Philosophy-RAG-demo/transcriptions/HoP 318 - Oliver Primavesi on Greek Manuscripts.txt

 Hi, I'm Peter Adamson, and you're listening to the History of Philosophy podcast, brought to you with the support of the Philosophy Department at King's College London and the LMU in Munich, online at historyoffilosophy.net. Today's episode will be an interview about Byzantine manuscripts with Oliver Primavesi, who is professor of Greek right here at the LMU in Munich. Together with me and Christoph Rapp, Oliver directs the Munich School of Ancient Philosophy, and I've wanted to have him on the podcast for a long time, so I'm glad to finally have him here. Hi, Oliver. Hi. Thanks for coming on the series. You're the perfect person to talk to about this topic because you work a lot with Greek manuscripts in the process of editing ancient philosophical works and you've worked on a variety of authors, including Empedocles, but you're most known for your work on Aristotle, or at least that's what you've been doing in the past years. Can you, though, before we get into talking about how you go about editing a work by Aristotle, can you just describe what a Byzantine manuscript looks like? So if I were allowed to pick one up and hold it in my hands, what would I be holding? Basically, it's a book with pages, and the writing material will be either parchment or paper. The envelope will be leather. And if it's paper, then an interesting feature of these pages will be the watermarks, which allow a precise date for the production of the paper, which is very important in establishing the sequence in which one manuscript was copied from the other. Another very important feature of the Byzantine book is that the pages, or rather the folia, i.e. the sheets, are combined into quires. So one double sheet or B folium will not be bound separately, but within a quire, a little booklet, as it were. And it's very important for the history of the text to take into account the loss of quires in such a book, and the re-entry of them perhaps at the wrong place, which explains many features of mistakes in our transmission. So pages don't just fall out one sheet at a time, they fall out in bunches of pages that were grouped together. Yes, that's it. And how fragile are they? I mean, do they crumble under the touch, or are they quite robust? No, especially parchment is a very robust material, and even if it's paper, the paper is stable. It's much better than paper from the late 19th and early 20th century. Okay, that's interesting. And there's, by the way, maybe we should also say something about the ink. So there's sometimes different colors of ink, right? Yes. Differences in the color of ink are very important for the distinction between various hands at work. Because what you will note in the Byzantine book is a scribe copies the text and then corrections, alterations are entered by second and third hands. And that's why microfilms are unsatisfactory. You need a color digital copy of the text if you are not working with the original. So this has actually become a lot easier, presumably just in the last 20 years. Yes. In a way, the technology of digitalization has created a paleographical paradise. Because you can, instead of actually having to go physically to every library to look at every... Which then may be closed just when you arrive. Right. Or the librarian is in a bad mood and doesn't want to show it to you. Yes. And the advance of the technology makes the production of such digital copies cheaper and cheaper every year. So you can, even if there are many manuscripts of the works you are going to edit, you can afford a full collection of digital copies as it were. And that's fantastic conditions of work, of course, compared with earlier ages. Right. By the way, I didn't want to conjure up this image of a grumpy librarian as a typical example. We love librarians, right? I just wanted to say that. Okay. So you were just talking about the scribes and, as you said, put it at the different hands. In other words, you can tell it's been written in the handwriting of different people. And the role of scribes in making manuscripts is obviously very important. They're the ones who copy the text from a previous copy, or maybe even compare several copies and produce a new copy in the manuscript you're looking at. And in the last episode, I tried to explain why it's important that scribes make mistakes when they copy, which is natural, right? If you try to copy out a whole book by hand, you'll make mistakes. Can you explain, since you can obviously explain it better than I can, how it is that the mistakes that scribes make help you edit the text and establish the best possible reading of the text? So the correct text can always go back to the author. The individuality of scribes who have copied this text rests on their mistakes. So if you want to describe relationships of dependence between various branches of the tradition, it's the mistakes which are the evidence for family relationships. The mistake once introduced then may stay in that branch and distinguish it from other branches where that mistake doesn't turn up. And the vital distinction here is between errors which create a bond between a group of manuscripts or conjunctive errors, and another type of error which entails that a certain group of manuscript which is free from the error in question cannot go back to a manuscript where this error was already there. And it's not the same errors which qualify as separative