Digitise an archive people can actually search and trust

Scanning turns paper into image files. Building a useful digital archive also requires selection, order, description, quality checks and continuing care. A folder containing thousands of files named “scan” may be harder to use than the boxes it replaced.

Start with the retrieval question: what should someone be able to find, and what will they need to understand when they open it? A family collection, a project archive and an accounts archive have different priorities and access rules.

Preserve the source before improving access

NARA advises careful handling and keeping originals after digitisation because digital files have preservation risks too. Fragile items and bound material need a capture method that does not damage them. NARA: digitising papers and photographs.

Give each paper series a retention decision: keep the originals, transfer them to storage, or dispose of them after the applicable retention and approval requirements are met. Record who approved the decision and when; finish the digital quality check before carrying it out.

A worked pilot: one box, twenty representative items

A fictional collection contains 240 items in one box. Select twenty for a pilot: ordinary pages, a faint handwritten note, a photograph with writing on the back, a folded drawing and a multi-page document. The purpose is to discover difficult cases before setting the full workflow.

Record the box and folder order before removing material. Assign a stable item identifier, with an optional local code such as “Box 01, item 017”. Capture fronts and backs where they contain information, and keep pages in their original sequence. Do not assume a blank-looking back is irrelevant without checking it.

For the folded drawing, choose equipment and support that capture the whole sheet without forcing or damaging it. A fragile bound volume should not be flattened simply to fit the scanner.

Choose settings by the information to preserve

Test whether fine handwriting, faint marks, colour annotations and small dimensions remain legible at normal and enlarged viewing. The required resolution depends on the original and the intended use; one universal scan setting is not a quality policy.

Retain a high-quality master and create smaller access copies where useful. Record the capture settings and any processing applied. Aggressive cleanup can erase faint pencil marks or make a damaged original look deceptively complete. Keep an unaltered capture when an enhanced derivative is produced.

The Library of Congress discusses retaining a suitable master and creating sharing copies, while noting that an existing JPEG need not be converted merely to change its extension. Library of Congress: personal digital archiving questions.

Describe enough to make retrieval work

Use a small consistent set of fields: identifier, title or description, date or estimated date, creator where known, collection location, access restrictions and related items. “Letter about roof repair, approximately 1984” is more honest than inventing an exact date.

Keep the connection between the digital file and the physical location. If folders are later rearranged, update the location register without changing the historical description. Separate what is written on the document from an archivist's interpretation.

Filenames help, but the index should carry longer descriptions and uncertainty. Avoid names that expose personal information unnecessarily when files are shared outside the intended audience.

Treat OCR as an aid, not a transcription guarantee

OCR can make typed documents searchable and reduce retrieval effort. Check names, dates, reference numbers and quantities against the image. A single wrong digit can send a reader to the wrong invoice or drawing.

Retain the page image alongside extracted text. If you correct OCR, record the correction method and avoid overwriting the original image. Searchability does not establish PDF accessibility; reading order and structure require separate attention. See Making a PDF usable.

Quality control before scaling up

Check every item's page count, sequence and completeness against the source. Review image quality across each batch and inspect every critical or difficult item in detail. Reopen saved files from their destination; a successful scanner message does not prove the archive copy is readable.

Keep a rescan queue with reasons, rather than silently accepting cropped margins or missing backs. Have another person retrieve several items using only the index. If they cannot find them, improve the descriptions before scanning the remaining 220 items.

Plan continuing custody

Maintain backed-up copies with different failure exposures, restrict access appropriately and test restoration. A checksum helps detect changed bytes; it does not prove the scan contains every original page. Assign someone to review storage, formats and the index over time.

The completed archive includes files, descriptions, quality records and an understood relationship to the originals. Continue with Keeping files usable for twenty years.