Who Copied the Books Before Printing? A Room, a Rule and a Lot of Time

Copying a book by hand meant preparing the writing surface, ruling lines, cutting and maintaining quills, mixing ink, and then writing at a rate that made a substantial volume the work of many months. Parchment was made from animal skin, scraped, stretched and treated, and a large book consumed the skins of a herd, which is why books were enormously expensive before paper arrived. Scribes worked from an exemplar, copying line by line, and errors entered at every generation, with a scribe's eye skipping between similar words, misreading unfamiliar terms and occasionally correcting what was not wrong. Illumination and binding were separate specialities. The whole operation was slow enough that a monastery's library grew by a few volumes a year.

Practice varied and certain arrangements recur across surviving descriptions:

• •Silence, with monastic sign language used to request materials
• •Work by daylight only, since candles near parchment and ink were prohibited as a fire risk
• •An armarius responsible for materials, assignments and the library
• •Division of labour between scribes, rubricators adding headings in red, and illuminators
• •Correction by a reader comparing the copy against the exemplar afterwards
• •Colophons at the end where scribes recorded their names and occasionally complained

The survival of ancient literature was decided by these choices and the filter was severe. Texts needed by the institution were copied reliably, meaning scripture, liturgy, commentary and the works of church authorities. Classical texts survived where they were useful for teaching grammar and rhetoric, which is why certain authors are abundant and others vanished. A text that fell out of use for a few centuries stopped being copied, and its remaining copies decayed, so a great deal of ancient literature is known only from titles and quotations. Parchment was also reused, with old writing scraped off and overwritten, and modern imaging has recovered several important lost texts from beneath later ones, including a mathematical work by Archimedes read in 1998.

The personal remarks scribes added at the ends of manuscripts are among the more human survivals from the period. Complaints about cold hands, poor light, bad parchment and the length of the work appear repeatedly across centuries and regions. Requests that the reader pray for the copyist are standard. Curses against anyone who steals or damages the book are common and occasionally elaborate. Marginal drawings, doodles and notes appear where attention wandered. One well-known Irish marginal poem compares the scribe's work hunting words to his cat hunting mice. These are not decoration but the traces of individuals doing a long tedious job, and they are frequently the only evidence that a particular person existed at all.

Because every surviving copy descends from earlier copies with errors accumulating, scholars reconstruct what an author wrote by comparing manuscripts. Shared errors indicate shared ancestry, which allows the surviving copies to be arranged into a family tree showing which descended from which, and a reading present in independent branches is more likely original than one confined to a single branch. That method, developed systematically in the nineteenth century, turns the accumulated corruption into evidence. It has limits, since manuscripts were sometimes corrected against others, which mixes the branches, and since the tree frequently converges on a single surviving ancestor beyond which nothing can be recovered. Editions print the reconstructed text with the variants recorded beneath, which is what the dense footnotes in a scholarly edition are.

Copying by hand was slow enough that a library grew by a few volumes a year, and parchment from whole herds made books enormously expensive. Texts survived if an institution kept using them, which is why certain classical authors are abundant and others are known only from quotations. Comparing shared errors between surviving copies reconstructs what the author wrote.