Recovering Hidden Witnesses in the Kurzgefasste Liste: The Possibilities of MSI

This is a joint post with Dr. Denis Salgado, who works at CSNTM and has agreed to share about MSI technology.

A sensational story recently made headlines: researchers at the University of Glasgow have recovered the text of 42 “lost” pages from one of the most important early Greek New Testament manuscripts— Codex H (GA 015), a 6th-century copy of Paul’s letters.

What makes this discovery remarkable is not that the physical pages were suddenly found — they have long since disappeared. Rather, a team led by Garrick Allen, in collaboration with the Early Manuscripts Electronic Library (EMEL), used multispectral imaging (MSI) to reconstruct the missing text from faint ink traces left. These traces had been left on facing pages centuries ago when the faded writing was re-inked. The codex was later disassembled at Great Lavra Monastery on Mount Athos, and its leaves were reused in the bindings and flyleaves of other books. These reconstructed pages survived only as residual ink traces — until now.¹

You can see the images here. This is also uncharted territory for the Kurzgefasste Liste: it will be very interesting to see how these reconstructed “ghost” pages will be catalogued therein.

This breakthrough has generated renewed excitement about the potential of MSI for studying early Greek New Testament manuscripts. GA 015 belongs to a larger and more complex history of manuscripts being dismantled, recycled, rebound, and ultimately preserving traces of text that can no longer be read clearly by the naked eye. The Liste includes a substantial number of Greek New Testament witnesses shaped by this history of reuse: manuscripts in which earlier writing survives beneath later writing, often hidden, damaged, or obscured. These are palimpsests.

While GA 015 is not itself a palimpsest, its story illustrates the importance and potential MSI has for this field. If one reused manuscript can preserve this much recoverable evidence, what might still be brought to light from palimpsests already registered in the Liste but not yet examined with MSI?

This post looks more closely at Greek New Testament palimpsests. It begins by explaining what palimpsests are, how many are currently registered in the Liste, and what broader patterns emerge from this group of witnesses. It then turns to multispectral imaging, considering what this technology can reveal, where it has already proved successful, and which Greek New Testament palimpsests have already been examined using it.

What Is a Palimpsest?

At first glance, a palimpsest looks like any other manuscript. Yet beneath the visible text lie faint traces of an earlier work, deliberately erased so the parchment could be reused. The newer writing is known as the overtext (or scriptio superior), while the older, partially erased text beneath it is called the undertext (or scriptio inferior). Most palimpsests contain only one erased layer, but some parchments were reused multiple times, resulting in several layers of text.

For example, GA 2769 is described in the Liste as “dreimal beschrieben” (written three times). Its lowest layer contains a Menologion, the middle layer preserves portions of Matt 1:1–24 and 7:14–8:11, and the uppermost layer contains a Parakletike. In this case, the New Testament text survives as the middle stratum of the palimpsest.

The word palimpsest derives from the Greek παλίμψηστος (palimpsestos), meaning “scraped again,” from πάλιν (“again”) and ψάω (“to scrape”).² The term can be slightly misleading, however, because the preparation of parchment for reuse did not always involve aggressive scraping. Scribes often used gentler techniques such as washing, soaking, or chemically treating the surface to prepare it for new ink. A 10th-century recipe from Tegernsee, for instance, instructs the scribe to soak the old sheet in milk overnight, dry it under pressure with flour to prevent wrinkling, and then smooth the surface with chalk (and pumice if needed) before copying the new text.³

The aim was practical: to create a usable writing surface, not necessarily to obliterate every trace of the previous work. Because the erasure was often incomplete, many undertexts remain recoverable today with modern imaging techniques.

Although palimpsests are sometimes imagined as a single book whose pages were simply scrubbed clean and overwritten—like an artist painting over an old canvas—this analogy is imperfect. The process of reuse was frequently far more invasive. An earlier manuscript might be dismantled, its quires separated, and its leaves cut, folded, rotated, or rearranged to serve a new purpose. Complete codices made entirely from recycled parchment were relatively rare. More commonly, scribes incorporated individual folios, bifolios, or fragments from older books into new ones.⁴ As a result, a single palimpsest often preserves only scattered remnants of an earlier manuscript: leaves from the same original codex may now appear out of sequence, embedded in different volumes, or dispersed across collections.

