Reading Early Modern Paleography: A Working Guide to the Hands, Letterforms, and Transcription

Early modern paleography guide to identifying major manuscript hands, decoding misleading letterforms and dating conventions, and using machine transcription without losing source integrity.

Leo Team

July 14, 2026

Reading Early Modern Paleography: A Working Guide to the Hands, Letterforms, and Transcription
Contents

Early modern paleography (also spelled palaeography) is the study of how people wrote between roughly 1500 and 1800, and how to read what they left behind. For a working historian, the practical problem breaks into three parts:

  • identifying the hand on the page (secretary, italic, the court hands, French bâtarde, German Kurrent, and their relatives)
  • decoding the letterforms, abbreviations, numerals and dates that no longer work the way ours do (the long s, the thorn-derived ye, the macron for an omitted nasal, li. s. d., a year that begins on 25 March)
  • choosing a transcription convention (diplomatic, semi-diplomatic, or normalized) and holding to it consistently

Get those three right and the rest is patience.

This is a reference for that work: the hands, marks, numbers, dates and documents you will meet in a European or North American archive, the conventions that make a transcription defensible, where to learn, and where machine transcription fits. It sits within the broader craft of reading old handwriting and paleography, and it assumes you would rather understand a page than be told a machine has understood it for you.

The hands you will meet, c. 1500–1800

Paleography begins with recognition. Before you can read a document you have to know what family of script you are looking at, because the letterforms (and the traps) differ by hand.

Secretary hand is the script most English researchers hit first and hardest. An offshoot of the early Tudor court hands, it was the everyday English documentary hand from about 1550 to 1650. Its features are unfamiliar enough to stop a modern reader cold:

  • looped ascenders and descenders
  • an e that looks reversed, like a Greek theta
  • a twin-stemmed r
  • a c built square, like a small r
  • an h that drops below the line like a butcher's hook

Italic displaced it over the seventeenth century, and by about 1700 secretary survived mainly as stray letterforms in otherwise italic writing. Cambridge's English Handwriting 1500–1700 course remains the standard training ground. See also the walkthrough of how to read secretary hand; to learn it letter by letter, Leo Academy's free English secretary hand course and its secretary hand alphabet chart use real scribes' hands from the period.

Italic arrived from Italy in the fifteenth century, derived from humanist minuscule, and English humanists took it up in the early sixteenth. It is the ancestor of modern English handwriting, which makes it the easiest of the period's hands to read at sight.

Mixed hands and round hand. Most seventeenth-century writing is neither pure secretary nor pure italic: writers kept a secretary e or r inside an italic frame, so a letter can switch forms within a line. Cambridge's guide to dating and describing hands calls these mixed or transitional hands. They led to the English round hand of the mid-to-late seventeenth century, which became the eighteenth-century standard and the ancestor of copperplate.

Court hands are a separate difficulty. This family of English legal scripts was deliberately conservative and, in the case of Chancery, famously hard:

  • Common Law (Common Pleas and King's Bench)
  • Chancery
  • Exchequer (the "pipe roll" hand)

The court hands outlived ordinary writing by generations, until the Proceedings in Courts of Justice Act 1730 required that, from 25 March 1733, court records be written in English, in "a common legible hand", and not abbreviated. If your material is equity litigation, see the guide to reading Chancery court records.

The Continental hands follow their own lineages, many of them descended from late-medieval gothic cursive:

  • French cursive, the bâtarde and notarial hands,: dense, heavily looped, thick with ligatures (reference: the École des chartes' Theleme)
  • German Kurrent and its angular chancery variant, Kanzleischrift, served German vernacular writing from the sixteenth century until 1941. Sütterlin (1911–c.1941) is a late chapter, not the whole story; see how to read German Kurrent.
  • Spanish procesal, a long-lived notarial cursive, descends from the fifteenth-century cortesana
  • Italian mercantesca (the medieval merchant hand) and cancelleresca (the chancery hand from which italic type derives)
  • Dutch chancery cursive has its own distinctive letterforms

All of these are Latin-alphabet scripts, whatever the language, and that distinction governs what any recognition system can do; see which languages and scripts HTR actually reads.

