The only errors are transmission errors. Everything else is evidence about the way the language worked at the time -- and even transmission errors can serve that function under some circumstances. Sometimes people will still refer to anomalous spellings, etc., as "errors", as a kind of convenient shorthand.
A simple example is the existence of alternate spellings, which can indicate that there was a sound shift going on. In Greek documents from the period of Roman rule onwards, we often find variation in how the Latin letter v is transliterated: Valerius became Οὐαλέριος, but Vergilius became Βεργίλλιος (or Βεργίλιος). This provides evidence about the pronunciation of both languages: (1) that Latin v represented two different sounds at the time, /w/ and /v/; and (2) that Greek β was being pronounced /v/ at the time (at least sometimes), as in modern Greek.
Transmission errors are the only cases that really ought to be called mistakes. And even then it's complicated, because of course a transmission error may not be just a slip, but may represent an effort to "correct" an older text: in that case, the error provides evidence of the state of the language at the time when that copy was made.
One really complicated case is the spelling that we see in Homeric epic. Linguists love to pull apart Homeric language because it's the most archaic form of literary Greek that we have -- in some ways, the closest extant relative of Mycenaean Greek, and also good evidence of variation in several different dialects of early Greek; that in turn makes it very important for the reconstruction of proto-Helladic forms and Indo-European. However, the history of the transmission of the text of Homer throws a real spanner in the works, and this is something I feel linguists don't adequately appreciate. To my eyes it looks very much as though the orthography of Homeric Greek only became really settled when 3rd century BCE Alexandrian critics were making standard editions of the text; prior to that, things were more fluid. In particular, the transmission of the text in Athens had a big impact on the Alexandrian text; but Athens didn't adopt the Ionian alphabet until 404 BCE. So what did the Athenian text look like before that date? To be frank we don't know. Most philologists just seem to assume that the Athenians used their own form of the alphabet for every other purpose, but they used the Ionian alphabet for Homer. You can probably tell that I don't think that's a safe assumption. And if you abandon that assumption, and imagine that it was only in the 3rd century that the standard orthography was invented, then some fairly important things that we "know" about earlier linguistic change go out the window: a fair chunk of evidence for a separate Aeolic phase of the Greek epic tradition disappears, and some points about the development of the Ionic dialect also become much more nebulous. (I can go into details if you like, but only if you don't mind having terms like "quantitative metathesis" and "genitives of a-stem nouns" thrown at you.)
That might be a bit more intricate than you were imagining, but it should at least serve to illustrate that a lot can hinge on spelling variations -- in terms of linguistic change, at least.
fyi, you'll find more comments in these earlier posts
I kinda love this question, and it's something I've been thinking about a lot recently. I've done most of my research thus far on England in the 700s, but I'm now starting to look more at France and Germany in the 800s. The Latin completely changes, and so does the handwriting.
What do we make of all this? Which group had it right: the monks of Anglo-Saxon England, or the scribes of Carolingian Francia? Or were both groups simply flubbing (or flouting) the norms of classical Latin? Were they both wrong?
I think it's better to think of them both as being right -- handwriting, grammar, and spelling. As historians, it tells us much more to consider what they were trying to do than to try to take account of everything they may (or may not) have gotten wrong. Let's take as an example two texts I've recently looked at.
First, the Ismere Diploma. One of the English kings had this charter written in 736 to give a plot of land to one of his lords. There's some quirky things about this. The handwriting is the same as what English monks used to copy the bible, which tells us that writing was still a rare art, and that the word of the king was in some ways as sacred as the word of God. The grammar was often structured to place verbs second last in a phrase ('it to the aforesaid pertains field'), which was considered an artful flourish and assumes the reader has mastered how to parse Latin phrases. And the spelling shows a unique hybrid of Old English conventions and classical Latin norms, so the name that we now spell Æthelbald could be spelled in Old English "Æþelbald" or in Latin "Aethilbaltus," but it appears here as "Aethilbalt," showing that the scribe wasn't quite comfortable writing in the Anglo-Saxon tongue.
In comparison, let's take a Carolingian manuscript. The text is Isidore's Etymologies, written around 633, but this particular manuscript was copied in France sometime in the 800s. The handwriting is rounder, easier to write, perhaps even rushed. And there's more regular spacing between words, making it easier to read. It's a functional hand, showing that the scribes were much more business-like and much less reverent with what they were working with. The grammar is copied directly from a text two hundred years old without emendation or gloss, showing a cultural confidence with Isidore's old-fashioned style. And the spelling relies on aggressive abbreviations: wavy lines indicate that a syllable has been omitted, p's with a bar can indicate any prefix beginning with p, and "Christ" is represented by the Greek "xpr" (the r is actually a Latin s). These abbreviations suggest a large community of writers and readers who are all sufficiently trained to recognize where letters are missing and to fill them in with the proper Latin syllables.
So is it useful to think of either scribe -- the English monk who wrote the Ismere diploma or the Frankish scribe who copied the Etymologies -- as having poor handwriting or making grammar mistakes or typos? I don't think so. But by placing these texts side by side, we see how an English monk was meticulous and reverent, whereas a Frankish monk was rushed and brash. And in their own small ways, these observations help us get closer to understanding the worlds these people lived in.