A more fitting analogy, then, is that of a building constructed from stones salvaged from an earlier structure. The old building is taken apart; its materials are reshaped and reused, yet traces of their former life often remain visible. In much the same way, the parchment carrying an undertext survives because it was repurposed after the original manuscript had been disassembled. These material traces rarely allow us to reconstruct the lost codex in its entirety, but they nevertheless offer valuable evidence of its original form, use, and history.⁵

Palimpsests in Byzantine Manuscript Culture

Palimpsests are a natural consequence of the practical realities of pre-modern book production and the constant demand for usable writing material. Parchment, made from the prepared skins of sheep, goats, and calves, was remarkably durable and could last for centuries. Yet it was also labor-intensive to produce and often expensive to acquire. When a text was no longer valued or useful, the writing surface itself frequently remained in good condition and could be repurposed — often a more practical solution than obtaining fresh parchment. Recycled parchment thus circulated as a valuable material resource in its own right. Byzantine bookbinders are even known to have stockpiled already-erased leaves and sold them for use in copying or binding new manuscripts.⁶

Economic factors alone, however, do not fully explain the phenomenon. The creation of palimpsests was also shaped by patterns of use and ownership, shifting cultural values, and — in the case of New Testament manuscripts — the practical and liturgical needs of Christian communities. Manuscripts copied in languages no longer understood, written in scripts that had fallen out of use, duplicated in newer copies, damaged, left unfinished, or rendered obsolete for other reasons could all become candidates for reuse.⁷

Every palimpsest, therefore, preserves at least two intertwined histories: that of the erased undertext and that of the people and communities who chose to repurpose it. These manuscripts reveal not only what was written, but also the judgments made about which books were still usable in a given setting, which copies could be sacrificed, and how scarce resources should be managed. Making a palimpsest was, in essence, a decision that the parchment remained useful even when the text of a particular manuscript no longer did.

Counting Greek New Testament Palimpsests in the Liste 

There are currently 246 Greek New Testament palimpsest manuscripts in the Liste: 169 in which the Greek New Testament is the undertext, and 77 in which it appears as the later overtext.⁸

Since this post focuses on the potential of MSI for recovering obscured material, the scope is limited to the 169 palimpsests in which the Greek New Testament survives as the erased, damaged, or only partially recovered undertext.⁹

Among these 169 undertexts:

• 67 are majuscule manuscripts,
• 11 are minuscule manuscripts,
• 91 are lectionaries.

Surviving Extent

The physical preservation of these witnesses varies significantly:

• 30 survive in only a single folio or small fragment.
• 22 preserve 100 or more folios.
• Just 4 exceed 200 folios.
• The largest is GA 025 (9th century), which retains 327 folios containing Acts, the Catholic Epistles, Pauline Epistles, and Revelation. It is currently housed in the National Library of Russia, St. Petersburg.

Chronological Distribution

The undertext witnesses span the 5th-13th centuries. Their chronological pattern reveals important trends about which kinds of Greek New Testament manuscripts were most frequently recycled as palimpsests.

Early Dominance of Continuous-Text Majuscules

The earliest palimpsests are overwhelmingly continuous-text majuscules, dating from the 5th to the 10th century (nearly all before the 10th). This is not a marginal phenomenon: 67 of the 286 continuous-text majuscule entries in the Liste survive as palimpsest undertexts. In other words, nearly one-quarter of all known continuous-text majuscule witnesses to the Greek New Testament survive in manuscripts whose original text was erased and reused.

Shift Toward Lectionaries

From the 8th century onward, the pattern shifts markedly toward lectionaries. Of the 98 undertext witnesses dated to the 8th–10th centuries, 67 are lectionaries. This group includes 64 palimpsest majuscule lectionaries—roughly one-fifth of all majuscule lectionaries recorded in the Liste. As a result, a substantial portion of the surviving evidence for the early Greek lectionary tradition is preserved only as erased undertext.

Overall Significance

Taken together, these figures demonstrate that palimpsest undertexts are not peripheral to the textual tradition. For both continuous-text majuscules and majuscule lectionaries, a significant percentage of the known evidence survives as palimpsests. Any comprehensive study of these early witnesses must therefore take seriously the material preserved in this form.

Greek New Testament Palimpsest Undertexts by Century

Greek New Testament Palimpsest Undertexts by Century

Century	Majuscules	Lectionaries	Minuscules	Total
V	11	1	—	12
VI	19	—	—	19
VII	8	—	—	8
VIII	10	10	—	20
IX	18	29	—	47
X	1	28	2	31
XI	—	9	6	15
XII	—	12	3	15
XIII	—	2	—	2
Total	67	91	11	169

*For manuscripts with dates split between two centuries, for example X/XI, the later century has been used for this visualization.

Script and Category

An important pattern emerges when the undertexts are analyzed by both script type and manuscript category: the lectionary tradition retained majuscule script significantly longer than the continuous-text tradition.

Greek New Testament Palimpsests by Script/Category

Greek New Testament Palimpsests by Script/Category

Century	Majusculecontinuous text	Minusculecontinuous text	Majusculelectionaries	Minusculelectionaries	Total
V	11	0	1	0	12
VI	19	0	0	0	19
VII	8	0	0	0	8
VIII	10	0	10	0	20
IX	18	0	29	0	47
X	1	2	22	6	31
XI	0	6	0	9	15
XII	0	3	2	10	15
XIII	0	0	0	2	2
Total	67	11	64	27	169

Continuous-Text Witnesses

Among continuous-text manuscripts, the use of majuscule script is concentrated almost entirely before the 10th century. Of the 67 continuous-text majuscule undertexts:

• 66 date from the 5th through the 9th centuries
• only 1 dates to the 10th century

Lectionary Witnesses

Among the lectionary undertexts dated to the 10th century, 22 are majuscule lectionaries, compared with only 6 minuscule lectionaries.