The letterforms that trip everyone up

Once you have identified the hand, the individual marks are where transcription errors are made. A handful recur across languages and hands often enough to be worth committing to memory.

The long s (ſ, U+017F). It was used at the start and in the middle of words in English, French and German from roughly 1500 to 1800, but not at the end, where the round s took over. A final double s is therefore often written ſs, while a doubled long s, ſſ, is common inside a word (poſſeſs). Misreading it as f is the single most common early modern transcription error. See the long s as secretary scribes wrote it. It was abandoned in English print around 1790–1810 and persisted in handwriting a little longer.

Thorn and the y-shaped article. Thorn (Þ/þ) had largely given way to th in English by the fifteenth century, but its ghost survives in early modern writing and print as the y-shaped abbreviations yᵉ, yᵗ and yˢ, meaning the, that and this. They are not "ye olde" anything: the y is a thorn in disguise.

Minims. The short vertical strokes that make up m, n, u and i are the classic ambiguity. A run of them (minimum, say, or a roman numeral) can be read several ways, and only context resolves it.

u/v and i/j. These were variant shapes of single letters. Before about 1630, English printers used v at the start of a word and u elsewhere, whatever the sound, and i stood for j. Expect vnto, loue and iustice: they are correct transcriptions, not errors.

Abbreviations. Early modern scribes and printers alike carried a large repertoire of space-saving marks:

  • A suspension omits letters at the end of a word
  • A contraction omits them in the middle
  • The macron or titulus (combining macron, U+0304) marks an omitted m or n, so cōmon is common and q̃ expands to que
  • A p with a bar through the descender (ꝑ) is per or par (ꝑishe, parish). A p with a loop back through the descender (ꝓ) is pro.
  • Raised letters shorten common words: wᶜʰ (which), yoʳ (your), Mʳ
  • Brevigraphs, standalone signs, stand for whole words or syllables: the Tironian et (⁊, U+204A), the -us and -rum signs, the -es sign

The Academy's abbreviation pages give about 190 of the commonest English and Latin ones, each with its expansion. There is a broader reader's guide to manuscript abbreviations and ligatures. For Latin, Cappelli's Dizionario di abbreviature latine ed italiane remains the reference dictionary. For Unicode code points, follow the Medieval Unicode Font Initiative.

Numbers and money

Accounts, inventories and wills put numbers on the page in forms a modern reader will misread.

Roman numerals, with a twist. Usually lower case, with a final i written as j: iij is 3, xiij is 13. Larger numbers are often built by multiplication with a raised letter: iiijˣˣ is four score, or 80, and vjᶜ is 600. A crossed l (ƚ) can stand for 50. Arabic figures spread through the period, and many documents mix the two.

Pounds, shillings and pence. English money ran in three columns:

  • 12 pence (d., denarii) to the shilling
  • 20 shillings (s., solidi) to the pound (li., l. or £, librae)

Smaller sums appear as ob. (a halfpenny, obolus) and q. or qa. (a farthing, quadrans). Two named sums turn up constantly: the mark, 13s 4d (two-thirds of a pound), and the noble, 6s 8d. Summa heads the total. The Academy's note on pounds, shillings and pence works an inventory total through line by line. The National Archives' currency converter gives a rough sense of what a sum would buy.

Dating: the trap hiding in plain sight

A correctly transcribed date can still be wrong by a year if you do not know the conventions.

The year began on 25 March. In England the civil and legal year began on Lady Day, 25 March, until 1752. A document dated in January, February or March in seventeenth- or eighteenth-century England therefore often carries what looks to us like the previous year. Historians double-date those months (16 March 1592/3). Scotland moved its New Year to 1 January in 1600.

