How does it know which of the three scripts it is looking at?

From the letterforms themselves, and from anything you tell it about the manuscript’s known origin. Insular, Visigothic and Beneventan use unrelated shapes for the same letters, so getting this right first matters more than it does for closely related hands.

Can it separate a Latin gospel text from a later vernacular gloss?

Yes, if you ask for it. An interlinear or marginal addition in another language, or in a visibly later hand, can be transcribed as a layer of its own rather than folded into the main text.

Does it recognise the Tironian sign and regional forms of qui and que?

Yes, and you can choose whether they are kept as written or expanded, with the same choice applied consistently across the whole manuscript.

How does it handle Beneventan’s overlapping “biting” letters?

As a deliberate feature of the script, not as damage. Two letters sharing a single stroke in Beneventan is where an Insular or Visigothic scribe would keep them apart, and the transcription reflects that difference.

Will it flag a manuscript that mixes more than one regional script?

Yes, if you ask it to. Composite manuscripts, and later additions written in a different regional hand from the main text, are common, and it is worth flagging that rather than transcribing the whole thing as if it were uniform.

Latin · c. 650–1100

Tell three regional Latin hands apart, then transcribe them

After Rome, Latin split into distinct regional scripts that look nothing alike: rounded Insular writing in Ireland and Britain, angular Visigothic in Spain, and the boxy, overlapping Beneventan of southern Italy. Evidano identifies which regional system a page belongs to before reading a single letter, because the same shape can mean something different in each.

Upload a page See the samples

Sample pages

Real pages from public collections, shown beside their transcriptions. Pages whose transcription is still being checked are marked.

Nine lines of broad Insular half-uncial Latin with a smaller, browner Old English gloss written between the lines — Sample 1. A column of Jerome’s preface to the Gospels, the letter to Pope Damasus beginning “Novum opus”, in the Lindisfarne Gospels (London, British Library, Cotton MS Nero D.iv), written at or near Lindisfarne around 700 in a broad and exactly formed Insular half-uncial. The priest Aldred added an Old English gloss word by word between the lines in the tenth century, in a smaller and lighter hand that uses the Tironian ⁊ for “and”; the Latin here also shows the hooked e caudata for ae.

Manuscript page of rounded Insular majuscule headed by a tall zoomorphic initial, each short summary opening with a coloured initial — Sample 2. Folio 24 recto of the Book of Kells, the Irish gospel book written around 800, showing part of the breves causae – the short chapter summaries that stand before the Gospel of John, running from the witness of John the Baptist to the wedding at Cana and the woman of Samaria. The rounded Insular majuscule keeps its letters separate and starts each summary with a coloured initial, while ornaments fill out the ends of short lines.

Part of a manuscript page in small upright Beneventan minuscule, with short notes added in the right margin beside the text — Sample 3. Part of folio 38 recto of Florence, Biblioteca Medicea Laurenziana, Plut. 68.2 – the Second Medicean, the only medieval copy of Books 11–16 of Tacitus’s Annals, written at Monte Cassino in the eleventh century in Beneventan minuscule. The passage describes Nero’s punishment of the Christians after the fire of Rome, and a later reader has picked out Nero, the Christians and Pontius Pilate in the margin. Reproduced from Enrico Rostagno’s 1902 phototype facsimile of the manuscript.

Two-column Visigothic minuscule text page from a Spanish manuscript of a commentary on the Apocalypse, with red rubrics — Sample 4. A page from the “Silos Apocalypse”, a copy of Beatus of Liébana’s Commentary on the Apocalypse made at the monastery of Santo Domingo de Silos in northern Spain between 1091 and 1109, in the angular Visigothic minuscule the house kept using long after most Spanish scriptoria had switched to Caroline script. Red rubrics mark the end of one book and the start of the next.

Three scripts, one language

Insular script developed in Irish monasteries from the seventh century and spread to Northumbria, producing grand majuscule gospel books like the Book of Kells and a more compact Insular minuscule and half-uncial for everyday copying, seen in the Lindisfarne Gospels. Distinctive features include a flat-topped g, a tall or looped s, wedge-shaped serifs, and the Tironian sign (⁊) for “and”. Visigothic script emerged in Spain around the same time and, remarkably, survived at conservative houses like the monastery of Silos into the twelfth century, long after most of the peninsula had adopted Caroline minuscule; it favours an open a and its own distinctive ligatures, and is the script of Beatus of Liébana’s widely copied Commentary on the Apocalypse. Beneventan script, centred on Monte Cassino and the cities of Benevento and Bari, was in use from the eighth century into the thirteenth for the Rule of St Benedict itself and the wider output of southern Italian monasteries.

Beneventan is the most visually distinctive of the three: its letters are compressed, angular and boxy, with tall serifs that end in a small hook, and certain letter pairs deliberately overlap, or “bite”, into a single compound shape rather than standing apart. Visigothic hands are more rounded than Beneventan but sharper than Insular, with a characteristic treatment of the combination ti that distinguishes a hard from a soft sound, something no other Latin script bothers to mark. Insular hands range from the large, decorated majuscule of a display gospel book to a small, current minuscule used for glosses and working copies, but keep their rounded bowls and distinctive g in either register.