This distribution aligns with the 10th-century evidence from the Liste as a whole:

• Majuscule lectionaries account for 128 of the total 166 lectionary witnesses.
• By contrast, continuous-text majuscules had declined sharply by the 10th century, accounting for only 21 of the 144 continuous-text witnesses.

The palimpsest evidence therefore fits the wider pattern: lectionaries preserved majuscule script about a century longer than continuous-text manuscripts did.

Languages and Overtexts

The overtexts also show that Greek New Testament manuscripts could be reused across linguistic boundaries. At least 27 witnesses were overwritten wholly or partly in languages other than Greek. These include Georgian (11), Arabic (5), Syriac (4), Coptic (3), Latin (2), Hebrew (1), and Middle Persian/Pahlavi (1).

GA 0225 offers a curious example. Its reused leaves were not simply erased and copied into a new codex; they were glued together into a roll and overwritten with Pahlavi/Middle Persian text.  

This demonstrates that the reuse of a manuscript could involve radical transformations in language, genre, and physical format.

Character of the Overtexts

Roughly one-third of the palimpsests have no identified overtext in the Liste. Among those that are identified, the vast majority are religious in nature. These include:

• 36 liturgical books (such as Menaia, Euchologia, and Old Testament lectionaries)
• 16 New Testament lectionaries

Taken together, this evidence shows that Greek New Testament manuscripts continued to have miltifaceted afterlives, moving across linguistic, geographic, and ecclesiastical boundaries long after their original production. The evidence suggests that these artifacts were often recycled within ecclesiastical settings, where they could be redirected toward the spiritual, liturgical, and pedagogical needs of later communities.

Double New Testament Palimpsests

A small but significant group consists of 21 palimpsests in which both the erased undertext and the later overtext received separate Gregory-Aland numbers. These manuscripts preserve two distinct instantiations of the New Testament within a single physical artefact: an earlier witness that was erased and a second New Testament text copied in its place. In one exceptional case, Selden Supra 2, there are even three distinct layers of New Testament text.

The following table lists the cases in which each layer of text has been separately catalogued.

Double New Testament Palimpsests

Double New Testament Palimpsests chart

No.	Undertext GA number and date		Overtext GA number and date	Shelfmark
1	GA 025IX c.	→	GA 18341301	St. Petersburg, National Library of Russia, Gr. 225
2	Ξ 040VIII c.	→	ℓ299XIII c.	Cambridge, Cambridge University Library, Add. 10062
3	GA 0132IX c.	→	GA 639XI c.	Oxford, Christ Church, Wake 37
4	GA 0209VII c.	→	ℓ1611XIV c.	Ann Arbor, University of Michigan Library, Ms. 8
5	GA 0233VIII c.	→	ℓ16841247	Münster, Bibelmuseum, Ms. 1
6	GA 0257IX c.	→	ℓ2094X c.	Zavorda, Nikanoros Monastery, 2
7	ℓ586X c.	→	GA 713XI c.	Birmingham, Cadbury Research Library, Peckover Gr. 7
8	ℓ668IX c.	→	GA 9821311	Athos, Esphigmenou Monastery, 27
9	GA 2854X c.	→	ℓ466XIII c.	Grottaferrata, Biblioteca Statale, A. α. 11.13
10	ℓ135VIII c.	→	ℓ136X c.	Vatican City, Biblioteca Apostolica Vaticana, Barb. gr. 472
11	ℓ269VIII c.	→	ℓ1944XIII c.	Venice, Biblioteca Nazionale Marciana, Gr. I,49 (1213)
12	ℓ397X c.	→	ℓ398XIV c.	Athens, National Library, EBE 177
13	ℓ1193IX c.	→	ℓ23431263	Athos, St. Panteleimon Monastery, 100
14	ℓ1214X c.	→	ℓ1235XIV c.	Athens, National Library, EBE 2112
15	ℓ1955IX c.	→	ℓ27XIV c.	Oxford, Bodleian Library, Selden Supra 3
16	ℓ1989VIII c.	→	ℓ1988XIV c.	Cambridge, Cambridge University Library, Add. 4489
17	ℓ2278X c.	→	GA 1897XII/XIII c.	Jerusalem, Greek Orthodox Patriarchate, Stavru 57
18	ℓ2309X c.	→	GA 220XIII c.	Ann Arbor, University of Michigan Library, Ms. 83
19	ℓ2393XII c.	→	ℓ494XIII c.	Grottaferrata, Biblioteca Statale, Γ. β. 12
20	ℓ2457XI c.	→	ℓ1065XI c.	Athens, Benaki Museum, TA 144
21	GA 0134; ℓ1954VIII c.; IX c.	→	ℓ26XIII c.	Oxford, Bodleian Library, Selden Supra 2

Script Transition

The pattern is striking. In nearly every instance, the overtext is written in minuscule script, while the undertexts are majuscule. The only exceptions are GA 2854, ℓ2393, and ℓ2457.¹⁰  These artifacts therefore visibly document the transition from majuscule to minuscule book production.