Old Style and New Style. Catholic Europe adopted the Gregorian calendar from 1582, when ten days were dropped. England did not change until 1752, when 2 September was followed by 14 September. Letters between England and the Continent may carry both dates, marked O.S. and N.S.

Regnal years count from a monarch's accession, not the calendar year. 14° Eliz. is the fourteenth regnal year of Elizabeth I, which runs from 17 November 1571 to 16 November 1572. Watch Charles II, whose years were counted from his father's execution in January 1649, so the Restoration of 1660 fell in his twelfth year.

Feasts and terms. Many documents date by saints' days and quarter days (Michaelmas, Lady Day) or by the law terms rather than by day and month.

The Academy's note on New Year, double dates and regnal years tabulates where each reign's years begin; C. R. Cheney's Handbook of Dates is the standard reference.

The documents you will meet

Knowing why a record exists tells you what to expect on the page. In English archives the commonest types are these.

  • Parish registers. Thomas Cromwell's injunction of 1538 ordered every parish to record baptisms, marriages and burials. An order of 1597–8 required parchment books, with older entries copied in, so many "early" registers are later copies. From 1598 a yearly copy, the bishop's transcript, went to the diocese, and it may survive where the register does not. Short, formulaic entries make them the classic first document; see transcribing parish records.
  • Wills and probate inventories. Before 1858, wills were proved in the church courts. From about 1530 neighbours listed and valued the dead person's goods, often room by room, so probate inventories are dense with money and numerals; see transcribing old wills.
  • Deeds and manorial records. Conveyances follow fixed clause patterns. Manor court rolls record tenants' land, rents and local order.
  • Quarter sessions and church courts. Justices of the peace met four times a year for lesser crimes and county business. Church courts handled probate, marriage, defamation and tithes, often in Latin.
  • Letters and accounts, in every hand from formal secretary to hurried italic

Latin remained the language of many legal and ecclesiastical records until 1733, so expect formulae such as anno regni and sepultus. A glossary of Latin genealogy terms covers the vocabulary of the registers.

Choosing a transcription convention

Reading the page is half the task. The other half is deciding how faithfully to represent what you read, and documenting that decision.

  • Diplomatic transcription preserves everything verbatim: original spelling, abbreviations, punctuation, deletions, insertions. It is the most defensible for scholarly citation and the most laborious.
  • Semi-diplomatic transcription expands abbreviations, usually with the expansion marked (as yo[u]r), but keeps original spelling and layout
  • Normalized transcription modernizes spelling, punctuation and layout. It is legitimate, but only if the convention is stated so a reader knows what has been changed.

The recurring error here is treating archaic spelling as a mistake. It is not. Publick, shew and yeomanry are orthographically correct for their date, because spelling was not standardized. Silently modernizing them is itself a transcription error, and one that quietly destroys evidence about scribe, region and date.

A second trap: editions disagree on signs. Square brackets mean supplied letters in one edition and deleted words in another, so read an edition's key, and state your own. Cambridge sets out a clear house style in its transcription conventions, and there is a fuller discussion in transcription conventions for old documents. When you formalize any of this for publication, TEI P5 chapter 12 provides the standard encoding for primary sources, and chapter 11 does the same for manuscript description.

Where to learn

Paleography is learned by reading a little every day against a key. Free resources:

In print: Preston and Yeandle, English Handwriting 1400–1650; Dawson and Kennedy-Skipton, Elizabethan Handwriting 1500–1650; Petti, English Literary Hands from Chaucer to Dryden; and Hector, The Handwriting of English Documents. For a general approach, see this method for reading an unfamiliar hand.

Where machine transcription fits, and where it does not

At some point the volume of material outruns the hours. Hundreds of images of notarial cursive or secretary hand cannot be transcribed by eye on any realistic timeline. That is the work Leo's own model, ATR-1, was built for: it is tuned specifically for historical documents, handwritten and printed. Machine transcription still has limits, and three misconceptions about it do real damage, but first, what a specialised model does differently.

