In 1906, the Danish classical scholar Johan Ludvig Heiberg examined a Greek book of Christian prayers in Constantinople and noticed older writing running across its pages. Beneath the prayers lay mathematical diagrams and passages copied in the tenth century from Archimedes, the Greek mathematician who had worked in Syracuse more than a millennium earlier. The older manuscript had been taken apart, its text erased and its parchment reused for a prayer book completed in 1229.

The attempted erasure made Archimedes difficult to read, but it did not remove every trace. Some of the writings preserved under the prayers have no other surviving manuscript witness. The new book travelled through monastic and private collections long after the earlier volume had ceased to exist as a separate object. Reusing the parchment destroyed the old volume as a book, yet left enough of the writing for later scholars to recover.

No single group of monks decided what should survive of Greek and Roman writing. Across the Middle Ages, scribes made fresh copies of some books, annotated or repaired others and dismantled volumes whose parchment had become more useful than their contents. Their choices depended on local needs, languages, resources and readers. What happened in seventh-century Italy might differ sharply from ninth-century Francia or thirteenth-century Jerusalem. Sometimes the physical books still show us both what a community chose to write and what it had tried to erase.

What Happened When a Scribe Reused a Manuscript?

A manuscript written over an erased earlier text is called a palimpsest, from Greek words relating to scraping a surface again. The material in many surviving examples is parchment, prepared from animal skin. It is tougher than ordinary paper and can sometimes be washed or scraped so that the surface can be used for new writing. Scholars call the older layer the undertext and the newer writing the overtext. One piece of parchment can therefore carry texts written centuries apart.

To make a palimpsest, a book worker might remove the old binding and separate the folded sheets. A book worker might wash the ink, abrade the parchment and smooth it before cutting or folding the sheets into a new arrangement. The second book could be smaller, differently ruled and intended for an entirely different use. Some leaves were cut down so severely that part of the old writing disappeared altogether. The Vatican Library’s study of methods of erasing manuscripts shows that washing and abrasion could leave very different amounts of the earlier surface intact.

Washing was not always effective enough to remove the text completely. Ink could enter the skin, or writing could leave changes in its texture that remained after the visible pigment faded. Reused pages were also sometimes rotated by ninety degrees, reducing the interference between lines of old and new text. Vatican researchers describe this as a practical solution to legibility. The crossing of two scripts need not have had a symbolic or religious purpose.

Ancient pagan literature was only one kind of material recycled in this way. A damaged Christian book, an outdated legal volume or an obsolete service text might also be reused. In later periods, discarded parchment found new jobs as covers, pastedowns or supports inside other bindings. Hannah Ryley’s study of medieval book recycling examines the persistence of this practice. Even after a text fell out of use, its parchment could be valuable for making or repairing another book.

Why Parchment Was Worth Reusing

Producing new parchment took resources and skilled work. Animal skin was treated using water, lime or other materials, then stretched on a frame while the worker removed flesh and hair and adjusted the tension during drying. Only afterwards could the sheet be cut, folded and ruled for writing. A substantial book could require skins from many animals and weeks or months of copying. Any precise figure for the number of skins in a ‘typical’ medieval Bible would conceal differences in page size, animal size and book design.

Communities did not have equal access to those supplies. A wealthy monastery with livestock and trained book workers could sometimes prepare its own materials. A smaller house might buy parchment, receive a finished book as a gift or borrow another manuscript to serve as a model. Reusing old leaves spared the community some of the expense of acquiring fresh skins and made use of material that had already endured long periods of transport and handling.

Copying took time and skilled labour as well as parchment. Scribes needed ink, pens, light, an exemplar to work from and the training to read and reproduce it. Even conscientious copyists could omit a line, duplicate words, misunderstand letter shapes or make corrections that changed the received text. Marginal notes might accumulate over generations. Each new copy gave a work another opportunity to survive, but it also created the possibility of transmitting fresh mistakes.

Parchment was valuable, although there is no reliable foundation for the often repeated claim that it ranked immediately after gold and silk among medieval commodities. Nor does a recycled manuscript automatically prove that its makers were impoverished or cut off from supplies by warfare. It tells us that they considered the material worth preparing again. To explain why, historians need to consider where and when the new book was made and what was written in place of the old text.

Advertisements

Which Ancient Books Were Chosen for Erasure?

What scribes erased depended partly on what they expected their community to read. Copies of the Psalms, Gospels and instructions for worship were useful in daily monastic life. An incomplete legal book, an old version of a text or a manuscript written in an unfamiliar hand might no longer have an obvious purpose. One ancient author could remain part of the curriculum in one house while an old copy of the same author’s writing was dismantled somewhere else.