Nature of the Reuse

These are not cases of New Testament manuscripts being casually displaced by unrelated texts. Rather, they show older New Testament artefacts being deliberately recycled in the production of other New Testament books, often in response to the demands of lectionary use. Such manuscripts were dismantled and rewritten into codices better suited to later scripts, formats, and liturgical practices.

What the Liste  Can and Cannot Tell Us

Although limited in scope, the data in the Liste offers a valuable starting point for the study of Greek New Testament palimpsests. It provides essential information on the number of catalogued witnesses, their current locations, classifications, surviving extent, and—in many cases—the nature of the overtext. It also helps reveal broader patterns of manuscript reuse.

However, the Liste cannot provide a full picture of the undertexts themselves. In some cases the surviving New Testament content remains largely unidentified; in others we have only a partial identification. Closer examination of the manuscripts themselves is therefore essential to verify and expand our knowledge of their precise content and textual significance.

MSI has emerged as one of the most powerful tools for addressing these challenges.

What Is Multispectral Imaging?

Multispectral imaging refers to the non-invasive imaging process that captures image data at isolated wavelengths across the electromagnetic spectrum, such as ultraviolet (UV), visible, and infrared (IR), thus revealing features of an object that are otherwise imperceptible to the unaided human eye. Materials respond differently to various electromagnetic wavelengths, so MSI technology exploits these differences to enhance or reveal characteristics that would be otherwise invisible under ordinary light.

Process

Light is electromagnetic radiation or transfer of energy without physical connection. Think of the sun, for a second. As a major source of energy, the sun emits radiation through the air in waves that come in various frequencies and lengths in what is called the electromagnetic spectrum. 

You may remember from school that sunlight is a form of electromagnetic radiation that can be split into colors. A glass prism separates full-spectrum sunlight into individual bands of red, orange, yellow, green, blue, indigo, and violet, each color virtually corresponding to a particular wavelength of light. This very small portion of the electromagnetic spectrum is known as the visible spectrum.

The human eye is designed to perceive objects as light illuminates them. Different materials exhibit characteristic patterns of reflectance and absorption, interacting differently with various wavelengths of light. Surfaces subjected to full sunlight will absorb some wavelengths and reflect others, depending on their composition and physical properties. Our brain processes the interaction between surface and wavelengths captured by our eyes, leading us to perceive the surfaces and the colors associated with each wavelength of light.

While the human eye is a marvelous instrument, it can perceive only a narrow band of wavelengths ranging from about 400 to 750 nanometers (nm). The rather large electromagnetic spectrum includes wavelengths invisible to our eyes. This is the case of radio waves, microwaves, infrared, ultraviolet, X rays, and gamma rays. If you move beyond violet in the spectrum to shorter wavelengths, you get ultraviolet, X rays, and gamma rays. On the other hand, if you move beyond red in the spectrum to longer wavelengths, you get infrared, microwaves, and radio. The diagram below displays the wide electromagnetic spectrum, highlighting the small section of the visible spectrum:

Electromagnetic Spectrum (image from here)

Because our eyes cannot perceive ultraviolet light, we are missing numerous data points generated right in front of us as objects interact with ultraviolet waves. The same is true if the same objects are subjected to infrared and other wavelengths outside the visible spectrum.

MSI technology is particularly valuable because it captures data that our eyes cannot see and regular cameras cannot capture. Unlike the regular imaging process under normal light, MSI uses sophisticated lights to emit discrete wavelengths which special cameras are designed to detect. This means that MSI equipment photographs the same surface using multiple wavelengths and records the unique ways in which the same surface reacts to those different wavelengths (a.k.a. spectral signatures). Put another way, MSI records distinctive interactions between surface and light occurring outside visible light and maps them into images humans can interpret.

To capture images of an object using MSI technology, the technician must select spectral bands and decide which filters, if any, will be used. Sometimes, depending on the thickness of the substrate, a technique called “transmissive light” is employed, which projects light through the artifact instead of on its surface. Decisions on technique, wavebands, and filters often vary because they depend on the nature and condition of the material or substrate, what types of ink are present on the substrate, and what technicians hope to capture. Since every surface contains its own spectral signatures, two surfaces, albeit alike to the human eye, might require different procedures.

Once all procedural decisions have been made and the equipment has been properly calibrated, the camera captures numerous images of the same surface as each wavelength of light is emitted in sequence, with or without filter. In a realistic scenario, the process might capture a total of 20 or more images of the same object. Since the surface reacts differently to the various wavelengths, some images are more useful than others. Thus, in post-processing, the technician selects the most promising images for digital manipulation to highlight, whenever possible, areas of interest for further study.

