How to Transcribe an Old Family Diary or Journal

This is a practical guide to transcribing an old family diary or journal: a grandmother's pocket diaries, a farmer's daybooks, a travel journal from an emigrant voyage. It covers surveying and photographing bound volumes without damaging them, dating the entries, reading pencil and private shorthand, writing a transcription of dated entries, indexing the people and places, and deciding what to share.

A diary answers questions no register can: who came to the funeral, when the family moved and why. It is also one of the hardest family documents to transcribe. A diarist wrote for one reader, so the entries are terse, the hand is often cramped into a small space, the names are initials, and the abbreviations are private.

Diaries share a method with all genealogy record transcription: copy what is on the page, and keep your own additions visibly separate. If the family also kept its correspondence, our guide on how to transcribe old family letters covers that side. This guide covers what is particular to diaries.

Survey the volumes before you start

Diaries rarely come as one book. More often there is a run of small volumes, some missing, with loose papers tucked inside, so take stock first.

List every volume with its dates, size, cover and condition, and give it an identifier, such as GRE-D-1871 for the 1871 diary. Write it on a slip or sleeve, never on the diary.
Note what kind of book it is, blank notebook or printed diary, because it changes how you date the entries
Map the gaps. Record whether a missing year is lost or was never written, if you can tell.
Leave inserts where you found them. Photograph pressed flowers, receipts and cuttings in place before you move anything.
Check for overlap. Diarists sometimes kept two books at once, such as a pocket diary and a fair copy written up later.

Photograph bound volumes without damaging them

You will transcribe from photographs, not the book. The general method for photographing archival documents applies: even light, no flash, the camera square to the page. Bound volumes add their own rules.

Never force a binding flat. The US National Archives advises photographing books with a copy stand and book supports, wedges or a cradle, so the binding opens without force, rather than pressing them onto a flatbed scanner. The Library of Congress likewise warns against a book lying flat open at 180 degrees. At home, rolled towels or two stacks of paperbacks under the covers do the job of a cradle. Open the book only as far as it goes comfortably, and photograph from above.

Shoot both pages of an opening together. A small pocket diary opened out fits in one frame, and keeping facing pages together keeps the week in one image.
Photograph the gutter. Writing runs into the fold in a tight binding. Take an extra close shot of the inner margin where it does.
Photograph everything, in order. That means the covers, the endpapers, the printed front matter, every opening including the blank ones, and the back pocket. Endpapers often hold addresses or birthdays.
Fill the frame. Pocket diaries are small and the writing smaller. If you scan loose inserts, see what resolution to scan old documents at.

For pencil, watch the light. Graphite has a slight sheen and can glare under direct light, so move the lamp until the writing reads dark. If a volume is falling apart, photograph it as it is and ask a conservator.

Three dates in every entry: printed, written and real

Diaries look easy to date because every entry has a date. In practice there can be three: the date printed on the page, the date the diarist wrote at the head of the entry, and the day the events actually happened.

Printed diaries. In the nineteenth century, pocket diaries became a common stationery item: publishers put almanac information at the front, followed by pages with a printed date for each day, which limited each entry to a page or less. That printed date is not always the date of the entry.

A diarist could run over. A long day spills into the next day's space, and the next entry starts lower down. Follow the sense, not the printed grid.
A diarist could reuse an old diary. A spare volume for one year was sometimes used for another, with the printed dates crossed out or ignored. If the weekday the diarist writes ("Thurs.") does not match the printed date, check which year it fits.

Blank notebooks and journals. Here the diarist's own date is all you have, and it may be partial ("Thurs. 14th"). A perpetual calendar tells you which years the 14th fell on a Thursday in a given month, which can narrow a run of undated volumes quickly.

Entries written up later. A run of entries in one even hand and one pen can mean the diarist caught up afterwards. Note it: an entry written a week later is weaker evidence.

Before 1752, an English diary may begin the year on 25 March, so entries from January to March carry the previous year's number. A diary kept through 1752 jumps from 2 September to 14 September. The Leo Academy Handbook's note on the calendar, New Year and Old Style dates sets out the rules.

Reading pencil and cramped hands

Pencil. Graphite marks resist moisture, light and ageing but rub off easily. A pencil diary that has been carried in a pocket for a year is usually smudged and worn where the pages rubbed, not faded. Work from the best photograph, zoom in, and adjust contrast on a copy of the image (never the original file); greyscale with stronger contrast often brings worn graphite back. Don't try to clean or fix the pencil on the page itself.