Christian priorities certainly affected copying. Communities needed scripture, commentary and books for the education of clergy. The manuscript evidence, however, does not support a rule that monks kept Christian writing and erased everything pagan. Greek and Roman authors were taught for their language and rhetoric, and Christian works could also become undertexts. The Vatican Library records Latin palimpsests containing both classical and Christian undertexts. These manuscripts document particular acts of reuse, not a universal policy of cultural destruction.

The language and script of a book also affected its usefulness. In the Greek-speaking eastern Mediterranean, ancient Greek works remained accessible to educated readers and continued to be copied. Latin dominated learning and worship across much of western Europe. A Greek manuscript might therefore be unreadable to a western community even while Greek readers elsewhere valued it. Changes in script and methods of teaching could make old volumes harder to use even when the language itself was familiar.

Virgil shows how readily a pagan author could remain part of medieval education. His Aeneid recounts the journeys of the Trojan Aeneas and connects him to Rome’s legendary origins. A British Library manuscript of the poem, probably used in teaching around 1175–1224, bears corrections and marginal comments accumulated over several centuries. Its distinctive format was suited to a teacher’s work. The continuing use of this Roman epic in classrooms is as much part of medieval literary history as the recycling of less frequently read ancient books.

Many Greek works also survived through scholarly traditions outside European Christian monasteries. Syriac- and Arabic-speaking scholars translated, studied and transmitted mathematics, medicine and astronomy, with especially important activity under the Abbasids. Richard Lorch’s study of Greek, Arabic and Latin mathematical transmission follows movements of mathematical writings between languages through the medieval period. A Greek text might disappear while an Arabic translation survived, or its different versions might preserve different terminology and readings. The history of ancient science cannot be explained by Latin monastic copying alone.

The fate of ancient books also depended on changes outside the communities that copied them. Ancient urban book markets declined, political boundaries changed and libraries passed into new hands or were dispersed. Copies required patrons or institutions willing to pay for them. Destruction could occur through fire, conflict, neglected storage, later rebuilding, sale or even the recycling of bookbindings. The number of extant palimpsests cannot be converted into a reliable percentage of ancient literature destroyed during a particular medieval century.

Cicero Beneath a Book of Psalms

A surviving copy of Cicero’s writings shows what reuse could mean in a Latin monastery. In the seventh century, probably at Bobbio in northern Italy, workers washed writing from part of a fourth- or fifth-century copy of De re publica, Cicero’s dialogue on government and political responsibility. They then wrote part of Augustine’s Expositions on the Psalms onto the parchment. Both texts were substantial works in Latin. This was a change of books suited to different needs, rather than the replacement of an unintelligible foreign language.

The surviving manuscript is catalogued as Vatican Latin 5757. The Vatican Library’s examination of its two texts identifies extensive portions of Cicero’s first two books, with smaller sections of later books, still present beneath Augustine’s commentary. A monastic house whose worship centred on the Psalms had obvious uses for Augustine’s interpretation. We do not know whether the discarded Cicero codex was complete when it was recycled, or precisely what its makers thought of it. The rest of that earlier volume is unaccounted for.

Only in 1819 did Angelo Mai recognize Cicero’s writing beneath the newer text. His discovery recovered large parts of a dialogue that had been known mainly through excerpts and quotations. The old volume had plainly not been kept intact for posterity, yet its remaining ink had preserved words that would otherwise have been lost. Cicero’s volume had been dismantled, but part of its writing remained in the parchment.

Nor should we assume that Cicero himself had been rejected. Medieval Latin readers continued to study and copy many of his works on philosophy and rhetoric. The Bobbio palimpsest records the fate of one copy of one dialogue, not the attitude of all monasteries toward his writings. Elsewhere, other Ciceronian books were being preserved and reproduced.

Advertisements

The Archimedes Book Reused in Jerusalem

The Archimedes palimpsest began with a Greek manuscript produced in tenth-century Constantinople. It contained mathematics attributed to Archimedes, who had lived in Syracuse in the third century BCE and whose works reached the copyists through many generations of transmission. Geometry, methods of measurement and the behaviour of bodies floating in liquids were among the subjects represented. By the time that manuscript was made, Archimedes himself had been dead for well over a thousand years.

The old book was dismantled for a prayer collection completed in 1229 by the scribe Johannes Myronas, probably in Jerusalem. This new Greek euchologion contained leaves taken from several earlier manuscripts. Pages from the Archimedes volume were cut, rearranged and refolded to fit the smaller prayer book. The Archimedes Palimpsest Project’s account of the book identifies seven treatises by Archimedes in its lower writing, alongside works by other ancient and Christian authors.