Application in New Testament Textual Criticism

In the field of manuscript studies and consequently New Testament Textual Criticism, MSI is often employed to recover texts that have become illegible to the naked eye, either by accident or on purpose. Often, water, moisture, or fire damage the substrate. While water might wash off a significant amount of ink on parchment and paper, moisture and fire damage darkens parchment to the point that it is impossible to read any text.

Here is an image of a Dead Sea Scroll fragment likely damaged by moisture. This same surface was photographed with MSI technology using near infrared light. At certain wavelengths, the darkened parchment reflects light in a different manner, thus appearing lighter than the text. The result is that the text becomes visible.

Other times, the text of manuscripts becomes illegible by design, as is the case with palimpsests. MSI technology can often uncover that first layer of text since the substrate and the particles of ink in the two textual layers react in unique ways to the various wavebands. With some image manipulation, many times it is possible to isolate the two layers of text.

What MSI Has Already Made Possible

Over the last two decades, MSI has transformed the study of palimpsests. Before the technological advances of the last century, undertexts were recoverable only with difficulty. In the 19th century, paleographers often applied chemical reagents to recover the undertext of palimpsests. The rationale was simple. The ink most widely used on parchment, and consequently for the undertext of palimpsest, contained iron. Various potions were then concocted to cause a reaction in the traces of the iron-based ink, thus enhancing the visibility of the first layer of text. This practical step, despite its benefits for paleographers, came with long-term destructive effects since some mixtures included highly corrosive chemicals (e.g., potassium sulfide and hydrochloric acid).¹¹ Additionally, some mixtures frequently altered the color of the substrate to shades of green, blue, and brown. It was this type of procedure that gave Codex Ephraimi Rescriptus its characteristic appearance.

The scenario has changed quite significantly since then. Photography, the same technological novelty that provided new ways to study manuscripts 150 years ago, has evolved to a great degree, so that we can now employ non-invasive imaging techniques to capture and enhance hidden writing.

MSI has done far more than simply improve the photography of challenging manuscripts. In several major projects, it has fundamentally changed the amount of text available to scholars, refined manuscript dating, and improved cataloguing. The case studies below illustrate the extraordinary potential of MSI and what it may yet achieve for the Greek New Testament palimpsests preserved in the Liste.

The Archimedes Palimpsest Project

Although it does not involve a New Testament manuscript, the Archimedes Palimpsest Project remains the landmark example that established the modern paradigm for recovering erased texts with MSI. The manuscript originally contained a 10th-century copy of several works by Archimedes. It was later erased and reused as a 13th-century Euchologion. While earlier scholars had partially identified the undertext, much of it remained illegible.

Beginning in 2000 at the Walters Art Museum in Baltimore, the project employed advanced MSI to recover substantial portions of previously unreadable text. The team’s digital images and transcriptions were made freely available online in 2008. The importance of this project lies not only in the recovery of classical texts but also in demonstrating that erased writing could be recovered systematically and without damaging the manuscript, setting a precedent for later work on other palimpsests.¹²

Codex Zacynthius (GA 040) 

One of the most important applications of MSI to a Greek New Testament palimpsest so far is Codex Zacynthius (GA 040). Its upper text is a later Gospel lectionary (ℓ299), while the erased scriptio inferior, copied around the year 700, preserves a substantial portion of the Gospel of Luke accompanied by extensive catenae (marginal commentaries). It is the earliest known Greek New Testament catena manuscript. Here an image of the catena of Luke.

The project led by David Parker and Hugh Houghton at the Institute for Textual Scholarship and Electronic Editing (University of Birmingham) ran from 2018 to 2020. Using MSI, the team succeeded in making the undertext fully readable for the first time since its erasure. This enabled the first complete transcription of the older text.

Key contributions include the recovery and identification of the catena: 343 commentary extracts, among them 300 passages from early Christian writings not otherwise preserved in Greek, and 38 extracts from Severus of Antioch that represent some of the most substantial surviving Greek evidence for his works. The resulting digital edition, with MSI images, transcriptions, and the first English translation of a Greek New Testament catena makes this recovered material available for further study.¹³

The Sinai Palimpsests Project

The Sinai Palimpsests Project is one of the largest and most ambitious MSI initiatives undertaken to date. Conducted by St. Catherine’s Monastery, EMEL, and the UCLA Library between 2011 and 2016, the project examined 74 palimpsests (approximately 6,800 pages) from the monastery’s collection of more than 160 known palimpsest manuscripts.

Using MSI, the team identified 305 erased texts from Late Antiquity and the Middle Ages, written in 10 different languages. These recovered texts encompass biblical, patristic, liturgical, classical, medical, and documentary material. Many are either previously unknown, uniquely preserved, or represent the earliest surviving witnesses to known works.

