I’ve now collated my Latin text – all 6 sentences of it – with 2 early editions and 24 manuscripts. I have at least another 6-10 manuscripts accessible to me to collate.
As I thought, this is a case where you have to learn by doing. You have to attempt to collate the text and manuscripts, somehow, anyhow, just guessing how, in order to learn how to do it, and how not to do it. Then you go back and do it properly.
I’d like to record a couple of things which are emerging from my first pass.
Point 1. Don’t alter your “base text” in mid-collate. If you do, the early parts of your collation become uncertain.
The “base text” does not matter at all. It is NOT your final text. Rather it is some “textus receptus” that you found in a crummy early edition. It’s a random text. But…. it’s the text against which you collate. Unless you want to record every word of every witness, you have to have something where your silence says “the text at this point in the manuscript is the same as the ‘base text'”.
I have indeed made this mistake already. What I should have done is have another line, which is my proposed text. That I could alter as I choose.
Point 2. Expect to collate more than once.
When you start, you don’t know how the text varies. You don’t know if it varies only in individual words, or if whole chunks are involved. So take this part of the text:
favet**, credite mihi, favet nostrae devotioni.
Now you have to compile your collation using a) the word or phrase, followed by b) the manuscript shelfmark. Doing it the other way doesn’t work for more than about three manuscripts. So you gaily put something like this:
favet – Angers 802, Arras 462, Milan P113supp
and you mark which “favet” in the base text.
But as you proceed, you find that you’ve goofed. You should have recorded both “favet”s separately. Your notes start to change, until you reach something like this:
REDO THE WHOLE THING “favet, credite mihi, favet” (he favours) – Mom.; erased BNF 2627, Fal., Angers 802, Arras 462, Milan P113supp, BNF 1864; BNF 5287, BNF 5296C, BNF 5344 RECHECK, BNF 5573, BNF 11750, BNF 12600, Saint-Omer 715
“adjuvat credite michi et favet” – BNF 5284, BNF 5345
“adjuvat…adjuvat” – BNF 5360, BNF 5607, BNF 18303, BNF NAL 2335, Rouen 1383
“adjuvat” (=he helps) – Wien ONB 12831 (15th), Orleans 342, BNF 989 RECHECK
The unit of variation isn’t the individual “favet” but the pair favet…favet. There are four possible combinations of “favet” and “adjuvat“. You’ve got to record this part of the text differently. You would have done it differently from the start, had you known. You’re going to have to go back to the first few manuscripts and see how they look against what you now know about the variability of the text.
So… your first pass through the collation process is really just a means to learn what the “units of variation” are.
Point 3. Note down any cases that explain a divergence of reading.
I’ve been wondering why I have variations between “expectet” and “expectat“. But I have now come across the practice of writing “expectet” as “expect&“. That abbreviation at the end, in some versions of book hand, looks almost identical to a lower-case “a” with a “t” right against it. So that by itself explains the “at” variant.
Likewise, as we saw in a previous post, in Beneventan minuscule I was really uncertain what the word was. It looked to me like “notata“. But someone who knew the hand told me it was “nacta“. The collation shows that these are the two main variants of the text at that point. Conclusion? An early ancestor of every manuscript of the “notata” family was written in a Beneventan book hand, and the scribe misread it – just as I did – when he made a copy in Carolingian book hand. The text was composed in Naples, so the existence of such a copy in Beneventan is almost inevitable.
I’ve also noted that some manuscripts use the Tironian symbol for “et” (=”and”). Usefully there is a unicode character for it: “⁊“. When the “⁊” is faint and narrow – and it often looks just like the lower portion of an “i” -, then the scribe may simply not see it. It can quite easily be mistaken for text on the other side of the leaf. I found myself looking twice, in one manuscript. So the presence or absence of an “et” is not necessarily significant.
Point 4. The early editions are bad stuff. The editors have introduced changes which are not found in any manuscript, and seem entirely unnecessary.
I have not, of course, examined every manuscript. But I’ve looked at rather a lot now. There is a consistent pattern of differences, between the manuscripts on the one hand, and both of the early editions (in different places) on the other. I know that people tend to assume that the editors just printed what they had before them. I am less sure of this now.
I’m going to carry on with the collate, and process the remaining half-dozen manuscripts into the working document. But once I reach the end, I will do it all again; except that this time, I know what to look for.