errors and as conjunctive errors. So there may be errors which are easily corrected, which are so obvious that they can be easily corrected by a later scribe. So these errors, errors which are easy to correct, do not count as separative errors because even a copy which doesn't feature them can still go back in principle to a copy which did feature it because it's so easy to correct. And similar things are true about conjunctive errors. Shared errors normally establish a suspicion that the manuscripts which share them go back to one of the same model where it turned up for the first time. But on the other hand, there are certain conditions in a text which invite certain errors to be committed more than once. Similar phrases, for instance, phrases of a similar structure tend to lose one or two of the similar items, and that can happen to the same result quite independently. Or maybe if Aristotle uses a really unusual word that looks kind of like a usual word, then two different scribes might write the usual word by mistake. For instance, yes, exactly. So one has to be clear about what kind of relationship one wants to establish and what kind of error is reliable evidence for that type of relationship. Would I be right in suspecting that because the same mistake can be made more than once and because scribes on their own can correct errors, so it's not like you just use one error to join two manuscripts in a family or to separate manuscripts in two different families. Is it more like you get lots of statistical evidence and then you see that there are quite a few shared errors between manuscript A and B and quite a few disjunctive errors between B and C? Or is it more like you can really do it with just a very small number of errors? If the errors are good errors, i.e. clearly separative and or clearly conjunctive, then a small number of decisively good errors is far better than a huge collection of individually weak errors, as it were. So it's really about quality and not quantity? Yes. I see. Okay, that's really interesting. And so what this allows you to do is basically figure out that manuscript B was copied from manuscript A or that A and B were both copied from the same model, whereas manuscript C wasn't. And then you build these branches and this is called a stemma, or stemma collodicum, so the branching tree of manuscripts. So once you've established that, you have an understanding of how the text was transmitted down through these many copies, but it seems like at the end you'd just be sitting there with lots of different versions of a text. So how do you go from the stage of having a stemma to the stage of actually telling your reader what you think Aristotle most likely wrote? There I would like to take up the notion of an archetype which we have already established in the previous issue, and that is a set of errors shared by all manuscripts of a given text. Defines the shape of the copy on which they all depend. So the archetype is the latest text which combined all errors shared by all excellent manuscripts. And what we are trying to get at by means of the stemma is then the first offspring of that archetype, i.e. the first division of the family tree, i.e. the immediate copies of that archetype, which we call hip archetypes. And so the aim of stemmatology in editing ancient texts is precisely to reduce the number of versions of texts available to the number of original copies made from the archetype. And given that especially the Greek texts have gone through the transliteration process, i.e. the change from the majuscule to the minuscule alphabet, a process which was very demanding and therefore not repeated more often than necessary, the number of hip archetypes which we can reconstruct will often be restricted to the number of transliterations made from minuscule copies of that text. And it's a fair guess that even in very popular texts like Aristotle, the most copied pagan texts in Byzantine manuscripts, there are up to three hip archetypes which we can reconstruct if the evidence is available. If it is, then the job of textual criticism is greatly facilitated because you have not to choose and to make up your mind between 50 or so variants if there are 50 or so manuscripts, but only between the two or three variants which can reasonably be ascribed to the first generation after the archetype. And it's this simplification which justifies the considerable labor implied in the establishment of a stemma. SHAYE SMITH Okay, but we still have to presumably be modest in our aspirations here in the sense that what you're reconstructing is, let's say, three different hip archetypes which would themselves have been copied in late antiquity, in other words, more than half a millennium after Aristotle lived. So we're talking about trying, in a way, we're trying to reconstruct copies of Aristotle's works which may already have had lots and lots of mistakes in them, in fact would have had lots and lots of mistakes in them. Yes, of course. What we aim at is to reconstruct the archetype on the basis of the reconstructed hip archetypes. What we must never have found is the archetype and even the first edition of Aristotle's extant works by Andronicus mid-first century BC, let alone Andronicus' edition with what Aristotle himself wrote. So that's another matter. That's another cup of tea. But the great progress is available even if the archetype of Aristotle's extant works were already be reconstructed. But we are far from that