An interesting Raman spectroscopic study of carbon black inks on several of these manuscripts held in the same collection as the Beato de Valcavado manuscript in the Monastery of Santo Domingo de Silos shows, firstly, the confirmation of their assignment as carbon-based inks and, secondly, information about the possible sourcing of the carbon from which the inks were made. All spectra show the characteristically broad D and G bands of amorphous carbon peaking around 1320 and 1590 cm−1. A spectral stack plot of four manuscripts from the Santo Domingo de Silos collection, comprising AMS2 from the 17th century, AMS4 from the 17th and 18th Centuries, AMS 6 from the 17th century and AMS10 from the 18th century. The spectra of AMS 2 and AMS4 have D and G bands at 1318 ± 2 and 1591 cm−1, indicative of a bone or ivory black, but the absence of the phosphate band at 960 should be noted, ascribed to the excitation wavelength of 785 nm used here.51 AMS6 has a pronounced high wavenumber shoulder at 1608 cm−1 on the G band centred at 1578 cm−1 and a lower D band centred at 1308 cm−1, indicative of a charcoal produced using a resinous material, and a higher relative band intensity for the G band over the D band. Finally, AMS10 shows a G band at 1601 cm−1 and a D band at 1316 cm−1 with similar relative intensity to that for AMS2, indicating a soot-based carbon source based on “black chalk”.
In contrast, the Raman spectrum of iron gall ink from the manuscript collection and the 10th century Beato de Valcavado is shown in Figure 1.4: characteristic Raman signature bands of iron gall ink are found at 1577 mw broad, 1480 s broad, 1424 mw shoulder, 1341 m broad, 937 w, 774 w, 705 w, 614/550 m broad, 402 w, all in cm−1. The band at 1341 varies in wavenumber25,26,30 between 1350 and 1310, whilst that at 614/550 can vary between 640 and 490 cm−1.
The identification of and analytical discrimination between ancient inks and the recognition of the corrosive nature of iron gallotannate inks is now well documented,52,53 but the dating of specimen manuscripts with any precision is still generally not feasible solely by examination of the ink composition alone. The consideration of associated information from radiocarbon dating of the manuscript substrates, eschatology, lexicography and chronological placement of historiated pigments materially assists in the verification of the antiquity of a manuscript. Analytical spectroscopy has demonstrated the novel presence of iron gall inks in an admixture with other inks such as carbon black and logwoods often in the same manuscript, reinforcing the tenet that it is not simply an either/or choice but perhaps reflects the availability of the inks available to the scribes at that time and their experience of temporary shortages. Finally, the use of iron gall ink as a black pigment for the historiation of written scripts alongside the more usual mediaeval pigments and even in an admixture with them is also recorded analytically.
© The Royal Society of Chemistry 2019 (2018)