For New Testament studies, the project yielded particularly important results. It revealed fragments of a previously unknown Old Syriac Gospel manuscript, significantly expanding our evidence for the early transmission of the Gospels in Syriac.¹⁴ Equally significant is Conrad Thorup Elmelund’s work on GA 0289. Using the new MSI images, Elmelund produced a far more accurate transcription than previously possible. The Liste had recorded only portions of Romans and 1 Corinthians, but the enhanced images revealed additional material from those books as well as text from Philippians and Colossians. His study also corrected some NA28 apparatus readings and the new images enabled him to date the undertext to the mid-5th century rather than the 7th or 8th. GA 0289 is therefore a particularly important reminder that MSI can affect not only what is visible, but how a witness is catalogued, dated, and cited.¹⁵

Vatopedi Palimpsests Project 

Another major initiative is the Vatopedi Palimpsests Project, launched in 2024 at Vatopedi Monastery on Mount Athos. This represents the first large-scale palimpsest digitization effort ever undertaken on Mount Athos. The project, a collaboration between Vatopedi Monastery, the UCLA Library, EMEL, and Aristotle University of Thessaloniki, aims to systematically examine all palimpsest manuscripts in the monastery’s library and recover their erased texts using MSI. 

For those interested in the Greek New Testament, the project holds particular promise. Although no Greek New Testament palimpsests from Vatopedi are currently recorded in the Liste, the monastery preserves nearly 200 Greek New Testament manuscripts. Any new discoveries emerging from this work could therefore be highly significant. Images and data from the project will be made publicly available through the UCLA Digital Library.

Beyond Imaging Alone: AI-Assisted Palimpsest Research

The field is rapidly moving beyond imaging alone. At the Centre for the Study of Manuscript Cultures (CSMC) at the University of Hamburg, the project “Generative AI for Revealing Palimpsests” (2025–2026) is exploring how computational methods can be combined with MSI to further enhance the readability of erased texts.

Even with high-quality MSI, a major challenge remains: when the undertext and overtext overlap heavily, the enhanced images may still not yield fully legible results. To address this, CSMC researchers are testing generative image-inpainting techniques designed to suppress the upper text and reconstruct clearer versions of the damaged or obscured undertext.

These AI-assisted tools hold great promise for complementing traditional philological, paleographical, and textual-critical work. The future of palimpsest research will likely rely on a powerful combination of methods: MSI to reveal what is invisible to the naked eye, AI-driven image analysis to process and separate complex layers of text, and human expertise to interpret and contextualize the recovered evidence.¹⁶

CSNTM and MSI

Besides capturing images of manuscripts using conventional photography, CSNTM has equipment to photograph manuscripts with MSI. The Center utilizes the MegaVision EV system, which captures high-resolution images across 12 or more spectral bands from the near ultraviolet to the near infrared. Depending on the degree of preservation of the ink, it is possible to make the faded text more legible.

In 2008, CSNTM photographed a manuscript housed at the National Center of Manuscripts in Tbilisi, Georgia using conventional photography. This manuscript is a palimpsest: while the upper text is in Georgian, the undertext, known as GA 0240, contains passages from the Pastoral Epistles in Greek. The palimpsest leaves were later photographed with MSI technology. The images below contrast the legibility of the undertext under visible light and the legibility of the same text under wavelengths outside the visible spectrum.

CSNTM has also employed MSI technology to photograph a manuscript housed at the Erzabtei St. Martin zu Beuron in Germany. This manuscript, which comprises only 4 bifolios, is a palimpsest containing two layers of text. The first layer, written in majuscule script possibly in the 9th century, contains passages from the Gospel of Matthew and is known as GA 0197. Around the 11th century, the biblical text was erased and liturgical material from a typicon was written in minuscule script. Reading the undertext is challenging because the ink has been erased to a great degree. With the assistance of MSI technology, however, scholars can access that earlier textual layer.

GA 0197 photographed under visible light and using MSI equipment:

In 2025, CSNTM photographed another artifact using MSI technology. This document is not a palimpsest. The ink, however, has suffered significant decay, making it impossible to read the text of many leaves under normal light conditions. Different wavelengths have now rendered the text legible.

Parchment leaf photographed under visible light and under UV light using MSI technology:

Because MSI equipment is expensive to acquire, highly specialized to use, and still not widely available, research projects often must rely on major grants, institutional partnerships, and long planning timelines.

It is at this point that CSNTM can assist institutions that hold manuscripts. Thanks to the financial support of individuals and private foundations, the organization has already acquired this technology and can utilize it to carry out the mission of digitizing and preserving all Greek New Testament manuscripts, including multiple layers of text preserved within a single document. Because CSNTM is privately funded, MSI technology can be made more readily accessible to holding institutions, thereby benefiting the preservation of cultural heritage and the broader scholarly community.

In the fall of 2026, CSNTM will engage in two projects involving multispectral imaging. One project entails the imaging of numerous Dead Sea Scrolls, while the other targets a 5th-century New Testament fragment in a private collection.

Collaboration with the holders of these manuscripts is one of the most important aspects of this work which we hope will continue to contribute to scholarship as the transmission of the New Testament is traced more accurately.

