Can one page contain both Gothic and Latin script?

Constantly, and that is the normal case in this period. Each passage is recognised on its own terms and labelled, so a Gothic remark and a Latin-script name in the same cell are both transcribed correctly instead of one of them being forced into the other alphabet.

How are å, ä, ö, æ and ø handled?

They are transcribed as written, including the older forms where a superscript letter stands in for the ring or the dots. Where a mark is ambiguous the reading is flagged rather than resolved from modern spelling.

What happens to the printed headings on a form?

They are read once and used to structure the output, so the values come back under the right field names. They are not repeated for every row, and they are never reported as something written by hand.

Are ditto marks and abbreviations expanded?

Only if you ask, and then with a mark showing that the value was inferred. Occupational and status abbreviations can be kept as written and expanded in a separate field, which is what most database projects want.

Does the same approach work on printed newspapers of the period?

Yes. Fraktur headings over roman-type columns are recognised as two founts, and the column structure is preserved, so an official gazette can be read as text with its sections intact rather than as one long block.

Swedish · c. 1750–1900

Two scripts, one writer, one line

Between about 1750 and 1900 a Swedish or Norwegian clerk could write the body of an entry in Gothic cursive and every personal and place name in Latin round hand, switching alphabet in the middle of a sentence and back again. Print did the same thing more slowly, keeping Fraktur for headings long after the text beneath had gone over to roman type.

Upload a page

Sample pages

Printed page in blackletter type with the heading Norges Riges Grundlov above a first paragraph beginning Kongeriget Norge — Sample 1. The opening of the Norwegian constitution as printed in the official edition of 1814, with the November amendments. National Library of Norway. Both the heading and the paragraph are set in Fraktur, with the letter-spaced words Norge and Sverige standing out; the section number is given in roman capitals.

Opening of a Swedish household examination roll with printed column headings and entries listing a family, their birth years and remarks — Sample 2. Household examination roll of Brålanda parish, Sweden, for 1813–19. The printed headings — Födelse, Läser, Inflyttad, Utflyttad — stand over entries written in a hand that keeps Gothic forms for the remarks and uses Latin letters for the names; the right-hand columns record attendance at communion year by year.

Namn	Födelse Året	Födelse Orten
[cancelled: Brukare af Lottbytorp [?]]
Anders Norling	1782
H. Maria Nilsdr	1783.
S. Olle	1802	adm
D: Cherstin	1807.
D: Kristin [?]	1810.
S. Eric	1813
D. Märta	1817.
[cancelled: Inhys. Pig. Anna Olofsdr [?]]	[cancelled: 1798 [?]]	[cancelled: adm]
Inhys. Pig. Maria Olsdr	1795	adm
½ Pär Börjesson	1786
H. Anna M: Hansdr	1788
[cancelled: D: Maja Stina]	1813
[cancelled: D: Maria]	1814
S. Anders	1817.
[cancelled: [illegible]]	[cancelled: 1803 [?]]
Moder Enk. [illegible] Andersdr	1764 [?]	adm
Gof. And: [cancelled: Andersson]	1799.	adm

Masthead and first columns of an 1836 Swedish newspaper, with the title in roman capitals and section headings in blackletter over columns of roman type — Sample 3. Sveriges Stats-Tidning, or Post- och Inrikes Tidningar, of Friday 17 June 1836. The masthead and the news columns are set in roman type; the date line and the section headings Officiella Afdelningen and Icke-Officiella Afdelningen keep Fraktur, which is exactly how the change of type ran in Sweden.

Left page of a printed register book of 1849 with ruled columns and a family listed in a fine sloping Latin hand — Sample 4. Personal book of the Swedish parish of St Catherine in St Petersburg, 1849, with the family of Immanuel Nobel among the households. Riksarkivet, Stockholm. The book is a printed form — Stånd, Namn och Kön, Födelse, Kommen ifrån, Anmärkningar — and the entries are written in a small Latin round hand, with the confirmation and communion columns beginning at the right-hand edge of the page.

