Being able to view a manuscript online does not necessarily make its identity clear. A cover title may not describe the contents, and medical material may sit at the back of a volume catalogued under another subject. I compared copies of a Joseon emergency handbook and developed a record that lets a reader return to the evidence behind a proposed textual relationship.
Collation means comparing copies to examine differences in wording and content. For this project, I recorded more than the passages being compared: the source locations, the reasons for linking them, and the judgments that remained uncertain or were rejected became part of the manuscript’s supporting material.
Similar headings do not settle what lies beneath them
I separated the comparison of headings and local order from the comparison of the instructions beneath them. Similar symptom headings in a similar sequence do not establish that the remedies or explanations are also the same. I recorded additions, substitutions, and changes in location alongside the evidence for a corresponding heading.
A unit in one copy can also be split across entries in another. Matching the same position on two pages would miss that possibility. I defined the unit being compared, considered its neighboring entries, and recorded how far the evidence supported the relationship. A shared arrangement was not treated as identity of the entire books.
Keeping rejected readings visible
The collation ledger connects image locations with the evidence and the classification of each proposed relationship. Clear readings are distinguished from judgments affected by an ambiguous character or entry boundary. I also kept proposals that lacked enough evidence, together with the reason for excluding them. This makes disagreement local: another reader can identify the particular passage or decision they would change.
The left- and right-page labels are working locations used to compare digital images, not folio numbers assigned by the holding institutions. Repository identifiers remain attached so the material can be found again. The digital structure supports that return to the source; it does not automatically resolve a difficult reading.
Even a persuasive parallel does not establish which copy was taken directly from the other. An intermediate compilation or shared source may still be involved. This work compares publicly accessible facsimiles. It does not include physical examination of the manuscripts or a test of the effectiveness of their remedies.
Manuscript and disclosure status
I brought the collation and source records into a manuscript submitted to Digital Scholarship in the Humanities on 24 August 2026; it was in review as of 3 September 2026. Neither journal acceptance nor an institutional catalogue correction has been confirmed.
Holding-institution images and specific unpublished readings are not reproduced here. Once a version of record and DOI are confirmed, I plan to explain the comparisons using images or transcriptions that can be shared under the applicable rights conditions.