What a model built for historical documents does differently

ATR-1 is a Transformer fine-tuned to read images of historical documents. A general AI model is built to produce fluent modern text; ATR-1 is trained to transcribe what is written, and not to over-correct or silently modernize it:

  • the long s stays a long s
  • the macron is not silently resolved to the letters it stands for
  • archaic spelling, u/v and i/j, and strikethroughs survive into the transcription

It reads the image at high resolution, where general models scale it down, so a run of minims is still there to weigh. It works zero-shot, with no per-hand model to train first, and reads Latin-script manuscript and print in English, French, German, Spanish, Italian, Dutch and Latin, among the many languages written in that alphabet.

It also checks its own work. Output that shows a known failure pattern is hidden and retried automatically, and if a job still cannot succeed it stops and the credit is refunded. The errors that do get through are the recoverable kind, a wrong letter or word that you check against the image shown beside the text, not fluent text the page never contained. When you correct a transcription in Leo, the correction feeds back into training, so the model improves from release to release.

"OCR can read old books"

Conventional OCR engines were built for modern print. On early modern type, their accuracy drops sharply even on a page a person reads easily (Springmann and Lüdeling, 2017):

  • the long s comes out as f
  • ligatures such as ct, ſt and ffl break apart
  • the dictionary correction built into OCR "fixes" period spelling into something worse

The Early Modern OCR Project's final report put the existing OCR of Early English Books Online at roughly 68% correct and Eighteenth Century Collections Online at roughly 86%. A 2026 benchmark of English print from 1612 to 1807 found the best vision-language models making about one character error in forty, against about one in five for Tesseract.

The picture changes as the type does. A clean nineteenth-century book or a typescript is now cheap to read well, with a small model or a conventional OCR engine. An early modern page with long s, ligatures and unsettled spelling is a different job (the Bodleian's guide to reading early printed English explains why), and so is anything you will quote diplomatically. That is the job ATR-1 was trained for: the long s, the macron and the period spelling come through as printed. It is also the first step of ExLatinis, Leo's project to translate the Latin books printed in Europe between 1450 and 1750. ATR-1 transcribes each page, a separate model translates the transcription, and the image, the Latin transcription and the English translation appear together so a reader can check one against the others.

"An LLM can just read it"

General multimodal models from OpenAI, Google and Anthropic will transcribe a historical page with no training, and on later, cleaner hands they can do well (published tests in late 2025 put Google's Gemini ahead of the others on eighteenth- and nineteenth-century English handwriting). Their characteristic failure is the one a historian can least afford: quiet normalization. A general model is trained to predict probable text, so it tends to put a more probable, modern spelling in place of the one on the page. Shew can become show, publick can become public, and an abbreviation can be expanded without a trace. The result reads smoothly, which is exactly what makes it hard to catch. ATR-1 is built not to do this: what you check is a faithful draft, not a tidied one. There is more on what published tests show about general models and on why fluent LLM output misleads.

"HTR removes the need for paleography"

It does not. Trainable handwritten text recognition (HTR) models reach low single-digit character error rates on hands inside their training data. On unseen hands they run far higher, and a trained model typically needs dozens of pages of checked transcription for each new hand. ATR-1 removes that training step, since it reads an unseen hand zero-shot, but it does not remove the reader. Whatever the tool, you need paleographic competence to check the output, resolve the minims and decide which readings to trust, which is why historians still need paleography. The mechanics are set out in how HTR actually works and OCR and HTR.

