It is hard to avoid assuming that early printers had the same concern for accuracy as a modern book printer does and that this meant that the switchover from script to print suddenly ushered in a new era of accuracy. We often casually assert that once Gutenberg had invented movable metal types it was possible to record texts with an accuracy as was never achieved by scribes. This is just plain wrong though: on two sides. Firstly scribes did make mistakes, omitting a word here, spelling one wrong there, even leaving out a line of copy from time to time, but book dealers were alive to the problem, and spent quite a lot of effort trying to ensure that copies were all the same. So it’s true that inaccuracies abounded, but people were constantly trying to improve texts or at least not to make them any more degraded. Secondly we overlook the fact that early printers themselves made lots of mistakes. Many did so despite their best efforts, while others were happy to produce slapdash work as long as it made money.
It is almost inevitable that the first printers would struggle to achieve accuracy. If a sheet of 16 pages were being printed the first task would be to cast type. Once you had the bits of type you needed you could start to compose it: i.e.to combine it into words, lines and pages. (Setting the type for Gutenberg’s 42-line Bible is estimated only to have started two or three years after the cutting of the types and their casting had begun! Cutting moulds probably started in 1449 or 1450, composing in 1452, and the printing probably wasn’t completed till 1455.) Most 15th century printers wouldn’t have had anything describable as a proofing press, so in order to pull a proof they’d put the forme of type onto the press and pull a couple of copies. Leaving the press set up but not in operation was hopelessly uneconomic, so such proofreading as was done would be done quickly. Any corrections called for would be made to the forme while it was on press — it’d be unlocked and a new bit of type inserted quickly (and of course occasionally wrongly).
Then the run would continue. Because paper was rather expensive, the printer would not discard the early sheets run off before corrections, so that the pile of copies of the first sheet would contain some copies from the first (uncorrected) state, as well as copies from the second state and even from a third or fourth state. When the desired quantity had been run the type would be distributed and used for setting the next eight pages, which would go through the same sort of correction cycle(s). When the job was finished and it came to binding you might end up with one section from the first state, another from the second, and two from the third — they’d all be mixed up, so that some pages would be corrected while others were from the first uncorrected run with different mixtures on front and back. So, contrary to what we would like to believe, early printed books would not all be identical — they’d be bewilderingly different. The Folger Shakespeare Library owns eighty copies of Shakespeare’s First Folio — no two copies are identical!
It took centuries for printers to work out the routines which we now take for granted which enabled them to (almost) guarantee accurate reproduction of texts. This didn’t really start to be true till the 17th and 18th centuries. When we look back we cannot imagine a situation where proofing routines weren’t firmly established — it’s our imagination that’s at fault.
Now although this error-proneness was a quality shared with manuscripts, the printed version did have one huge advantage. There were multiple copies, so, even if there might be mistakes these were less significant than the fact that many people actually got to see the books, whereas in the past getting to see a manuscript was not an everyday experience. For example, as Elizabeth Eisenstein: tells us in The Printing Press as an Agent of Change (pp. 580-1), Nicolaus Copernicus, who was born twenty years after Gutenberg had started composing his Bible, “as a student at Cracow in the 1480s . . . probably found it hard to get a look at a single copy of Ptolemy’s Almagest — even in a corrupted medieval Latin form. Before he died, he had three different editions at hand.”
Making Book 
Fascinating stuff! Especially in comparison with where we are now with ebooks, POD, et al.
I think we’ve almost gotten to the point of fetishizing accuracy. I’m not advocating spelling reform, but I do think that communication is the aim, and if someone misspells “ottomobile” the reader isn’t really harmed. Down with spellcheckers!?
Even the spelling of the same word could be different throughout the folio. In comparison it makes what you read on-line look an like an academic thesis.
Errors yes, but I think we need to bear in mind that simple misspellings don’t really get in the way of comprehension, and I suspect that lots of these early typos were of the sort that anyone would immediately recognize as being wrong, and effortlessly correct as they read. It’s a bit like the Erratum post I did recently where the author in effect says casually, “well we don’t think we really need to point these out, but here are the mistakes we made”.