About this edition
What the edition contains, where the documents are held, how the text was made and how far it can be relied on.
What this edition is
Traces in the Sand publishes the papers that survive on a group of estates around Trąbczyn and Zagórów, in the part of Poland that Prussia annexed in 1793. In 1796 the King of Prussia granted the estates to one of his generals, Prince Friedrich Ludwig of Hohenlohe-Ingelfingen. Some of them had been confiscated from a Polish nobleman, Antoni Prusimski. Hohenlohe-Ingelfingen’s agent, Peter Friedrich von Triebenfeld, managed the estates and divided them among settlers. Hohenlohe-Ingelfingen lost them in 1807, and the disputes over them and over the debts he had secured on them ran on until 1832. The story tells this in order.
The edition holds 613 documents on 2,005 pages, dated from 1173 to 1936. The oldest are medieval charters of the abbey of Ląd and of the knights of Trąbczyn, from the printed edition of the charters of Greater Poland. Then come entries from the court books of Konin and Kalisz on the quarrel between Trąbczyn and its neighbour Łukom, when Trąbczyn belonged to the Prusimski family: a commission ruled on the boundary in 1775, and the Crown Tribunal settled it in 1782. The documents are letters, royal grants, leases and contracts, court papers, and the files of Prussian ministries. For each document it gives:
- the text in its original language, transcribed from the manuscript or the print;
- an English translation;
- a short summary, in English and in German;
- the image of each manuscript page, beside its text.
People, places and estates are indexed across all the documents, so that one person can be followed from a letter to a contract to a court file. Words of the period that need explaining are marked in the text and defined in the Glossary.
Where the documents are held
The documents come from forty-five holdings in six archives and a library:
- the Hohenloher Zentralarchiv Neuenstein: Triebenfeld’s correspondence with Hohenlohe-Ingelfingen and his officials, 1798 to 1816 (318 documents); two volumes of grants, title deeds and leases; and a royal charter of 1805;
- the Geheimes Staatsarchiv Preußischer Kulturbesitz in Berlin: files of the Prussian administration of its Polish provinces, of the foreign ministry and of the state treasury;
- the State Archive in Poznań: entries from the court books of Konin and Kalisz on the boundary between Trąbczyn and Łukom and the lawsuits around it, from 1589 to 1788 (21 holdings, 92 documents); thirteen pages of the hearth-tax register of the Kalisz voivodeship for 1789; a court record of 1806; copies made about 1930 of papers on the division of the estates; a printed summons of 1860 in a lawsuit between the owner of Trąbczyn and its colonists; and two notebooks of 1936 in which the historian Albert Breyer wrote down what the German settlers of Łazińsk and the places around it told him, and a third, of 1935 and 1936, into which he copied the Russian tables of 1868 that gave the settlers of Łazińsk, Emilienheim and Orliny their land;
- the Central Archives of Historical Records in Warsaw: two files of 1807 on the return of the confiscated estates to Prusimski’s daughter; and six pages of a Prussian table of the villages around Konin, of about 1797;
- the State Archive in Łódź: the liquidation table of 1868 by which the colonists of Łazińsk received their land, its copy of 1869, and the entries to 1901 on their rights in the manor forest, their exchange for land and the survey of 1896;
- the Diocesan Archive in Włocławek: nine pages of the visitation of the Konin deanery of 1790 and 1791, on the parish of Trąbczyn and the souls of the neighbouring parish of Zagórów;
- the Kórnik Library of the Polish Academy of Sciences, in its digital copy: a report in the Warsaw newspaper Kuryer Polski of 1754 on the funeral of Paweł Prusimski at Trąbczyn, and twenty-one entries of the Kodeks Dyplomatyczny Wielkopolski, the printed edition of the medieval charters of Greater Poland, from 1173 to 1440, given from its electronic text, without page images.
Sources lists every holding with its archive reference. Each holding has a page of its own that describes what it contains and how its text was prepared.
Most of the documents are in German. Twenty-two are in French, nineteen in Polish and a hundred and seven in Latin, the court entries often with passages in the other of the two, and some of the German deeds have passages in Polish and Latin. Each is translated from its own language.
How the text was made, and how far to trust it
The originals are handwritten, most of them in Kurrent, the German cursive of the period, which is hard to read and easy to misread.
The editor is an amateur palaeographer. Because of the number of pages, the work for this edition was done with the support of AI, in three steps:
- An AI system for reading handwriting produced the first transcription. The editor corrected names and layout by hand.
- An AI model (Claude) then went through the text and corrected it. A word was changed only where something other than the sense of the sentence supports the change: a second copy of the same text, the same word written clearly elsewhere, a fixed formula, or, in the holdings added most recently, the page itself, compared with the text line by line. Each change is recorded with its reason.