Ruled census page of 1865 listing residents of a Christiania address with their occupations, marital status, ages and birthplaces — Sample 5. The Norwegian census of 1865 for Kristian Augusts gate 17–19 in Christiania, the house where Gina Krog lived. Digitised by the Digitalarkivet. Names and birthplaces are in a Latin round hand, while abbreviations such as ugift and ditto keep the shapes of the older cursive, and every column is filled to the edge of the ruling.

A century and a half of changing over

The change from Gothic to Latin script in Scandinavia was not an event but a long overlap. In the eighteenth century the everyday hand is still Gothic cursive, but writers already reach for Latin letters — latinsk skrift — when they write a proper name, a foreign word, a title or a technical term, because those are the things a Gothic hand renders badly. In the nineteenth century the proportion tips: names and headings are Latin, then whole entries, and by the 1870s most new registers are written in a round Latin hand with only a few Gothic habits surviving in the abbreviations.

The two systems have to be read by different rules. In Gothic, e, n, u and r are the same short stroke and a small bow distinguishes u from n; long ſ reaches below the line and is easily mistaken for f; h, k and l loop above it. In Latin round hand the e is open, c and e are close together, and the capitals I, J and S are drawn with sweeping flourishes that can carry over the neighbouring word. The vowels å, ä and ö, and Norwegian æ and ø, appear in both systems, sometimes with the ring or the dots and sometimes as an a with a superscript letter left over from older practice.

The paperwork changes shape at the same time. Household examination rolls, parish registers, censuses and emigration lists move from ruled columns drawn by hand to printed forms with fixed headings, then to forms printed with a heading in Fraktur and filled in Latin script. Newspapers and official gazettes make the same move: by the 1830s a Swedish state newspaper sets its news in roman type and keeps Fraktur for the standing section headings. Notes are added to older entries for decades afterwards, so one register page can contain a Gothic entry from 1815, a Latin-script correction from 1850 and a stamp from the 1880s.

Why ordinary OCR struggles here

Alphabet changes inside a line

A single entry may run: Gothic for the occupation, Latin for the name, Gothic again for the remark. Recognition that assumes one script per page produces gibberish at every switch, and the switches fall precisely on the names an index is built from.

Two sets of confusable letters at once

Gothic contributes the e/n/u/r problem and long ſ against f; Latin round hand contributes flourished capitals and open e against c. A page with both needs the right resolution rules applied to the right stretch of writing.

Printed forms with handwritten fillings

From the middle of the nineteenth century most of these records are forms. The printed headings must be read once and used as structure, not repeated as if the clerk had written them, and empty cells have to be reported as empty rather than filled from the row above.

Layers of later additions

Household rolls were kept in use for years and annotated continuously: a move, a death, a certificate of literacy, a stamp. The additions are in different inks, different scripts and different hands, and a transcription that merges them into the original entry invents a family history.

Who works with this material

Genealogists working nineteenth-century records

Household rolls, censuses and emigration lists are the standard route from a living family back to the 1700s. What matters is that the names come out in the script they were written in and that the columns stay aligned, because an age or a birthplace in the wrong row sends the research down a false line.

Emigration and population historians

Moving-out lists, passenger records and parish notes track a whole generation across the Atlantic. Working at that scale needs consistent transcription of place names and dates from forms that changed layout every few years.

Editors of nineteenth-century correspondence

Letters, diaries and drafts from this period mix Gothic and Latin script according to the writer’s schooling and mood. An edition has to record which script each passage is in, because the choice often marks a quotation, a name or a shift of register.

Archives publishing digitised registers

The Swedish and Norwegian national archives have put these series online in full. Adding searchable text to them, with the printed headings used as field names, is what turns a shelf of scanned volumes into a database people can query.

