The work on this site rests on documents: letters, registers, certificates, official files, photographs, and the papers my family kept. This page explains how I read them, how I present them, and how I tell what a record says apart from what I infer from it.
Records, memory and inference
I draw on three kinds of source: records made at the time, the family’s own memories, and published scholarship. I keep them apart. Where a record says something, I cite it. Where the family remembered something, I say so and quote them. Where I am inferring, I say that I am, using words such as “probably”, “perhaps” or “we can picture”. When two sources disagree, I set the disagreement out rather than choosing the version that reads more smoothly. Questions I cannot yet answer are listed as open.
Transcription
Transcriptions keep the writer’s own spelling, grammar and punctuation, including mistakes. Nothing is silently corrected. Uncertain readings are marked [?], illegible words are marked as such, and anything I have added for clarity appears in square brackets. Struck-through words and underlining are shown where they matter to the reading.
Most of the German letters are written in Sütterlin, the old German script taught until 1941. They were transcribed by the volunteer transcribers of the Sütterlinstube Hamburg e.V., and their readings are the basis of the German texts here. They also transcribed a set of letters in a fast Latin-script hand, which they took on although they normally work only with the old scripts. Letters and documents in ordinary handwriting or typescript were transcribed with AI from the original scans and checked by me; where a family member who reads German has corrected one, their reading stands.
Translation
German texts appear beside their English translations, paragraph by paragraph, so the two can be read across. The translations aim to be faithful rather than literary. Where the original is broken or unclear, I say so in a note rather than smoothing it in English.
I read only a little German and no Hungarian. Most of the translations on this site and in the books were made with AI, and each book says which. Some are my own. Printing the original beside the English is deliberate: anyone who reads German can check every line. Where I can, I have passages that turn on idiom or tone checked by German speakers in the family.
Oral history
Family recordings in German were transcribed and translated with AI (WhisperAI). I treat those transcripts as a guide to what is on the recording, not as quotations. A line is quoted only after it has been checked against the audio by someone who speaks German.
Citations
Claims are cited to their source: archive and collection references, register entries, census pieces and folios, newspaper titles and dates. Online family trees are used as leads, and are cited only where they are the sole source. Where a tree and a register disagree, I follow the register and note the difference. Images carry their source and permission.
How I use AI
I use AI as a tool, and I say where. Across this site and the books it has:
- transcribed typed and plainly handwritten documents, and recorded conversations;
- translated German and Hungarian into English;
- helped draft and structure text from my research notes and the documents;
- helped compile lists of names and places from the letters, such as the Who’s Who sections; and
- checked my drafts for inconsistencies against the sources.
The tools were Claude (Anthropic), using several model versions, 2026, and WhisperAI for recordings. The research itself, finding and reading the records, deciding what they show and what to publish, is mine, and so are the conclusions and any errors. My 2020 biographies of Georg and Charlotte Schliewen, from which several books grew, were written before I used AI tools. I’m still learning how to use these tools well, and I’ve looked for guidance from people who think carefully about it: the Coalition for Responsible AI in Genealogy’s published principles, Lynn Broderick’s Guidelines for the Responsible Use of Artificial Intelligence in Genealogy, and the Verein für Computergenealogie’s articles on AI in family research. Where their advice and my practice differ, I’d like to hear about it.
Living people
Material about living people is published only with their consent, and some family members are deliberately not named.
Corrections
This is research in progress, and archival inquiries are ongoing. If you have a correction, a document, or knowledge of anyone named here, please get in touch through the Contact page.