Can the transcription keep yat, the yus letters and omega as written?

Yes. The default is a diplomatic reading that reproduces ѣ, ѧ, ѫ, ѡ, ѳ, ѵ, ѕ and the ѹ digraph as they appear, including where the scribe used the historically wrong letter. A normalised version can be produced from the same pass.

How are titla and superscript letters handled?

They are treated as abbreviations, not as noise. A word under a titlo is expanded from the sense of the passage, and letters written above the line go back into the word they belong to. You choose whether the result shows the closed form or the expansion, with or without the supplied letters marked.

Will a date written in Cyrillic numerals come out as a number?

Letters under a stroke and fenced by points are recognised as numerals and read in the order the system uses, including the teens, where the unit precedes the ten. The year is given as written and, on request, with its common-era equivalent beside it.

Does this work on birch-bark letters as well as parchment codices?

It does, though bark is scratched rather than inked and the grain competes with the strokes. A raking-light photograph gives far better results than a flat one, and readings from cracked sheets are proposals for a specialist to confirm.

What happens to text lost to holes, stains or flaked ink?

Illegible stretches are marked rather than filled in by guesswork, and partly visible letters are flagged as uncertain. Where a parallel copy supplies the missing words, the editor adds them in brackets afterwards.

Russian · c. 1000–1400

Read the ustav hand of early Rus

Ustav draws every letter separately and leaves no gap between words, so a line of a gospel book arrives as one unbroken string. Evidano transcribes the folio as it stands, keeps or expands the titla as you instruct, and reads a letter under a stroke as the number it is rather than as a stray consonant.

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.

Last leaf of the Ostromir Gospel in two columns of eleventh-century ustav, with the words AZ GRIGORII DIAKON in large red letters near the head of the second column — Sample 1. The scribe’s record at the end of the Ostromir Gospel, St Petersburg, National Library of Russia, F.п.I.5, f. 294v, finished in 1057. Deacon Grigory names his patron, the Novgorod posadnik Ostromir, and dates his work by saints’ days; both columns are pure ustav, with titla over the sacred names and a red display line.

Two columns of eleventh-century ustav below an ornamental frame, with drawn animals and a small human figure in the lower margin — Sample 2. A page of the Izbornik of Svyatoslav, dated 1073, State Historical Museum, Moscow. Below the framed headpiece two columns run in an even ustav with almost no descenders, and the marginal beasts and the small figure at the foot sit directly against the written area.

Sheet of birch bark broken across the middle, with nine lines of Cyrillic letters scratched into the surface — Sample 3. Novgorod birch-bark letter no. 109, first quarter of the twelfth century: Zhiznomir writes to Mikula about a slave woman bought in Pskov and the dispute that followed. The letters are scratched with a stylus, in a plain book hand without word division, and the sheet has split along the grain.

Single column of large brown ustav on parchment, twenty-one lines with several small letters and numerals set above the line — Sample 4. A folio of the Arkhangelsk Gospel, dated 1092, Russian State Library, Moscow: the parable of the talents in a wide, slightly sloping ustav. Letter-numerals under strokes carry the sums, superscript letters sit over their words, and the leaf number was added centuries later.

Parchment chronicle page with continuous prose above and four short annual entries below, each opening with a red initial and a date in Cyrillic numerals — Sample 5. The Synodal copy of the Novgorod First Chronicle, f. 1v, State Historical Museum, Moscow, thirteenth to fourteenth century. The lower half shows the chronicle’s shape: each year opens with a red В, the date follows as letters under titla, and a year with nothing to report stays a bare heading.

What an ustav page looks like

Ustav is the formal book hand of the earliest East Slavonic manuscripts, modelled on Greek uncial and used from the conversion of Rus until the fourteenth century. Its letters are drawn rather than written: each stands separately and upright between two ruled lines. Little descends below the line, nothing separates one word from the next, and the alphabet is far larger than the modern one — yat (ѣ), the yus letters (ѧ, ѫ and their iotated pairs), omega (ѡ), fita (ѳ), izhitsa (ѵ), zelo (ѕ), the digraph ѹ for u, and iotated ѥ and ꙗ. The nasal vowels the yus letters once stood for had already left East Slavonic speech, so scribes copying South Slavonic exemplars swap ѧ for ꙗ and ѫ for ѹ.

Frequent and sacred words are contracted under a titlo, a stroke that often shelters a letter lifted clear of the line: б҃ъ for богъ, г҃ь for господь, ч҃лвкъ for человѣкъ. The same stroke turns letters into numbers — а҃ is one, і҃ is ten, ф҃ is five hundred — and a ҂ sign multiplies by a thousand, so a chronicle year written ҂ѕ҃ фк҃є is 6525 from the creation of the world, our 1017. Points set at mid height, singly or in clusters, carry the whole burden of punctuation, and headings and initials are laid in cinnabar, which reproduces paler than the black ink.

Large gospel books are ruled in two columns and smaller codices in a single block, the prickings that guided the ruling often still visible at the edge. At a line end the scribe shrinks the last letters, tucks them above the line or joins them into a ligature; lection numbers, liturgical directions and later readers’ notes crowd the margins. The same letterforms travel outside books: the everyday letters scratched into birch bark at Novgorod use a stripped-down ustav with no ink at all. From the late fourteenth century the hand grows narrower and quicker and becomes poluustav.

Why ordinary OCR struggles here

Lines with no word division

A scribe writing scriptio continua expects the reader to know where each word ends. Layout-driven OCR splits the line at random and yields plausible-looking nonsense. Segmentation has to come from the language, which means a model that knows Church Slavonic morphology rather than one that only knows letter shapes.

