The ink used on a historical manuscript may appear to be an insignificant detail compared with the text it carries. For the archaeologist, the conservator, and the historian of material culture, however, ink composition is a detailed record of where, when, and how a document was produced. Different inks use different recipes, and the recipes differ systematically by region and period. Matching the ink of an undated document to a known ink history can often place it within a few decades and a few hundred kilometres of its origin, and the techniques required to do so have become an established part of heritage science over the last thirty years.

The most widely used ink in medieval and early modern Europe was iron-gall ink, produced by combining an extract of oak galls (swellings on oak trees caused by wasp larvae) with a solution of iron sulphate. The mixture produced a soluble compound that was easy to write with, but when applied to paper or parchment the compound slowly oxidised and became insoluble, forming the characteristic black mark that characterises iron-gall writing. Iron-gall ink was cheap, readily available, and remarkably resistant to fading, but it has one significant drawback: the same iron ions that give the ink its colour can, over time, catalyse the degradation of the paper or parchment, producing a pattern of browning and, in extreme cases, complete loss of the written area. Many medieval manuscripts show various stages of iron-gall damage, and conservators have developed specific treatments - often involving a controlled application of antioxidant compounds - to slow the process.

Carbon ink, the older alternative, consists of carbon black suspended in a gum binder. Carbon ink does not damage its support in the way that iron-gall can, because carbon is chemically inert. It was used throughout the ancient Mediterranean and East Asia, and it remained common in some contexts - including Chinese calligraphy and Korean woodblock printing - into the twentieth century. Carbon ink's main limitations are that it can be removed relatively easily by water and that it does not bond as tightly to parchment as iron-gall, so carbon-ink manuscripts can show fading or smearing where iron-gall manuscripts would not.

The Dutch conservation scientist Liesbet Koopmans has argued that ink analysis is most valuable when several techniques are used in combination. X-ray fluorescence detects the elements present in the ink, distinguishing iron-gall from carbon at a glance and often identifying trace elements that can narrow the recipe further. Raman spectroscopy identifies the molecular compounds, useful for distinguishing between different types of iron-gall ink and between different pigments that might be added for colour effects. Microspectroscopy of tiny samples can identify the binder used. The combination of these techniques can produce a detailed chemical portrait of an ink with almost no damage to the document.

Archaeological applications have been striking. A disputed letter supposedly written by a fifteenth-century Florentine merchant was shown by ink analysis to be a nineteenth-century forgery: the iron-gall recipe used contained a chemical signature characteristic of a particular early industrial process that did not exist in the fifteenth century. A medieval manuscript whose precise origin was uncertain could be placed at a specific monastery by matching its ink to inks used in other documents definitely produced there. The reconstruction of the so-called Vinland Map - a controversial document claiming to show Norse knowledge of North America before Columbus - has depended extensively on analysis of its ink, with continuing debates about whether the chemical evidence supports or disproves its authenticity.

Colour inks introduce additional complexity. Red ink, used for decorative initials and rubrication (the marking of sections in red for easier navigation), was typically based on vermilion or red lead, both of which can be identified chemically. Blue inks used in later manuscripts came from a variety of sources, including lapis lazuli, azurite, and synthetic prussian blue; each has a distinct signature. The choice of colour ink was sometimes governed by local availability and sometimes by cost, and variations in ink colour within a manuscript can indicate either that the work was produced over time, as ink supplies changed, or that it was produced by several scribes or painters using different materials.

The ink itself has sometimes preserved texts that the original writing did not. In a famous case from 2012, a previously undocumented poem by Archimedes was recovered from a thirteenth-century Byzantine prayer book under which the older text had been scraped and rewritten. The iron traces of the original ink, invisible to the naked eye, could be mapped by X-ray fluorescence imaging, and the resulting map reconstructed the erased text. Koopmans's laboratory has worked on several similar palimpsests - manuscripts whose original text has been overwritten - and has shown that careful imaging can often recover substantial portions of the erased content, sometimes including mathematical and scientific material unknown from any other source.

A recent trend in the field is the construction of reference databases of ink signatures. Each characterised ink recipe, tied to a documented source, contributes to a library against which unknown inks can be compared. Koopmans's group has contributed a database of Dutch and Flemish inks from the fifteenth to the seventeenth centuries, with more than a thousand samples from securely dated documents. The database has been used to date several undated works at national archives, and she has argued that similar regional databases for other parts of Europe would substantially improve the precision of ink-based dating. The construction of such databases is cumulative work, requiring the analysis of many documents whose date and provenance are already known. It has been slow, but the contribution to understanding of historical production is significant enough to justify the investment.