Current Imaging Status of Greek New Testament Palimpsests

So far only a small portion of the known Greek New Testament palimpsest undertexts appear to have been photographed with MSI. Of the 169 undertext witnesses considered in this post, we have only been able to confirm 10 which have been photographed with MSI. These are:

• Codex Zacynthius (GA 040)
• Portions of Codex Ephraemi Rescriptus (GA 04)
• Sinai Palimpsests Project: GA 0281, GA 0284, GA 0288, GA 0289, and ℓ907.
• CSNTM: GA 0197 and GA 0240. 
• British Library: ℓ1317.

We would welcome any additions to this necessarily provisional list, which will be updated as further information becomes available. The evidence remains striking: only a little more than 5 percent of the known corpus of Greek New Testament undertexts has so far been photographed with this technology. 

Conclusion: The Work Still Ahead

The Greek New Testament palimpsests recorded in the Liste represent one of the most promising yet underexamined frontiers in the field of New Testament textual criticism. While important research projects such as Codex Zacynthius and the Sinai Palimpsests Project have begun to examine these witnesses, many have received only minimal scholarly attention and considerable foundational work remains.

As this post has shown, these witnesses are far from peripheral: they comprise nearly one-quarter of all continuous-text majuscule witnesses and roughly one-fifth of all majuscule lectionaries in the Liste. A significant portion of our early evidence for both the continuous-text tradition and the lectionary tradition thus survives only in erased or damaged form. To neglect them is to leave a substantial part of the early Greek New Testament tradition unexplored.

MSI offers two important possibilities in this regard. First, it can recover text still hidden beneath erasures or overtexts, potentially expanding the material recorded in the Liste and refining our understanding of these witnesses. Second, it can illuminate the manuscripts’ complex histories—why they were erased and the social contexts in which they were produced, used, and repurposed.

While MSI is uniquely positioned to advance this work, the task ahead can only succeed through close and sustained collaboration among holding institutions, curators, librarians, New Testament textual critics, paleographers, and specialists in MSI.

The work that remains is substantial, yet the potential value is equally compelling. These hidden texts can refine our knowledge of specific witnesses, expand the recoverable textual evidence, and make a vital contribution to our knowledge of the New Testament’s transmission history.