Cramped hands. When the space is small, letters shrink, loops close and words run together. Two habits help:

Build the diarist's alphabet from words you know. Days of the week, months, the weather words that recur daily, and family names each give you letter forms you are sure of.
Use the repetition. Diaries repeat themselves. "Fine", "wet", "to chapel", "to mkt." appear dozens of times, so a doubtful word on one day is often clear on another.

For the general method, see how to read old handwriting. Mark what you still cannot read: "[illegible, 1 word]" in the right place is better than a guess presented as fact.

Private shorthand, initials and codes

A diarist wrote for themselves, so expect three kinds of shortcut.

Everyday abbreviations. Diaries share most of the abbreviations of letters and accounts: recd for received, wd for would, mkt for market, Do for ditto. The family-letters guide lists the common ones, our guide to manuscript abbreviations explains the older marks, and the Leo Academy Handbook has abbreviation pages for early modern forms.

Initials and nicknames. "M. poorly again" or "W. to mkt." is clear to the diarist and opaque to everyone else. Don't expand an initial in the text. Keep a running key of who each initial seems to be, with the entries that support it, and confirm it against registers and census returns before you rely on it.

Shorthand and codes. Some diarists used a published shorthand: Pitman's, first presented in 1837, or Gregg's, from 1888. Both write sounds rather than spellings, so a reader trained in the right system and edition is needed. Others invented private symbols for money, illness or a courtship. Mark these as [shorthand, 3 lines] or [symbol], and list every unexplained sign with the entries where it appears. Patterns across a year often explain a symbol that one entry cannot.

Weather, farm and household entries

Many diaries are mostly routine: the weather, work done, money in and out. These entries date events and show how a household lived.

Weather. Transcribe it as written, abbreviations included ("Fine. Wind N.E."). It can confirm a date against other diaries or local records.
Farm work. Field names and crops ("Sowed oats in Long Close") locate the family on the land and can be matched to maps and estate records.
Money. Keep sums exactly as written, in the units the diarist used, for example "Recd of Mrs B. 2/6" in shillings and pence. Don't convert or total them in the transcription. Put any conversion in a note.

Don't summarise routine entries in the faithful transcription: the one day that matters is often hidden in a month of "Fine. At home".

Travel journals

Travel journals, from an emigrant voyage to a honeymoon tour, follow the same method with a few additions.

Keep place names as spelled. A phonetic spelling of a port or village is evidence of what the writer heard. Identify the place in a note, never in the text.
Track the route. List each stop with the entry date and the place as written. It makes the journal easy to map and to check against passenger lists.
Watch the dates on a long voyage. A day can slip. Note any inconsistency rather than silently correcting it.

How to transcribe a dated entry

A diary transcription should let a reader find any entry and check it against the image. Set out each entry the same way:

Begin each entry with its date as written, then the printed date in brackets if it differs: Thurs. 14th [printed: Wednesday 13 March]
Mark the volume and the image, for example [GRE-D-1871, image 34], so every entry leads back to its photograph
Keep the diarist's spelling, capitals and punctuation. "Recieved" stays as written, and a "corrected" name can send research the wrong way.
Keep line breaks, or mark them with a vertical bar ( | ), so the text can be checked line by line against the image
Show changes of medium. Note [in ink] or [in pencil] where the diarist switched, and [added later] where an addition is clearly in another hand or pen.
Mark empty days. [no entry] tells the reader the gap is the diarist's, not yours.
Mark uncertainty. Use [?] after a doubtful reading and [illegible] for what you cannot read.

Editors use different signs for the same thing: one edition's square brackets are another's deletions. Write a short key at the top of each volume and keep to it. Our guide to transcription conventions for old documents covers the choices in more depth.

Index people, places and dates as you go

Build the index while you transcribe, not afterwards.

People. One line per mention: the name or initial as written, who you think it is, and the entry date. Keep "as written" and "identified as" in separate columns.
Places. Farms, chapels, towns and ports, with the spelling as written and the modern name beside it.
Events. Births, marriages, deaths, moves, illnesses and purchases, each with the entry date. This list is the one most relatives will want.

Then test the diary against the records: a birth should match a baptism, a death a burial. Our guide to transcribing parish records covers the other side of that check.

Sensitive content and family privacy

Diaries record what letters left out: illness, debt, grief, quarrels, sometimes secrets. Decide how to handle them before you share anything.

