How to organize your manuscript research photos
A reading-room session can leave thousands of camera files in one folder. Organize them while shelfmark boundaries, retakes, details, and omissions are still documented in notes and memory.
Begin with object identity, then preserve filenames, metadata, rights evidence, and backups as connected records.
Why "I'll remember" fails
A reading-room session does not produce a tidy sequence. It produces a dump: three manuscripts in one afternoon because the second one turned out to be misdescribed in the catalog; overview shots of bindings and spines mixed in with leaf-by-leaf runs; retakes where your shadow fell across the page; frames that are mostly ruler and color bar; a shot of the shelfmark label so you'd know later where one book ends and the next begins. That last habit is the tell: even in the room, you knew you would not remember.
You won't. Camera filenames encode nothing but sequence, and sequence decays fastest. Six months later, the boundary between manuscript two and manuscript three is a guess, and the guess costs an afternoon of squinting at incipits. The moment to impose order is as soon as the images exist, while the session is still in your head, because everything you fail to record now must be reconstructed later from the images alone.
Triage first: sort by shelfmark before anything else
A shelfmark identifies an object within a repository or collection, but the same string can occur elsewhere. Use a composite identity: repository, collection when needed, shelfmark, and a stable catalog URI when one exists. The first pass over a photo dump establishes which images belong to which object and which institution holds it.
Do this while your memory of the session is freshest. Walk the images in shooting order, find the shelfmark labels or the opening leaves, and split the run. Keep the ugly frames for now: the ruler shots, the retakes, the accidental photograph of the desk. Deleting is a later, calmer decision, and a retake you delete today is sometimes the only frame where the marginal note is in focus.
Choose a folder unit that matches the object
For a bound codex, one folder named with repository and shelfmark is a useful default:
Walters_W102BnF_fr1450Bodleian_Ashmole304Cambridge_CUL_Add1000
Fragments, composite codices, disbound leaves, archival series, multispectral sets, and several renditions of one folio may need subfolders or a different unit. Trips and sessions can remain one level above the object folder when capture context matters.
Use only verified information for partial or uncertain identifications. A wrong shelfmark can propagate into filenames, metadata, and citations. A temporary label such as Unidentified_privatecoll_hours_01 keeps the uncertainty visible until the object is identified.
Rename to foliation
Within a manuscript folder, a filename such as W102_024r.jpg is easier to identify than IMG_4771.jpg. Treat the filename as an access label rather than the sole catalog record. Repeated numbers, roman flyleaves, unnumbered leaves, object photographs, details, and repository-specific suffixes may require notes or labels that the filename pattern cannot express. The renaming guide covers the supported workflow.
Put the metadata in the files
Repository, shelfmark, rights evidence, notes, and keywords can be stored in a catalog and selected IPTC/XMP fields. Embedded metadata may be stripped by an export or service, so retain the catalog and any sidecars. The metadata guide covers field meanings and preservation limits.
Record the rights as you go
Record the reader agreement, item-level rights statement, repository terms, license, permission, or unresolved status that applies when images enter the project. A record may apply to a whole manuscript or to named images and uses. Recheck it before publication. The rights-tracking guide explains which evidence and conditions to retain.
Back up like the images are unrepeatable
Research photographs may be expensive or impossible to retake. The 3-2-1 backup pattern keeps three copies on two kinds of storage, with one copy off-site. The working copy counts as one.
A laptop and an external drive kept together remain vulnerable to one event. Include the untouched camera or repository files, working derivatives, sidecars, notes, rights documents, and the catalog database in the backup set. Use checksums where practical and test that a sample can be restored.
When a photo manager is enough
Folders can support a small project when names and records remain consistent. Photo managers add browsing, ratings, and metadata editing; available batch-renaming and IPTC features vary by application and version.
A general photo manager can search any metadata its user records, including rights terms, shelfmarks, and subject tags. It may lack dedicated validation for folio sides, numbering sequences, manuscript relationships, collation, transcription passes, and source-linked figures. Compare those concrete needs with the fields and queries your existing system already supports.
All guides · Related: Renaming manuscript photographs to folio numbers · Embedding shelfmark and rights metadata · Tracking image reproduction rights · Foliation vs. pagination · Recto/Verso vs. Tropy
Recto/Verso: manuscript photo cataloging for historians. macOS, in beta.