point. CURTIS SMITH Right. And then ultimately you're still going to be stuck in some passages with a situation where you may be tempted to just guess or hypothesize a different reading of the Greek which is not found in any manuscript. JENS SACHS Of course, there are these mistakes of the archetype. And there you can either resort to conjecture or to despair, i.e. to mark the corrupt section by two cruxes, the cruces desperationes, where you feel not entitled to guess anything with a reasonable degree of certainty. But still, the improvement of the text and the intelligibility of the text, if you dispose of the archetype, is very considerable. So for all its shortcomings, the great problem in Aristotelian scholarship is not the difference between the archetype and Andronicus nor the difference between Andronicus and Aristotle, although both are considerable or may be considerable. But our main problem is that the work of reconstructing the archetype has been done only for a small fraction of Aristotle's works. CURTIS SMITH Right. So one text that you have worked on very intensively is a short work by Aristotle called De Motua Animaleum, which means On the Motion of Animals. So it's often referred to under its Latin title, De Motua Animaleum. And I was wondering if you could give us an example from that text about how this kind of detective work actually improves our understanding of what Aristotle wrote. JUERGEN SANDERS Yes, the De Motua Animaleum is an interesting case, because the last editor of that work, Martha Nussbaum, had postulated the existence of a second hip archetype, independent of all manuscripts and their archetype, which she was working on. But for certain reasons, she postulated that some very good readings in otherwise unexciting manuscripts must go back to inspiration by a second hip archetype, which she deemed lost. And the funny thing with De Motua is that, in the meantime, we have identified among some quite late 15th century manuscript, the direct descendants of that second hip archetype, which Martha Nussbaum only postulated. And if we are to evaluate the impact of such a discovery, one is well advised in not stressing problems which nobody has noticed before, and where the new text brings a solution to what nobody had perceived as a problem so far, but to concentrate on well-established so-called cruxes, i.e. corrupt passages already discussed for a century or so, and to look whether your new hip archetype brings a solution for this kind of problem. So that will carry more conviction, as it were. And one example for that is the famous passage where Aristotle illustrates his basic theorem that animal self-motion is impossible without a stable external resting point, i.e. a point external to the animal which is stable. And he gives two examples for that. The first example is that if you walk in loose sand, you will suffer from the unstability of the sand. And the second example was unintelligible so far, because it would be, according to the text available to Martha Nussbaum 40 years ago, that mice are unable to walk on earth. And that strikes us as unreasonable, because mice are very good in walking on earth. Absolutely. They're fantastic at it. They're very quick. Yes. Now the new hip archetype has another version of that very passage, which is that it's difficult for mice to work on pitch. And that's an allusion to a famous proverb where a pitch trap is constructed by means of a piece of cheese which attracts then mice, but they have to walk through a zone of smooth pitch before. Like tar, sort of sticky stuff. Yes, where they are blocked, as it were. And that substance is unstable in two respects. First, it gives way a bit when the paw, the little paw of the mice sinks into it. But then if the mouse wants to go on, it uncannily follows the paw. So it's stable in both directions. And of course, it's fun to be able to allude to such a well-known proverb, well attested in the 30 years before and after Aristotle in that context. And so I'm sorry to interrupt, but the point here is that the two phrases, on earth and in pitch, turn out in Greek to be very similar. So you could... A paleographical similar. So entege and entpite in Greek metascule are close enough to each other to invite a mistake. And of course, if you don't know the proverb, you might not like the example. But given that it is a well-known proverb, one might now criticize my enthusiasm for that new reading by suggesting that there was some reason for which the text was corrupt in the first place. And then a scribe could have arrived and said, well, let's take the well-known proverb of the mice in pitch in order to solve that problem. This would be a case of something we talked about before where later scribes themselves make corrections to a corrupt text. Yes. But in this case, we were very lucky in that when we had discovered this variant in the new second hyperarchetype of the transmission of the demotoanimalium, we had a fresh look at the oldest manuscript of the first hyperarchetype, which was already known, but which had not been examined with sufficient care. Because what we saw is that the oldest manuscript of the demotoanimalium extant, the Paris manuscript 1853, there, the well-known but meaningless text of mice walking on earth, was written only by later hand after the first text had been erased. Now, examination years yielded the surprising result that the first script could be made readable again and yielded the reading mice