Getting the best transcription

Step 1

Warn the model that scripts alternate

Say that Gothic cursive and Latin round hand may appear in the same line, and ask for each stretch to be labelled with the script it is written in. That label is also useful evidence later, when you want to know which parts of an entry the clerk treated as names.

Step 2

Use the printed headings as the schema

Give the column headings from the form and ask for one row per person, with empty cells reported as empty. On a census or a household roll this turns a page into records that can be loaded straight into a database.

Step 3

Ask for later additions to be dated where possible

Request annotations, stamps and corrections as separate items, with a note on ink and script. In a household roll kept in use for twenty years, that separation is what stops a later move from being read as part of the original entry.

Step 4

Test on the messiest page in the volume

Pick an opening with crossings-out, marginal notes and a change of clerk. If the transcription keeps the rows straight and marks what it cannot read, the clean pages will need no more than a glance.

Frequently asked questions

Can one page contain both Gothic and Latin script?: Constantly, and that is the normal case in this period. Each passage is recognised on its own terms and labelled, so a Gothic remark and a Latin-script name in the same cell are both transcribed correctly instead of one of them being forced into the other alphabet.
How are å, ä, ö, æ and ø handled?: They are transcribed as written, including the older forms where a superscript letter stands in for the ring or the dots. Where a mark is ambiguous the reading is flagged rather than resolved from modern spelling.
What happens to the printed headings on a form?: They are read once and used to structure the output, so the values come back under the right field names. They are not repeated for every row, and they are never reported as something written by hand.
Are ditto marks and abbreviations expanded?: Only if you ask, and then with a mark showing that the value was inferred. Occupational and status abbreviations can be kept as written and expanded in a separate field, which is what most database projects want.
Does the same approach work on printed newspapers of the period?: Yes. Fraktur headings over roman-type columns are recognised as two founts, and the column structure is preserved, so an official gazette can be read as text with its sections intact rather than as one long block.

Related scripts and pages

Danish-Norwegian Parish, Court, and Administrative Gothic, c. 1700–1875The Gothic hand these records are moving away from, in the same series of books.Swedish Neo-Gothic and Fraktur Administrative Hands, c. 1500–1750Where the Swedish side of this story starts, in charters and Fraktur print.Sütterlin and Late Kurrent, c. 1911–1941How the same change played out in German, where the Gothic hand survived a generation longer.All Swedish and Danish-Norwegian handsEvery domain from medieval charters to nineteenth-century parish books.Image to TextSearch all supported languages, scripts and document types.

Transcribe a register that changes script

Upload a household roll, a census page or a nineteenth-century letter and get the Gothic and the Latin passages labelled and kept apart.

	Stånd, Namn och Kön.	Födelse Ort.	Födelse År och dag.	Vigd, Enkl. Enka eller skiljd.	Kommen ifrån.	Bokläsning.	Christend. kunskap.	Anmärkningar och hänvisningar till andra Kyrkoböcker.	Confirmerad.	Transport	1841
1844. 170	Mechan. Hultman, Carl Petter	Djurö [?] Sverige	14/6 813.		St. Cl. [?] 43, 44; 47.				C.
	Apoth. El. Wickman, Carl Joh. Gust.	S. Borg [?]	18/2 827.		S. O. [?] 42. 44; 64.			t. Wib. [?]	C.		15/2
	Apoth. El. Harthin, Carl Lud. Theod.	[illegible]	22/7 826.		S. O. 44; 65.				44. 64.
	Apoth. El. Andersin, Carl Ulric	[illegible]	22/5 826.		[illegible] 44; 35. 36. 46; 288 [?]			t. M[?]	C.
	Apoth. L. Rungain [?], Joh. Ludv. [?]	Parikkala [?]	1/5 815.		S. O. 40, 44; 33.				C.
	Elev i Lyceum. Steven, Alexis	Christinestad [?]	825		40; 118.				C.
419	Mechan. Nobel, Immanuel		24 [?]/3 801.	18/[?] 27	St. Pbg [?] 42, 49; 248.				C.
	H. Ahlsell, Andr. Carol.		803		D:o				C.
	S. Emil Oscar	St. C.	29/10 843.
	S. Robert								45; 33.
	S. Rolf	St. C.	20/10 845.
	S. Ludvig		27/7 831.		St. Pbg [?] 42. 49; 248.				47; 62.
+	D. Betty Carolina Charlotta [?]	St. C.	[?] 849.
	S. Alfred Bernhard		21/10 833.		D:o				50; 65.
	Bokhåll. [?] Låser [?], Johan			6/6 44.				t. S. [illegible] L. 44; 40.
	H. Hogmark [?], Sophia Euphros.	St. C.	7 [?]/7 817.			X	X	[illegible] p. 234 [?]	38; 38.	3/2 37.	13 [?]/2

