C2From Microfilm to the Scanned Page

Large-scale copying of printed material for preservation and access is older than computing. Microfilm established both the practice and most of its arguments, and reading that history makes the current one much less surprising.

What microfilm solved

Microfilm answered a genuine crisis. Newspapers and much late-nineteenth-century printing were produced on paper that becomes acidic and brittle, and a run of a daily paper consumes enormous shelf space in a form that cannot be handled without damage. Photographing each page onto film reduced the space requirement dramatically, produced a copy stable for a long period under proper storage, and allowed a fragile original to be withdrawn from use.

It also introduced the pattern that has governed every reformatting project since: the copy is easier to store and harder to use. A reel must be mounted, wound and read on a machine in a fixed location, under conditions the reader does not control. Browsing is slow, and comparison between two points in a run is awkward. The material became safe and became less consulted.

The disposal argument

The controversial part was never the filming. It was that filming was frequently used to justify discarding the original. The objections raised then are exactly the ones raised now: the copy loses colour, loses paper evidence, loses marginal material at the gutter, sometimes loses whole pages through operator error, and freezes the copy at whatever quality the equipment of that decade allowed. A photograph is a decision about what matters, taken once, on behalf of every future reader.

The counter-argument is equally serious and is about cost. Preservation of originals at scale is expensive, and an institution that refuses every reformatting decision until it can be made perfectly will lose material to decay in the meantime. Neither position is simply correct, which is why the argument recurs rather than resolving.

Scanning, and the layer of text

Digital scanning inherits all of this and adds one new thing: optical character recognition. A scan is an image, and an image is not searchable. Recognition produces a text layer beneath the image, which is what makes a scanned collection usable at all.

That text layer is an interpretation, and its quality varies enormously with typeface, paper condition, layout complexity and language. Clean modern type recognises well. Broken type, close-set columns, tables, marginal notes, older letterforms and mixed scripts recognise badly, and the failures are silent: a search simply returns nothing, and the reader concludes the material does not exist rather than that it was not read correctly.

What this means for a reader

The practical consequences are worth stating plainly. A negative result in a scanned archive is weak evidence of absence. A quoted passage taken from a text layer should be checked against the page image, because recognition errors produce plausible-looking wrong words. And the visible extent of a digitised collection is a statement about what was funded, permitted and physically robust enough to be handled, not a statement about what was published.

Microfilm made a generation of readers accustomed to consulting a surrogate. Scanning made the surrogate searchable and portable, which is a genuine transformation. It did not make the surrogate complete, and the habit of treating it as complete is the most common error in using these collections.