walking in pitch without any doubt. So what the second hyperarchetype has brought to light was then found to have been once the reading, the original reading of the oldest manuscript of the first hyperarchetype II. That's amazing. It's like when they found Pluto by figuring out that it must be there and they looked for it and it was there. Yes. Something else we should mention at least briefly is what sometimes is called the indirect transmission of Greek text. So this is where you don't work with a manuscript of the text you're interested in editing, but you work with a translation of the text or maybe a quotation of the text in some other author. So maybe let's focus on what we can learn from translations of Greek works into other languages. How much do those help in establishing the Greek text that you're actually editing? Well, I think there are two cases, two quite different cases. The first is the medieval translations of Greek manuscript into Latin. And there, it turns out that the model of the Latin translations, for instance, made by William of Moorbaker around 1260 AD, rest on manuscripts which are among our oldest manuscripts, i.e. venerable 9th century parchment manuscripts, some of which are wholly or partly preserved and others are lost. So even when the text is well attested by Byzantine manuscripts, still a branch of it may be better represented by the Greek model of Latin medieval translation than by any excellent Greek manuscripts. But even greater help is to be expected from the Syriac and Arabic translations, especially those which were made in the Dark Ages or towards the end of the Dark Ages, between 600 and 800 AD, when the copying of pagan texts had almost ceased in the Greek, i.e. Byzantine, word for religious and economic reasons. So there it may happen that the manuscript acquired in a Greek library, being it in the Byzantine emperor or being in some other part of what used to be the Greek world by the Arabic translators, may be older than and independent of not only our extant manuscripts but also their archetypes. And there are famous examples, like for instance Aristotle Poetics chapter 1, where the loss of one out of two, in principle, extant Greek branches has made the Greek text virtually unintelligible, and where it's only by heavy conjecture that previous generations of scholars could make sense of it, and where the good conjectures, i.e. a subset of conjectures, has been brilliantly confirmed and others astonishingly improved by the use of the Arabic translation. So there's really no doubt that at least in that case the Arabic translation is of huge help in establishing what Aristotle actually wrote. Yes, there was this translation movement in Baghdad, which at the beginning of the 9th century turned its attention either to confine to medicine, science, and so forth, also to philosophy. And the manuscripts they use for that may be vastly superior to what may be preserved in the individual cases in Greek. And of course those manuscripts are otherwise lost. Yes. So this is the only access we have to them. Yes. Okay, let's finally turn to another example from Aristotle, because this is actually the texture most working on now, and this is a slightly longer text than they put to Adam-Alba. Nineteen times longer to be precise. Namely the metaphysics. Ninety times longer. Nineteen. Oh, nineteen times. Oh, that sounds so bad. Okay, so this is the metaphysics, and it's obviously one of Aristotle's most important works. What was the editorial situation with this text before you came to work on it, and why were you convinced that it was worth going through a year's worth of effort to produce a new edition? So the editorial situation was that the two standard editions, the edition plus commentary by William Ross and the OCT, the Oxford classical text by Jaeger, were prepared before the scholarly edition of the medieval legend translations came out, and before, or at least without making use of the first edition of the Arabic translation of the metaphysics by Ibn Rushd of Cordoba, normally called a veris in the Latin world, came out. So the two main sources of the indirect tradition did not go into our standard Greek text of the metaphysics. And if you add to that, that out of a number of fifty-something Greek manuscripts, only three were used and studied by these editors, William David Ross and Werner Jaeger, then you see that on practically all counts, which we have so far mentioned in this conversation, the evidence used for the standard text, and of course for all the translations depending on the standard text, were obviously deficient. So why did they do that? Is it because they had a stema that they thought suggested that everything other than the three manuscripts they used was irrelevant? They simply worked on the hypothesis that if you stick to the oldest manuscripts, then you are likely to get everything which is worthwhile. But I believe there's a rule in paleography that the older manuscripts are not necessarily better. Yes. Look at the case of the Demoto, where two very young manuscripts were, or one of them, was directly copied from what was obviously a direct product of transliteration, i.e. a 15th century manuscript can be the copy of a 9th century manuscript. And then of course, your general inclination to use old manuscripts and prefer them over recent ones collapses. It will be often the case that later manuscripts only add further errors, but there may be decisive exceptions from that