---

1. Only fragmentary portions of 015 (Codex H) survive today, dispersed across collections in Paris, Turin, Kyiv, Moscow, Saint Petersburg, and Great Lavra Monastery. The significance of the recent digital recovery is not only that additional text from Paul’s letters can now be studied, but features of the Euthalian apparatus, which GAn015 is an important witness to, have also been recovered. See Garrick V. Allen and Kimberley W. Fowler, “The Ghost Leaves of Codex H and the Euthalian Apparatus,” Journal of Biblical Literature 144 (2025): 629–653. ↩︎
2. On the ambiguity of defining palimpsest, see Jost Gippert, “Removed and Rewritten: Palimpsests and Related Phenomena from a Cross-cultural Perspective,” in Palimpsests and Related Phenomena across Languages and Cultures, ed. Jost Gippert, José Maksimczuk, and Hasmik Sargsyan (Berlin: De Gruyter, 2025), 1–2. Gippert notes that the Greek term emphasizes erasure, rubbing, or scraping more directly than overwriting, raising the question of whether a manuscript whose text was erased but never subsequently rewritten should still be considered a palimpsest. Ted Erho holds to a stricter definition, reserving the term for a codex containing at least one quire in which an earlier text has been erased and overwritten by the main text of a newly produced manuscript. Ted Erho, “Ethiopic Palimpsests,” in Palimpsests and Related Phenomena, 393-432. ↩︎
3. For the Tegernsee recipe and for a description of several methods used to remove earlier writing from parchment, see András Németh, “Methods of Removing the Ancient Texts,” in Vatican Palimpsests: Digital Recovery of Erased Identities, Vatican Library. ↩︎
4. Claudia Rapp, “A Cache of Palimpsests and Christian Manuscript Culture across the Medieval Mediterranean,” 41. Rapp notes that “entire codices made up of palimpsest are relatively rare” and that it is “much more frequent” for individual quires to be copied on palimpsested material. She also notes that some palimpsests originated as scribal mistakes—folios whose ruling was poorly executed—or from manuscripts that were never completed, such as books left unfinished before illumination. A 15^th century example from Constantinople illustrates this process vividly. Examining the palimpsest manuscript Trin. 102, Binggeli showed that leaves from an earlier 11th-century codex were reused in several different ways: some were folded into new bifolios, others cut in half, others trimmed, and still others incorporated as complete bifolia. The scribe appears to have worked from complete leaves or bifolios of the earlier manuscript and adapted them according to the needs of the new codex. See André Binggeli, “Greek Palimpsests in the Patriarchal Library in Istanbul and the Making of a Palimpsest in Fifteenth-Century Constantinople,” 294–295. ↩︎
5. For a concrete example of this sort of reuse, see Grigory Kessel’s study of Sin. geo. 49, a 10th-century Georgian manuscript from St. Catherine’s Monastery. Kessel notes that the manuscript was copied using parchment from several earlier manuscripts in different languages and that some of those earlier manuscripts had themselves already been copied on reused folios. The case gives a good sense of how diverse elements of earlier books could be dismantled, rearranged, and assembled into a new codex. Grigory Kessel, “Membra disjecta sinaitica III: Two (Palimpsest) Fragments of Sin. geo. 49 and Their Four Syriac Undertexts,” The Vatican Library Review 1 (2022): 257–270. ↩︎
6. Described in Binggeli, “Greek Palimpsests,” 285–295. ↩︎
7. See Claudia Rapp, “A Cache of Palimpsests,” 40–44, especially her discussion of palimpsests as evidence for manuscript production, use, circulation, and textual preferences within medieval Christian communities. ↩︎
8. In the NTVMR, these entries can be found by selecting the manuscript feature “Palimpsest”; the results can then be further limited by selecting “Undertext” or “Overtext.” The count currently returns 248 entries but this includes GA 0168, which has been removed because it was entered without sufficient information to identify it again, and GA 2385, whose undertext was erroneously identified as New Testament but is, in fact, Septuagint. ↩︎
9. The 169 witnesses are currently distributed across 47 institutions in 39 locations. The largest holdings are the Vatican Library (26), British Library (19), National Library of Russia (17), St. Catherine’s Monastery (12), and the Grottaferrata State Library of the National Monument (12). Together, these 5 institutions hold 86 witnesses, just over half of the corpus. ↩︎
10. GA 2854, a complete tetraevangelion, was repurposed as the minuscule lectionary ℓ466. ℓ2393 consists of only 8 folios (likely a damaged or incomplete manuscript) that were washed and reused for ℓ494, a Euchologion. (Note: ℓ494 should actually be removed from the Liste, as it is not a New Testament lectionary proper; see Georgios Andreou, “Liturgical Manuscripts and the Siglum Lit in the Kurzgefasste Liste: Classification and Transmission,” Ex Fonte 5 (2026): 125–163, at 144.) The case of ℓ2457 is similar: it comprises only two folios that were reused in the full Gospel lectionary ℓ1065. ↩︎
11. For a brief overview of some chemical reagents paleographers applied to manuscripts in the 19th century, see Oliver Bock, “C. Maier’s Use of a Reagent in the Vercelli Book,” The Library 16 (2015): 249–81, pages 254–59. ↩︎
12. On the history and significance of the Archimedes Palimpsest project, see Roger L. Easton Jr. and William Noël, “The Multispectral Imaging of the Archimedes Palimpsest,” Gazette du livre médiéval 45 (2004): 39–49; and Reviel Netz, William Noel, Natalie Tchernetska, and Nigel Wilson, eds., The Archimedes Palimpsest, 2 vols. (Cambridge: Cambridge University Press for the Walters Art Museum, 2011). ↩︎
13. On the Codex Zacynthius project, see H. A. G. Houghton, “The Codex Zacynthius Project,” in H. A. G. Houghton and D. C. Parker, eds., Codex Zacynthius: Catena, Palimpsest, Lectionary, Texts and Studies (Piscataway, NJ: Gorgias, 2020), 9–18; and H. A. G. Houghton, Panagiotis Manafis, and A. C. Myshrall, The Palimpsest Catena of Codex Zacynthius: Text and Translation, Texts and Studies (Piscataway, NJ: Gorgias, 2020). ↩︎
14. For more on the Sinai Palimpsests Project, see Michael B. Phelps, “The Sinai Palimpsests Project (2011–2016): Goals, Methods, and Contributions,” in New Light on Old Manuscripts, 31–38; Claudia Rapp, “A Cache of Palimpsests and Christian Manuscript Culture across the Medieval Mediterranean: First Results of the Sinai Palimpsests Project,” in New Light on Old Manuscripts, 39–53. On the Old Syriac Gospel material, see Sebastian P. Brock, “Two Hitherto Unattested Passages of the Old Syriac Gospels in Palimpsests from St Catherine’s Monastery, Sinai,” Deltion Biblikōn Meletōn 31A (2016): 7–18. ↩︎
15. Conrad Thorup Elmelund, “The Undertext of Greek NF MG 99 from Sinai (GA 0289),” TC: A Journal of Biblical Textual Criticism 27 (2022): 51–68. Elmelund follows Guglielmo Cavallo’s paleographical assessment in redating the undertext to the 5th century. ↩︎
16. On the Hamburg project, see here. The project is led by Mahdi Jampour and runs from 2025 to 2026. On reducing interference from the overtext to make the undertext clearer for expert readers, see Mahdi Jampour, Hussein Mohammed, and Jost Gippert, “Enhancing the Readability of Palimpsests Using Generative Image Inpainting,” in Proceedings of the 13th International Conference on Pattern Recognition Applications and Methods 1 (2024): 687–694. ↩︎