Scriptorium

Methodology

How Scriptorium prepares, transcribes, checks and presents historical charters while preserving uncertainty and provenance.

From source to research record

The workflow proceeds through source metadata and image discovery, image assessment, preparation of a suitable image region and resolution, paleographic transcription, representation of abbreviations and uncertainty, regest and entity extraction, translation, consistency checks and targeted repair of suspicious results.

Transcription and uncertainty

The canonical transcript preserves abbreviated and uncertain readings with compact markup. Parentheses mark expanded abbreviations; paired symbols distinguish degrees of uncertainty, conjectural reconstruction, unreadable text and physically missing passages. Readable views are derived from this canonical evidence without overwriting it.

Automated quality control

A low-cost visual scout estimates the text region and expected extent before expensive transcription. The transcript is then compared with source descriptions, regests and extracted data. Strong discrepancies are retained as review signals instead of being silently normalised.

Limitations

Historical handwriting, damage, faded ink, seals, incomplete scans and catalogue inconsistencies can all affect results. Confidence labels are decision aids, not guarantees. Serious research should cite and inspect the scan and the original archival record.