Among the mathematical works are The Method of Mechanical Theorems, in which Archimedes uses physical reasoning to help discover geometrical conclusions, and a surviving portion of the Stomachion, associated with a geometrical dissection puzzle. The palimpsest is also our only surviving Greek manuscript witness to On Floating Bodies, though the work was transmitted in other forms. It is misleading to suggest that every treatise was wholly unknown before its recovery, or that the collection simply amounts to the discovery of modern calculus ahead of its time. Its importance lies in the arguments and textual versions preserved by these particular leaves.

Archimedes was not the only author concealed under the prayers. Some leaves preserved speeches by Hyperides, the fourth-century BCE Athenian orator, and others carried commentary on Aristotle’s Categories. The Walters Art Museum’s account of the conservation describes the identification of such writings. A single medieval prayer book had absorbed parchment from a number of quite different older volumes.

The prayer book’s later movements were complicated. It was documented in the Metochion of the Holy Sepulchre collection in Constantinople by the late nineteenth century. A catalogue transcription drew Heiberg’s attention to the mathematics, leading to his visit and photographs in 1906. Later the manuscript entered a private collection in France, and some pages suffered damage or were covered with paintings in the twentieth century. Sold at auction in 1998, it was then deposited at the Walters in Baltimore for investigation and treatment. The project’s manuscript history follows this sequence while acknowledging gaps in what can be established.

How Scientists Read the Writing Beneath the Prayers

Heiberg’s photographs made some erased passages readable, but many lines remained hidden by overlapping writing, damage and later paint. At the Walters, conservators worked alongside imaging specialists to recover more information without needlessly harming the leaves. Multispectral photography records the parchment under different wavelengths of light. Processing those images can make older lettering stand out against the newer script and the surrounding material.

This work is more exacting than adjusting the brightness of a photograph. Ink from two writing campaigns may respond in similar ways, while stains, wrinkles and natural variations in the skin complicate the results. Specialists align and compare images produced under different conditions before deciding which marks belong to which writing. Even then, scholars must read the Greek, recognize the handwriting and judge damaged passages. An enhanced image is evidence for an edition, not an automatically trustworthy transcription.

Some folios presented a different problem: modern painted pictures concealed the old text. X-ray fluorescence allowed researchers to map elements, including iron associated with underlying ink, through layers that visible-light photography could not penetrate. The Walters’ technical report on X-ray imaging describes this approach and its results. It can reveal otherwise hidden writing, but the signal does not make every letter equally clear, and it cannot promise a complete recovery of all erased passages.

Archimedes Palimpsest page showing barely visible erased text under normal light compared to clearly revealed text under ultraviolet illumination
Archimedes Palimpsest photographed in visible light (top) and ultraviolet illumination (bottom), revealing erased Greek mathematical text. Source: Knox et al., Image Restoration research.

The team also published imaging and transcription material through the digital manuscript project. Making the evidence available allows other researchers to compare readings and review what different imaging methods reveal. In a mathematical manuscript, the position of a line in a diagram can change how a proof is understood. Digital access preserves a record of the researchers’ observations as well as their reading of the ancient words.

Research has produced discoveries even very recently. In March 2026, France’s national research centre, CNRS, announced that Victor Gysembergh had recognized a missing Archimedes leaf in the Musée des Beaux-Arts at Blois by comparing it with Heiberg’s old photographs. One face contains propositions 39–41 of Book I of On the Sphere and the Cylinder; a later painted image covers the other. The CNRS announcement of 9 March 2026 presents the identification as established but describes further imaging of the covered side as work still to be undertaken. That hidden writing has not yet been read simply because the leaf has been identified.

Advertisements

What the Sinai Palimpsests Preserve

The Archimedes manuscript is one example of a wider eastern Mediterranean practice of reusing parchment. Saint Catherine’s Monastery at Sinai holds a particularly extensive collection. Older and newer scripts there include Greek, Syriac, Arabic, Georgian and other languages used in communities connected with the region. One piece of parchment can therefore carry evidence for the movement of writing across languages, places and centuries.

The Sinai Palimpsests Project identifies more than 160 known palimpsest manuscripts at the monastery. Its researchers combine spectral imaging with the expertise needed to decipher unfamiliar scripts and languages. Some recovered passages come from Christian scripture or theology, while others illuminate literary, historical and linguistic traditions. Their range again shows why palimpsests cannot be treated simply as a record of monks erasing pagan classics.