1	[illegible] Krog	logerende, Leutnant	ugift	23 1/2	[cancelled: 22]	Nordrehoug Ringerike
	Anna Krog	Pensionist Huusmoder	Enke		52	Throndhjem
	Gina Krog	Datter	ugift		19	Flakstad Præstegjeld
	Fredrik Krog	Søn Student	ugift	22		dito
	Diesen	logerende Leutnant	ugift	24 1/2		— ?
	Oskar Tibring	dito Student	ugift	19		Selbo Præstegjeld
	Marie Schmith	logerende	ugift	[cancelled: 18]	18	Levanger
	Gina Didrichsen [?]	logerende	ugift		15	Øvre Telemarken
	Dikka Sejer [?]	logerende	ugift		15	Christiania
	Malia [?] Sejer [?]	logerende	ugift	[cancelled: 11]	11	Gausdal
	Sophie Andersen	Tjenestepige	ugift		28	Vaaler [?] Præstegjeld
	Johanne Caspersen	dito	ugift		35	Drammen
1	Henriette Wegner	Kone Huusmoder	Enke		61	Hamburg
	Anna Henriette Wegner	Datter	ugift		25	Christiania (Aker)
	George Mygind Wegner	Søn Student	ugift	19		dito (Aker)
	Henriette Catrine Bidenkap	Pleiedatter	ugift		42	Drammen
	Annette Olsen	Tjenestepige	ugift		30	Ringsaker Hedemarken
	Berthe Olsen	dito	dito		22	dito
1	Fritz Wedel-Jarlsberg	Huusfader - Capitainlieutenant	gift	47		Fredriksværn (Fredriksværn)
	Julie Wedel-Jarlsberg	Huusmoder - Postmesterinde	do		47	Christiania
	Juliane Wedel-Jarlsberg	Datter - Skolepige			13	do
	Fritz Wedel-Jarlsberg	Søn - Skolegut		11		do
	Frantz Jacobsen	Tjener	Ugift	27		Askim Smaalenene
	Susanne Gulbrandsen	Tjenestepige			32 [?]	Christiania
1	Ingeborg Larsen	Tjenestepige			40	[illegible]

Two scripts, one writer, one line

Sample pages

Evidano Transcription

Evidano Transcription

Evidano Transcription

Evidano Transcription

Evidano Transcription

A century and a half of changing over

Why ordinary OCR struggles here

Alphabet changes inside a line

Two sets of confusable letters at once

Printed forms with handwritten fillings

Layers of later additions

Who works with this material

Genealogists working nineteenth-century records

Emigration and population historians

Editors of nineteenth-century correspondence

Archives publishing digitised registers

Getting the best transcription

Warn the model that scripts alternate

Use the printed headings as the schema

Ask for later additions to be dated where possible

Test on the messiest page in the volume

Further reading on this hand

Frequently asked questions

Related scripts and pages

Transcribe a register that changes script

Company

About

Preview Evidano

Account

Newsletter