Transcribe everything in the faithful version. Leaving out a painful entry in the record itself loses evidence and is hard to undo.
Consider the living. Entries about people who are still alive, or their parents, deserve more care than entries about the long dead. When in doubt, ask the family member most affected.
Make a reading version for sharing. A reading version can leave out or summarise sensitive passages, provided it is labelled as a selection and the faithful transcription is kept.

Most families want something they can read, not a scholarly edition. Keep the faithful transcription as the record, and make a reading version with modern spelling, abbreviations expanded and short notes on who people were. Label both. A family book can add photographs of a few openings and the index of people and places.

Where Leo fits

Twenty years of pocket diaries are a project, and a first machine transcription gets you to the reading much faster. Leo is built for that work.

Its transcription model, ATR-1, is specially tuned for historical documents. It is trained to transcribe what is on the page rather than tidy it, so the diarist's spellings, capitals and abbreviations are kept, not quietly modernised the way general-purpose AI models tend to do. In a diary, a phonetic place name or an odd spelling of a neighbour's name is evidence. ATR-1 reads the image at high resolution, where general models downsample it, and needs no training on the diarist's hand. Built-in failure detection hides output that looks wrong, retries it, and refunds the credit if it cannot succeed. The errors that do get through are the recoverable kind: a wrong character or word that you check against the image shown beside the text.

The workflow fits a run of diaries:

Upload photographs straight from your phone, scans or PDFs. An opening of two facing pages photographed together counts as one image.
Organise one document per volume in a folder for the diarist, with the creator, the dates and your identifier in each document's metadata
Search every volume at once, with fuzzy search to catch a name the diarist spelled three ways
Start the index. The Extract named entities Transformation lists the people and places in a transcription in its own annotation tab. Check every name it finds against the text.
Keep layers separate. The Modernize Transformation writes a modern-spelling reading version into its own tab, and the faithful transcription is never overwritten.
Export and share. Export to Word, PDF, HTML or TEI for a family book. A read-only share link needs no account, and you choose which tabs it shows. It is a public link, so share a reading version rather than the whole diary if parts are private.

Leo never sells user data, and it is UK and EU GDPR compliant; the privacy policy has the details.

Whatever produces the first draft, check it against the image, and give names, dates, places and sums the closest look. Our method for verifying transcription accuracy shows how. Every correction you make in Leo feeds back into training, so later readings improve.

Leo has a free plan, so you can start with one volume and see how far a first transcription takes you.

Frequently Asked Questions

How do you transcribe an old diary?

Give each volume an identifier, photograph every opening with the binding supported, never forced flat, and work from the images. Date each entry carefully, since the printed date, the diarist's date and the real day can differ. Copy each entry exactly as written, keeping spelling, abbreviations and line breaks, and mark anything you add or cannot read. Then check the transcription against the image, especially names, dates and sums.

How should I handle the printed dates in a pocket diary?

Treat the printed date as a guide, not a fact. Diarists ran long entries into the next day's space, and sometimes reused a spare diary from another year. Head each entry with the date as the diarist wrote it, add the printed date in brackets where it differs, and check the weekday against a perpetual calendar if anything looks wrong.

How do you read a diary written in pencil?

Pencil rarely fades, but it smudges and rubs off where pages have touched. Photograph with the light at a slight angle so the graphite doesn't glare, then zoom in and raise the contrast on a copy of the image. Learn the diarist's letter forms from words that recur, such as weekdays and weather words, and compare doubtful words with clearer entries elsewhere in the diary.

What if the diary uses shorthand or private abbreviations?

Keep a key as you go. Note every abbreviation, initial and symbol with the entries where it appears, because patterns across a year often explain them. A published system such as Pitman or Gregg needs a reader who knows that system. Mark untranscribed passages clearly, for example [shorthand, 3 lines], rather than guessing.

Transcribe the whole diary faithfully, then decide what to share. Entries about living people, illness, money or family quarrels deserve care, so ask the relatives most affected. A labelled reading version can leave out or summarise sensitive passages while the faithful transcription is kept as the record.

Can AI transcribe an old diary?

Yes, as a first draft that you then check. General-purpose AI models tend to quietly modernise a writer's spelling, which loses evidence a family historian needs. Leo's ATR-1 is specially tuned for historical documents and trained to transcribe what is on the page: it keeps the diarist's spellings and abbreviations, reads the image at high resolution, and needs no training on the diarist's hand. Output that looks wrong is hidden, retried and refunded if it cannot succeed. Check every entry against the image beside it.