Especially important are undertexts written in Caucasian Albanian, the language associated with an early medieval kingdom in the eastern Caucasus. Very little of its written record otherwise survives, making fragments from Sinai unusually valuable. The project also works on early Syriac gospel texts and material in Christian Palestinian Aramaic. Even a relatively small recovered passage can greatly enlarge the evidence for a language or the people who spoke and wrote it.

Such leaves have more than one geographical and chronological history. The earlier work might have been copied in one region, carried into another and recycled somewhere else generations later. The writing placed over it belongs to a different moment and sometimes a different community. Identifying scripts, scribal hands and related texts helps establish those movements. The different writing layers can thus reveal where books travelled, changed hands and were taken apart.

Why Monasteries Copied Ancient Literature

Erasing old volumes was only part of what monks did with books. Monastic libraries also held works for repeated reading, repair, lending and copying. In the sixth century, the rule associated with Benedict of Nursia set aside time for reading within the community’s routine. Its chapter on daily work and reading directs monks to receive a book during Lent. Benedict did not ask his followers to collect every ancient author, but regular reading became an explicit part of monastic discipline.

In sixth-century Italy, Cassiodorus also connected book production with learning. Formerly an official in the Ostrogothic kingdom in Italy, he withdrew from government during the sixth century and founded the community of Vivarium. His Institutiones guided readers through sacred and secular subjects. In Book I, chapter 30, he praised scribes and advised them to check older authorities on correct spelling so that mistakes would not damage the words they copied. Classical grammatical learning had a practical place within his Christian educational programme.

The contents of a monastic library usually followed the tasks of its readers. Scripture, commentaries, saints’ lives and service books supported worship and instruction. Histories, poetry and Latin works of grammar or rhetoric could help educate people who needed to interpret difficult passages or write well themselves. An ancient work survived when generations of readers found enough reason to make and keep copies, whether for Christian learning, literary education or another purpose.

Some communities devoted considerable labour to making and decorating books. Around 800, scribes and artists in the Insular Christian tradition made the Book of Kells, a Latin Gospel manuscript with elaborate decoration and exceptionally intricate lettering. It is a Christian work rather than an example of classical literature, but it shows how much labour could go into a book that a community valued. Not every older work commanded the same resources, and volumes regarded as less useful could consequently go uncopied.

Book production also extended far beyond the early medieval monastery. A fifteenth-century image of the Burgundian author and translator Jean Miélot represents a later, professional world of manuscript making. Courts commissioned books alongside churches and monasteries, and copying, translation and illumination supported a broader market for written works.

Medieval scribe Jean Mielot working at desk surrounded by manuscripts books and writing materials in monastery scriptorium
Jean Mielot in his scriptorium, from Miracles de Notre Dame, after 1456. Source: Bibliotheque nationale de France

Cathedral schools, bishops’ households, royal courts, commercial copyists and private owners all played roles in preserving writing. The British Library and Bibliothèque nationale de France’s medieval manuscript project encompasses classical, scientific, legal and historical books as well as devotional ones. Ancient writings circulated through these collections as well as through monastic libraries.

Charlemagne and the Expansion of Book Copying

In the later eighth and ninth centuries, political support helped expand book copying across western Europe. Charlemagne and the other Carolingian rulers promoted reforms in education and Christian worship, bringing scholars such as Alcuin, who had been educated at York, into their service. Schools and monasteries worked to produce more accurate Latin texts, improve the training of clergy and distribute books. The increased copying also benefited secular works inherited from antiquity.

One result was the spread of Caroline minuscule, a relatively regular Latin script with letter forms that readers in different regions could more easily recognize. A study of Carolingian book production in the New Cambridge Medieval History estimates that about 7,000 manuscripts made between roughly 750 and 900 still survive, against approximately 500 from earlier Merovingian Gaul. Neither number represents all the books originally made. Together, they nevertheless indicate the remarkable expansion visible in the surviving record.

Important Latin books were copied in places such as Tours, Corbie and St Gall. Once a work existed in several manuscripts, readers beyond its original home could use it, and scholars had more than one witness with which to identify errors and differences in wording. The reforms had religious and educational aims, but the expanded copying networks also helped ancient works circulate.

Charlemagne surrounded by court officers receiving scholar Alcuin presenting manuscripts copied by monks in painting
Charlemagne receives Alcuin presenting manuscripts, painted by Jean-Victor Schnetz, 1830. Source: Musee du Louvre, Paris.

Increased production never removed the possibility of loss. Old volumes still wore out, disappeared or were reused, and some ancient authors had vanished long before Carolingian copyists could have encountered them. Conflict and attacks on religious houses caused further damage, but there is no reliable evidence for a universal renewed surge of palimpsest-making to some earlier level after the Carolingian empire divided. The fortunes of actual texts and libraries have to be traced individually.