All three traditions share features that set them apart from the Caroline minuscule that eventually replaced most of them: heavier and more idiosyncratic abbreviation, regional preferences for shortening common words like qui and que, and punctuation systems that do not map cleanly onto later conventions. Because Insular gospel books remained in active use for centuries, many later gained interlinear glosses in the vernacular – Old English at Lindisfarne – written by a different, much later hand between the original Latin lines.

Why ordinary OCR struggles here

Same letter, three different shapes

g, r, t and common ligatures are built differently in each tradition. A model trained on one regional system misreads the other two, because the shapes are not variations on a theme but genuinely separate solutions to the same letter.

Regional abbreviation marks and forms of qui and que

Each tradition has its own preferred shorthand for extremely common words like qui and que, plus its own use of the Tironian ⁊. Recognising the mark requires already knowing which regional system is on the page.

Overlapping “biting” letters, especially in Beneventan

Certain letter pairs share a single stroke rather than standing apart, a deliberate feature of the script rather than damage or a ligature in the ordinary sense. Segmenting these into separate letters needs a model that has actually seen Beneventan.

Later glosses layered onto an older text

Insular gospel books often gained an interlinear vernacular gloss centuries after they were written, in a visibly different and later hand. Separating the original Latin from the addition needs explicit handling, not a single reading pass.

Who works with this material

Editors of Insular gospel books and glossed texts

Anyone preparing an edition of an Insular manuscript needs the Latin and any later interlinear vernacular gloss kept apart, letter for letter, rather than merged into a single reading.

Hispanists working on early medieval Spain

Visigothic-script cartularies and biblical commentaries from houses like Silos and León need a transcription that recognises the script has nothing to do with the Caroline hands used everywhere else in western Europe by that date.

Historians of Monte Cassino and southern Italian monasticism

Beneventan charters and liturgical books need a tool that has actually been shown the script’s overlapping letters, rather than a generic “medieval Latin” model that treats them as damage.

Palaeography students learning to date manuscripts by script

Checking a claimed regional attribution against an actual line-by-line transcription is a direct test of the skill a palaeography course is trying to teach.

Getting the best transcription

Step 1

Name the regional script before you upload

Say whether the manuscript is Insular, Visigothic or Beneventan. A general “medieval Latin” prompt defaults to assumptions built for Caroline minuscule that fit none of the three.

Step 2

Choose how to treat the Tironian sign and regional forms of qui and que

Decide whether these should be kept as written or expanded, and apply that choice consistently across the manuscript rather than deciding page by page.

Step 3

Separate any later gloss from the main text

Where a page carries an interlinear or marginal addition in a different language or a visibly later hand, ask for it to be transcribed as its own layer rather than merged into the main reading order.

Step 4

Test on a page with a rubric or colophon first

Red-ink section headings and closing notes often name the scribe, house or date, and reading one correctly is a fast check that the model has recognised the right script before you run a whole manuscript.

Frequently asked questions

How does it know which of the three scripts it is looking at?: From the letterforms themselves, and from anything you tell it about the manuscript’s known origin. Insular, Visigothic and Beneventan use unrelated shapes for the same letters, so getting this right first matters more than it does for closely related hands.
Can it separate a Latin gospel text from a later vernacular gloss?: Yes, if you ask for it. An interlinear or marginal addition in another language, or in a visibly later hand, can be transcribed as a layer of its own rather than folded into the main text.
Does it recognise the Tironian sign and regional forms of qui and que?: Yes, and you can choose whether they are kept as written or expanded, with the same choice applied consistently across the whole manuscript.
How does it handle Beneventan’s overlapping “biting” letters?: As a deliberate feature of the script, not as damage. Two letters sharing a single stroke in Beneventan is where an Insular or Visigothic scribe would keep them apart, and the transcription reflects that difference.
Will it flag a manuscript that mixes more than one regional script?: Yes, if you ask it to. Composite manuscripts, and later additions written in a different regional hand from the main text, are common, and it is worth flagging that rather than transcribing the whole thing as if it were uniform.

Related scripts and pages

Late Antique Uncial and Half-Uncial Codices, c. 300–700The half-uncial book hand all three regional scripts on this page descend from.Carolingian and Romanesque Minuscule, c. 750–1150The reformed hand that replaced most regional scripts except Beneventan and Visigothic Silos.Old English Insular Manuscripts, c. 700–1066The same Insular letterforms used for vernacular English rather than Latin.All Latin hands and periodsEvery Latin domain from Roman cursive to Renaissance humanistic script.Image to TextSearch all supported languages, scripts and document types.

Transcribe your Insular, Visigothic or Beneventan manuscript

Upload a folio, choose Insular, Visigothic, and Beneventan Latin as the domain, and name the regional script before you run the rest.

Start transcribing Create a free account