rule. And have you been able to revise the picture of the transmission of the metaphysics? Yes, insofar as the two families already assumed by the editors mentioned, i.e. Ross and Jager, and by and large confirmed by our study of all Greek manuscripts. So there are these two branches, but the relationship between them has been very much or is likely to be clarified very much by the use of Arabic translation. Because what you get and what looks like two branches, two 9th century branches of the text, may in fact be the combination of two original branches and then additions, mistakes, and so forth added to either. And if you now have an independent witness, which may be related to one or the other of your two branches, he gives you a portrait of an earlier stage in the history of that branches. He adds a kind of third dimension to that picture of there being two branches. And that's likely to be the case with the metaphysics. I.e. we establish a difference between if you call the two branches alpha and beta, then you get your Byzantine version of alpha and beta, and by means of the Arab, you can then distinguish between original alpha and additional alpha. I see. So it basically gives you a kind of distinction within one of the two main branches. Yes. Right. Okay. So stepping back from these specific examples now, just in conclusion, I'm wondering what you would say to people who are interested in Aristotle, whether they're reading him in English translation or translations into other languages, or people who can work with the Greek, or if they're reading a secondary literature that's of course based on the additions that have been available for the last hundred and more years. Would you say that we kind of need to put an asterisk next to all of that? And at the bottom of the page, the asterisk should say, by the way, bear in mind that this was all based on inadequate additions and that all of the text may need to change once we actually go in and do the work properly? Or is it not that bad? The possibility exists of that radical deficiency, but it will turn out that the situation is very different in different works. Nobody can tell in advance whether the full study of the whole tradition, both direct and indirect, will have a huge impact with 20 changes per Becker page, being the standard edition of Aristotle from 1831, or no change per Becker. Everything is possible depending on the open question to what extent the manuscripts already used represent the whole of the accessible tradition or not. You cannot tell that in advance. But if in a situation like the metaphysics, you know, and you have known that there is an Arabic tradition, extant even in Arabic and not as in other cases, only in the Latin translation made from the Arabic. And if you know that this Arabic translation has not been used ever before for the establishment of the Greek text, then you can be pretty confident that using this extra evidence will change your text considerably. Okay. And maybe it's also worth noting that the situation with Aristotle is different from, for example, Plato. So for one thing, there are no full Arabic translations of Plato anyway. And for another thing, I know from talking to you before that the editorial situation with Aristotle, perhaps surprisingly, is not nearly as good as with Plato. Is that right? Yes, it's a problem of Aristotle's popularity in Byzantine times, that there are roughly 1000 manuscripts containing works of Aristotle, and only something like 200 containing works of Plato, which means that to establish the whole family tree is far more complicated in the case of Aristotle than it is in the case of Plato. So ironically, the situation with Plato is better because there's less evidence. Yes. And the evidence of the single manuscript can be called better in so far as those branches which happen to be extant in the Plato tradition tend to be represented by old manuscripts. Whereas in Aristotle, that's only true for part of the branches. So you can have branches represented by 9th century manuscripts, but there may be other branches. And one case in point was the De Motto animalium, where the second branch of equal importance is only attested in full Greek text by 15th century manuscripts. And so, whereas in the case of Plato, the choice of manuscripts was basically established by the end of the 19th century, so that the OCT, the Oxford classic text by Bernard, can be called with some exceptions reasonably reliable, that's far from the case for the average edition of Aristotle. Right. Nevermind other philosophers who wrote in Greek, so Plotinus, Proclus, I mean, there's editorial work potentially to be done on a very wide range of authors, right? Yes. Although the case of Plotinus there, we have already an edition which takes into account the far more restricted number of manuscripts, sufficiently I would say. Okay, well that's good news. And speaking of Neoplatonism, next time I'm going to be looking not at the manuscript tradition so much as the reception tradition of a Neoplatonist philosopher, namely Proclus. We're going to be looking at two Byzantine authors, or at least Eastern Christian authors, who reacted to Proclus one critically, one more positively. So that's what I'll be talking about next time. For now, I'll thank Oliver Primavesi very much for coming on the podcast. You're welcome. And please join me next time to look at the reception of Proclus in Byzantium here on the History of Philosophy Without Any Gaps.