Choosing a model in Leo, and keeping the source intact

The governing principle is source integrity: a transcription is evidence only if it preserves what is on the page. ATR-1 is the default in Leo's transcription window; the same menu lists general models from OpenAI, Google Gemini and Anthropic Claude for a cheap read of clean print or a second opinion, each run in its own tab. Choose by document:

  • Handwriting, secretary hand included: ATR-1, specialised for historical documents and built to keep the page's spelling
  • Early modern print, or anything you will quote diplomatically: ATR-1
  • A clean, legible later printed page (a nineteenth-century book, a typescript): A small model such as GPT-6 Luna, or an OCR engine, for a fraction of a credit an image
  • A hard reading: Run a general model as a second reading into its own tab and compare the two transcriptions

None of that removes the reader from the loop. It moves the bottleneck from deciphering every character to verifying a draft, which, against a 900-page diary transcribed by hand, is what lets the analysis begin.

The skill that remains yours

Tools change the economics of transcription; they do not change the discipline. That knowledge lets you catch the machine's errors, defend your readings, and understand what a scribe was actually doing on the page:

  • ye is a thorn
  • a January date in 1650s England may mean 1651 to us
  • iiijˣˣ li. is eighty pounds
  • publick is not a typo

Paleography was never only about deciphering marks. It is about reading a document closely enough to know what it is. That competence is what makes a transcription citable, and it stays with the historian, whatever reads the first draft. If you are starting from nothing, the Academy's secretary hand course is a gentle way in.

Frequently Asked Questions

What is early modern paleography?

Early modern paleography is the study of how people wrote between roughly 1500 and 1800, and how to read what they left behind. For a working historian the task breaks into three parts:

  • identifying the hand on the page (secretary, italic, the court hands, French bâtarde, German Kurrent, and their relatives)
  • decoding the letterforms, abbreviations, numerals and dating conventions that no longer survive in modern writing (the long s, the thorn-derived ye, the macron for an omitted nasal)
  • choosing a transcription convention (diplomatic, semi-diplomatic, or normalized) and holding to it consistently

Get those three right and the rest is patience.

Why is the long s so often misread in old documents?

The long s (ſ) is the single most common early modern transcription error, because it looks like an f without the full crossbar and is routinely misread as one. It was used at the start and in the middle of words in English, French and German from roughly 1500 to 1800, but not at the end, where a round s was used, so a final double s is often ſs. English print abandoned it around 1790–1810, and handwriting held onto it a little longer.

Does "ye olde" really mean "the old"?

Yes. The y in forms like ye, yt and ys is a thorn in disguise, standing for the, that and this. Thorn (Þ/þ) had largely given way to th in English by the fifteenth century, but its ghost survived in handwriting and print as these y-shaped abbreviations. So "ye olde" was never pronounced with a y sound; it is simply an old spelling of "the."

Can OCR or an LLM transcribe early modern documents accurately?

They give a useful first draft, not one to cite unchecked, and the tool matters. Conventional OCR, built for modern print, tends to read the long s as f, break ligatures and "correct" period spelling on early modern type. General AI models read more of the page but tend to quietly modernize spelling, putting a more probable word in place of the one written. Leo's ATR-1 is tuned for historical documents, handwritten and printed, and trained to transcribe what is on the page without silently normalizing it; it also detects and retries suspect output. Whatever the tool, a paleographically literate reader still checks the output against the page.

Why does a January date in 1650s England sometimes mean the following year?

England kept the Julian calendar until September 1752, and its legal year began on 25 March (Lady Day) rather than 1 January. A document dated in January, February or March in seventeenth- or eighteenth-century England therefore often carries what looks to us like the previous year. Historians write such dates double, as in 16 March 1592/3. Catholic Europe had adopted the Gregorian calendar from 1582, so the two systems ran in parallel for well over a century. Ignoring the convention can put a correctly transcribed date wrong by a full year.

Where can I learn early modern paleography for free?

For English hands: Leo Academy's free English secretary hand course, Cambridge's English Handwriting 1500–1700, The National Archives' palaeography tutorial, and the Folger's Early Modern Manuscripts Online for practice. For Continental hands, the Newberry Library's French and Italian handbooks and the École des chartes' Theleme.

Share this article