The Treaty of Verdun in 843 divided the Carolingian realm among Charlemagne’s grandsons and changed the political world in which schools and monasteries operated. Copying did not cease at the treaty: it continued under different rulers and in different religious communities. For historians, that persistence is another reason to follow networks of books and readers rather than imagine that a single imperial centre directed literary preservation.

Map showing division of Carolingian Empire into three kingdoms after Treaty of Verdun in 843 CE
Territorial divisions of the Carolingian Empire following the Treaty of Verdun, 843 CE. Source: Wikimedia Commons

The Carolingian script reforms had consequences well beyond the Middle Ages. Renaissance humanists later admired the appearance of Caroline letters, in some cases mistaking them for the writing of ancient Rome. Their adaptations contributed to the lowercase forms used in early printed books. The drive to improve the legibility of manuscripts therefore influenced both which Latin texts endured and the shapes of letters familiar to readers today.

Byzantine Scholars and the Greek Classics

The Greek-speaking Byzantine world maintained different institutions and patterns of transmission. Ancient Greek remained important in the learning of Constantinople and elsewhere, with Homer, philosophy and rhetoric taught and copied alongside a great deal of Christian writing. Books moved through monasteries, schools, official households, imperial circles and private collections. Teachers and lay scholars shared in their preservation with monks; there was no single Byzantine institution responsible for keeping the classics alive.

Greek book production also changed during the late eighth and ninth centuries, when compact minuscule hands increasingly replaced older majuscule forms. The new writing belonged to shifting practices of education, administration and manuscript copying. Filippo Ronconi’s study of the adoption of Greek minuscule connects the change with administrative elites and monastic networks, rather than attributing it to one centrally organized programme.

Recopying an old text in a new script helped keep it usable, but it was never a completely neutral act. Scribes made choices about spelling, abbreviations, page design and marginal notes, and mistakes could enter the text. If an older exemplar was then discarded, its particular readings might be lost even when the work survived. To recover an ancient author’s wording, modern editors often have to compare manuscripts separated from the original by several generations of copying.

Byzantine illuminated manuscript page showing David and Goliath with gold background and Greek text from Paris Psalter
David and Goliath scene from the Paris Psalter, 10th century Byzantine manuscript. Source: Bibliotheque nationale de France

Byzantine religious houses sometimes copied explicitly secular literature. The Cambridge Intellectual History of Byzantium describes a tenth-century monk called Ephraim who copied Aristotle’s Organon and Polybius’ Histories as well as a Christian text. That example is difficult to reconcile with the claim that monks worked only on theology. The same scholarship discusses books owned by later Byzantine officials and intellectuals, including libraries eventually placed in monastic care.

Catastrophes also changed the fate of Greek books. The sack of Constantinople in 1204 and the Ottoman conquest in 1453 disrupted institutions and scattered collections, but neither event supplies a measurable total for the works lost. Greek manuscripts travelled to Italy through contacts and scholarly migration before and after 1453. Their survival emerged from many movements of books over centuries, rather than one rescue from a single doomed library.

What the Surviving Books Can and Cannot Tell Us

The works we now call the Greek and Roman classics are only a fraction of what ancient people wrote. Some disappeared before medieval scribes ever handled them, some went out of circulation when copying stopped, and others are known only from later quotations or papyri found by archaeologists. Medieval manuscripts remain indispensable to the study of classical literature, but their survival cannot reveal an exact proportion of writings lost in any particular century.

Records of particular libraries reveal more about book disposal than general claims of cultural destruction. Krista Milne’s 2025 study of book destruction in medieval England uses inventories, institutional documents and recycled fragments to show how communities removed books as practices and priorities changed. Medieval libraries were working collections that gained and lost volumes, not permanent vaults that kept every book once placed on a shelf.

The Archimedes and Cicero palimpsests each preserve writing from a book that was dismantled for reuse. In both cases, later readers obtained parchment by dismantling an older work. The books ceased to exist in their former shape, but traces of their writing survived long enough to be found. No imaging method can do the same for an ancient work that has left neither manuscript, fragment nor quotation. In those cases, the loss may result from many decisions and accidents rather than one identifiable act of erasure.

Medieval scribes chose books according to the needs and possibilities of their own times. Some works they copied became the main route through which ancient literature reached modern readers; others they recycled, neglected or left to disappear. The resulting record includes clear medieval copies, quotations in later writings and barely visible words under prayers. To understand what survived of antiquity, we have to consider all those paths through which a text could be kept, changed or lost.