Titla that swallow whole syllables

Under a single stroke чл҃вкъ stands for человѣкъ and гл҃а for глагола. Reading these is vocabulary, not pattern matching: the same skeleton of consonants expands differently in a rubric and in a chronicle entry, and a silently wrong expansion passes unnoticed unless the transcription records what was supplied.

Letters doing duty as numbers

Dates, chapter and lection numbers and sums of money are written with letters under a titlo and fenced by points. Generic recognition reports them as text, so a year becomes a nonsense word and a register loses its figures. The teens also put the unit before the ten.

Cinnabar, damage and later hands

Red ink flakes, parchment cockles and loses corners, and offsets from the facing leaf print backwards across the page. Later owners add foliation, marginal crosses and sixteenth-century notes in quite different hands. A usable transcription keeps the rubrics, marks the losses and leaves the additions outside the medieval text.

Who works with this material

Editors of early Slavonic texts

Anyone editing a gospel lectionary, a florilegium or a homiliary needs a first pass that preserves the manuscript’s own letters and does not quietly modernise ѧ into я. The earliest codices survive in one or two copies, so every variant must be traceable to a line on a folio.

Church historians and liturgists

Lection headings, saints’ commemorations and calendar rubrics are what make an eleventh-century gospel book datable and localisable. They are also the most heavily abbreviated parts of the page, so they gain most from a transcription that treats rubric as text rather than decoration.

Archaeologists publishing birch bark

Excavation in Novgorod, Staraya Russa and Smolensk still turns up scratched bark every season. Field publication needs a fast, honest first reading with doubtful letters flagged, which the specialist then corrects against the object under raking light.

Libraries and digitisation projects

A catalogue entry needs the incipit, the colophon and any owner’s notes, not the whole codex. Running the first and last folios of each manuscript makes a collection findable by its opening words rather than by shelfmark alone.

Getting the best transcription

Step 1

Settle the alphabet before you start

Say whether ѣ, ѧ, ѫ, ѡ, ѳ, ѵ and the ѹ digraph should be kept or mapped to modern equivalents. Keeping them suits a linguistic corpus; mapping them helps a search index. Fix the rule once so that a hundred folios come out consistent.

Step 2

Decide what happens to the titla

Ask for contractions to be left closed, expanded silently, or expanded with the supplied letters in brackets. The bracketed form costs nothing to strip later and shows that г҃ь on the page was read as господь, which is what an apparatus needs.

Step 3

Handle the numbers explicitly

Tell the model to keep letter-numerals as written and add the Arabic value beside them rather than replacing them. A chronicle year then stays ҂ѕ҃ фк҃є, its conversion to 1017 is visible, and an error in the arithmetic can be caught without returning to the parchment.

Step 4

Test on a rubricated, damaged folio

Pick a page that mixes red and black ink, has a hole or a stain and carries a later marginal note. If that page comes back with rubric separated, loss marked and the annotation kept out of the main sequence, the clean folios will need almost no correction.

Frequently asked questions

Can the transcription keep yat, the yus letters and omega as written?: Yes. The default is a diplomatic reading that reproduces ѣ, ѧ, ѫ, ѡ, ѳ, ѵ, ѕ and the ѹ digraph as they appear, including where the scribe used the historically wrong letter. A normalised version can be produced from the same pass.
How are titla and superscript letters handled?: They are treated as abbreviations, not as noise. A word under a titlo is expanded from the sense of the passage, and letters written above the line go back into the word they belong to. You choose whether the result shows the closed form or the expansion, with or without the supplied letters marked.
Will a date written in Cyrillic numerals come out as a number?: Letters under a stroke and fenced by points are recognised as numerals and read in the order the system uses, including the teens, where the unit precedes the ten. The year is given as written and, on request, with its common-era equivalent beside it.
Does this work on birch-bark letters as well as parchment codices?: It does, though bark is scratched rather than inked and the grain competes with the strokes. A raking-light photograph gives far better results than a flat one, and readings from cracked sheets are proposals for a specialist to confirm.
What happens to text lost to holes, stains or flaked ink?: Illegible stretches are marked rather than filled in by guesswork, and partly visible letters are flagged as uncertain. Where a parallel copy supplies the missing words, the editor adds them in brackets afterwards.

Related scripts and pages

Church Slavonic Poluustav and Vyaz, c. 1300–1700The narrower, faster book hand that replaced ustav, with interlaced vyaz headings above it.Greek Biblical Majuscule and Uncial Codices, c. 300–900The Greek uncial book hand from which ustav letterforms and its abbreviation habits descend.All Russian and Church Slavonic handsEvery Russian domain from eleventh-century ustav to pre-reform records of 1918.Image to TextSearch all supported languages, scripts and document types.

Transcribe your ustav folios

Upload a scan, choose Church Slavonic Ustav Manuscripts as the domain, and try the prompt on one folio before running a whole codex.

Start transcribing Create a free account

Read the ustav hand of early Rus

Sample pages

Transcription

Transcription being verified

Transcription

Transcription being verified

Transcription being verified

What an ustav page looks like

Why ordinary OCR struggles here

Lines with no word division

Titla that swallow whole syllables

Letters doing duty as numbers

Cinnabar, damage and later hands

Who works with this material

Editors of early Slavonic texts

Church historians and liturgists

Archaeologists publishing birch bark

Libraries and digitisation projects

Getting the best transcription

Settle the alphabet before you start

Decide what happens to the titla

Handle the numbers explicitly

Test on a rubricated, damaged folio

Further reading on this hand

Frequently asked questions

Related scripts and pages

Transcribe your ustav folios

Company

About

Preview Evidano

Account

Newsletter