Bound material is a different job from loose paper, and the reason is mechanical rather than technical. A stack of A4 can be squared up and fed through a transport. A book cannot. Every page is anchored along one edge, the text nearest that edge curves away from the lens, and the spine will only open so far before the binding starts to fail. Book scanning is the set of methods for getting a flat, complete, legible image out of a volume you are not permitted to take apart.
Destructive and non-destructive capture
The fast way to scan a book is to cut the spine off and feed the loose pages through a production scanner. That is a legitimate choice for a disposable reference copy of something reprintable. It is the wrong answer for a minute book, a statutory register, a signed volume, or anything with evidential or heritage value, because the artefact is part of the record. Non-destructive capture keeps the volume intact and accepts a much slower page rate in exchange.
The problems a binding creates
- Gutter shadow. The page curves down into the spine, the light falls away, and the inner margin goes dark and distorted just where the text still is.
- Text running into the binding, which is normal in registers and ledgers where entries were written straight across the centre of an opening.
- Tight bindings that will not open flat and have to be supported at an angle rather than pressed down by hand.
- Loose plates, foldouts and inserts that belong to the volume but do not sit in the page sequence.
- Printed page numbers that do not match the physical leaf count, once unnumbered plates, blanks and mis-numbered folios are counted.
- Paper thin enough that the reverse of the sheet shows through the capture and has to be handled in the imaging rather than accepted.
How the capture runs
- 1
Collation
The volume is checked through before anything is captured: total leaves, foldouts, inserts, loose material tucked between pages, and existing damage. This record is what tells you at the end whether anything was missed, and it is worth the time it costs.
- 2
Support and controlled opening
The volume is supported so the binding opens no further than it safely can, with the page held flat against glass rather than pressed down by hand. A volume that will only open partway is captured at that angle instead of being forced.
- 3
Opening capture and page splitting
Each opening is captured, then split into two single pages, deskewed and cropped to the page edge so the delivered file reads as a sequence of pages rather than a run of spreads.
- 4
Gutter and curvature correction
Distortion in the inner margin is corrected so the text nearest the spine reads straight rather than curving away and dropping into shadow. This is the step that decides whether ledger entries near the centre are usable.
- 5
Sequence verification
The image sequence is reconciled against the collation record. Foldouts are captured separately at full extent and placed at the correct point in the sequence rather than left at the end of the file.
- 6
OCR and output
A searchable PDF is produced carrying the page image with a text layer behind it, and page numbering that matches the volume rather than the image count, so a citation to a printed page number still lands on the right page.
What to expect from searchable text
OCR performs well on printed body text in a clean, well-opened volume. It performs poorly on handwritten ledger entries, faded ink, and ruled columns where the rules confuse line detection and entries get merged across cells. For handwritten registers, plan on a page image plus typed index fields covering the values people actually search: date, name, reference number, entry type. That is a description exercise, and it should appear in the scope rather than being expected as a by-product of the scan.
Where this comes up
Board minute books and statutory registers that have to be produced on request but must survive intact. Property and land record volumes held by government entities. Financial ledgers predating the current accounting system. Bound medical case books still referenced for historical patient history. Institutional library holdings being made available without putting the physical copy back into circulation. In each case the volume itself is worth more than the speed lost by keeping it whole.
Frequently asked questions
Do you have to cut the spine off to scan a book?
No. Non-destructive capture keeps the binding intact: the volume is supported, opened only as far as it safely allows, and each opening is captured against glass rather than fed through a transport. Cutting is faster and remains an option for disposable reference copies, but it is never the default for registers, minute books or anything with evidential value.
How is text near the spine captured?
By supporting the volume so the inner margin lies as flat as the binding permits, then correcting the residual curvature and shadow in processing so the text reads straight. Entries written across the centre of an opening, which is normal in ledgers and registers, are the reason this step exists and the first thing to inspect on a sample.
Can foldouts, maps and plates inside a volume be scanned?
Yes. They are captured separately at their full extent rather than folded, then placed at the correct point in the page sequence so the digital volume reads in the same order as the physical one. Foldouts are recorded during the collation check at the start, which is how they avoid being missed or dropped at the end of the file.
What format are scanned volumes delivered in?
Normally a searchable PDF per volume, carrying the page image with a text layer behind it and page numbering matched to the printed volume rather than the image count. Where the volume is historically significant, an unprocessed preservation master is kept alongside the access copy, in line with the approach used for archive collections.
Are handwritten registers searchable after scanning?
Not through OCR alone. Handwriting, faded ink and ruled columns all defeat reliable recognition, so the searchable route is typed index fields: date, name, reference number and entry type, captured by a person from the page image. That is planned work with its own cost, and it should be in the scope from the start rather than discovered at handover.
Is book scanning priced per page or per volume?
Per page, after a sample assessment, because volumes vary far more than boxes do. The factors that move the rate are page count including blanks, binding tightness, condition, the number of foldouts and inserts, whether index fields are captured by hand, and whether an unprocessed preservation master is required alongside the access copy.
Related reading
- Archive digitizationBound volumes usually arrive as part of a wider historical collection.
- Document scanning and digitizationThe broader service, for loose files and operational records.
- Off-site document digitizationBound-volume capture needs bench equipment that does not travel.
- Ask for a sample assessmentVolumes vary too much to price without seeing a representative few.
- Athena Global TechnologiesOur parent company, whose clients include archives and libraries.