- The editor checked samples of the changes against the scans and ruled on readings that remained in doubt.
Three things were left alone. The spelling of the period stands as written (laßen, seyn, nöthig). Words that could not be read are marked as doubtful and not guessed at. Text that a writer struck out is not transcribed.
One holding differs: the boundary decree of 1775 is given in modern Polish spelling, as the editor transcribed it. Its page says how far it was checked. In the Latin entries from the court books the clerks’ abbreviations are written out.
One thing was changed throughout: a person or place that the writers spell in several ways is given in one spelling, so that it can be found.
Reading handwriting by machine gets a share of the characters wrong. Going through the text again with a language model, which weighs each word against its context, removes many of these errors but not all of them, and it can leave plausible German that is not what the page says. Some inaccuracy therefore always remains. Given the number of documents, the editor has accepted that for this project.
Anything taken from these texts for academic work should first be checked against the original document. The image of each page is shown beside its text.
Reading a document
The buttons above the text switch between three views:
- Transcription follows the manuscript line for line. The lines are numbered, so a passage can be cited by document, page and line. A few documents are transcribed by paragraph, and say so.
- Reading is the same text run together as prose, with words that were broken at a line end joined.
- English is the translation, page by page.
The image of the page stands beside the text; select it to see it at full size.
Two kinds of text are set apart on a page under a label: notes that an office wrote on a paper it received (the date it arrived, a file number, an instruction), and text written sideways in a margin.
| Mark | Meaning |
|---|---|
[?] |
A word, or part of one, could not be read. Straight after a word, it marks that word as doubtful. |
[word?] |
A reading offered with doubt. Where only the first letters are read: [Rev…?]. |
[...] |
A gap: the page is damaged, or the text breaks off. |
G[oldbeck] |
Letters in square brackets inside a word are supplied by the edition: an abbreviation written out, or a signature that consists of an initial. |
ſ |
The long s, as written. |
(missing), (skipped) |
The archive’s number exists, but no text survives under it. |
There are 883 marks of doubt in the edition. Each document’s page says how many it has.
The English translations
Every document that has text has an English translation: 608 of the 613. They were made by an AI model under rules set by the editor, except those of the boundary decree of 1775 and of the entries from the court books, which are the editor’s own, revised where the text was corrected against the scans.
The translation shows a doubt wherever the original has one, because smooth English would hide it:
| In the original | In the English | Meaning |
|---|---|---|
[?] |
[illegible] |
nothing could be read |
[word?] |
[uncertain: word] |
a reading offered with doubt |
[...] |
[text lost] |
a gap, or a damaged page |
Figures and names are translated as they stand in the transcription, even where they look
wrong; the translation does not correct the text in passing. [uncertain: …] can also
appear where the original shows no mark: the transcription reads as ordinary German there
but makes no sense in its sentence, and was probably misread.
The summaries
Each document has a short summary to help a reader find it. It is a finding aid and not part of the text. The summaries were drafted by an AI model. In every holding they were written from the original language and each statement was then checked against the document; the page of each holding says how its summaries were made.
Dates, names and money
Dates. About nine documents in ten carry their own date. The others are dated by the editor from the manuscript, from the letters on either side, or from a second copy, and each document says which. Nine remain undated.
People. People lists everyone named often enough to follow, with the documents that name them. Some identifications rest on context alone; each entry says what it rests on.
Places. A place now in Poland is listed under its Polish name, with the German name in parentheses, and has its Polish name in the English translations. The original keeps the name its writer used.
Money. Sums are in the Prussian money of the time: 1 Reichsthaler (Rthl) is 24
Groschen (g), and 1 Groschen is 12 Pfennig (d, for denarius). The French documents
count in écus, which are the same thalers. The Polish documents also count in Polish
florins (złote), six to the thaler.
How the documents are numbered and grouped
A document is cited by its archive, its holding and its number within the holding. The numbers are the archive’s own and are not always in order of date.
Where one archive number covers several documents, they are numbered 72a, 72b and so on. A few documents survive in two copies. Both are given, because each copy shows where the other was miscopied or misread.
The edition is arranged by the owners of the estates. Nearly all the documents published so far belong to the years of Hohenlohe-Ingelfingen and their aftermath. One belongs to the earlier boundary dispute between Trąbczyn and its neighbour Łukom, and more are to follow. The years after the estates were returned to the Prusimski family in 1807 are still to come.
What is still open
- The marks of doubt wait to be checked against the originals.
- Thirteen letters of the correspondence are still largely an uncorrected machine reading. Their pages say so.
- Further files on the estates in the archives at Berlin are still to be added.
Citing and reusing the edition
Licences, credits and privacy says who runs the edition, how to cite a document, and what may be reused.