From my diary

In my last post I summarised as best I could the textual tradition of Plato, much of it from Pasquali’s Storia, written in 1934 (!).  After that, I returned to Pasquali, to see whether any other useful summaries of textual traditions lay hidden therein.  Sadly “hidden” is the right word.  The book is divided into sections, and the sections into chapters, and neither indicates the gold buried within.  The material about Plato was in one of those chapters.

The chapter itself was divided into 13 sub-sections, each headed only with a numeral.  One of these gave the tradition of Plato, but if I had not known that it was there, I should not have known.  Each sub-section was itself divided into sub-sub-sections, separated only by three asterisks.  In fact the sub-sub-sections formed a pattern; stuff in the manuscripts, stuff in commentaries, etc.  But no titles or marginalia indicated this.  This is very hard on a non-Italian speaker.

Worse yet, while each sub-section is indeed concerned with an author, such as Plato, or Homer, the reader must read almost a paragraph of piffle before any name is mentioned.  Three of these mention some other author first, just to confuse.

It is really important, when writing something intended for more than one language group, to provide an apparatus of headings and numberings allowing the reader to find what he is looking for!

Since then I have been reading into the textual transmission of Homer.  The sheer quantity of talk is extraordinary, and most of it inward-looking.

I was amused by some words of T. W. Allen, Homer: The Origins and the Transmission (1924), online here, whose edition of the text is still fundamental.  Page 6:

The removal of ‘literature’ and the neglect of the principles of investigation imposed upon us these hundred years has led with me to remarkable results, namely the discredit of contemporary method and the rehabilitation of tradition. With whatever refraction and inaccuracy classical and mediaeval mentality present to us the ancient world, the image produced by modern philological method is more distorted, and is in fact in most cases completely false. The reasoning applied to ancient records in the last hundred years is not only baseless, but it has cumbered the old world with lumber which makes the study of it a difficult and certainly most tedious matter. The repulsive jargon in which ancient history and literary criticism are conveyed, the narrow outlook, low vision and ignorance of human nature and the human mind—its working and possibilities— have turned classical philology into ridicule.

A century later, the same might be said again, I suspect.  Indeed another scholar writing a couple of decades ago suggested that there is no such thing as “Homer”, only a mass of oral material more or less edited ad hoc by persons unknown in antiquity.  From the lofty heights of his ivory tower, so far up in the clouds that the ground is no longer visible, he actually mocked a fellow scholar for expressing concern about whether we lose our Homer altogether in all this learned hairsplitting.  It did not seem to occur to this gentleman, in his privileged little world, that somebody has to pay for his ivory tower, and that this someone might rather object to paying for the privileged to study something that doesn’t actually exist.  The job of the text critic is to heal the text, not destroy it.

There is a persistent lack of reference to the primary facts about the tradition in much of what I have read so far.  This is not a good sign, in my experience.  Whatever we say about antiquity must be grounded in that data.


How did the works of Plato reach us? – The textual tradition of the dialogues

Plato’s works have reached us in medieval handwritten copies, the earliest written around 900 AD. The dialogues are arranged into nine groups of four dialogues, or “tetralogies.”[1] These give us the works in complete form, from direct copying down the centuries. But there are also surviving fragments of ancient copies on papyrus, found in rubbish dumps in Egypt where the climate is dry, which sometimes give a better reading in this passage or that, where the text has become corrupt in the centuries. Plato also is quoted at great length by other ancient authors, and sometimes these also have readings to contribute. Finally there are ancient translations of Plato into other languages.

The witnesses to the direct tradition, the medieval manuscript copies, are very numerous; more so than for any of the Greek classics other than Homer. One article suggests at least 250 manuscripts survive[2]; and a search of the Pinakes database gave 439.[3] Most are merely copies of other manuscripts, so it is important to identify the primary manuscripts.

The 19th century study of the transmission of the text proved to be unsound, and the whole task had to be started again just before WW1. In 1959 Dodds could write that critical work on the text is still in its early stages, and that, for the first 7 tetralogies, nobody could say how many of the manuscripts were primary – based on no other manuscript – or how they related to each other, or to the secondary manuscripts. And why? Because scholars lacked accurate collations of the manuscripts. Indeed the collations that were available proved to be full of errors.[4]

Key Medieval Manuscripts [5]

For the text of individual dialogues additional manuscripts are important, but these are the main ones for the tradition as a whole.

B – Oxford, Bodleian, E. D. Clarke 39 (= “Clarkianus”). The oldest extant witness. Written in 895 AD by “John the Calligrapher” for Arethas of Caesarea, according to a subscriptio. It contains the first 6 tetralogies, and never contained more. It was probably the first volume of a two-volume Plato. It was discovered in 1801, lying on the floor of the monastery of St John the Apostle on Patmos, and Clark purchased it. By looking at medieval catalogues of the monastery library, it seems that the monastery acquired it sometime between 1201 and 1355, and it remained largely unknown thereafter. It’s not clear that any other manuscript derives from it. B is online here:

The top of the first page of B – Bodleian MS E. D. Clarke 39, folio 1r.

A – Paris, BNF graecus 1807. Ca. 900 AD. Today contains only the 8th and 9th tetralogies, and the Spuria. Probably the second volume of a two-volume set. Not online. Online here.

T – Venice, Marcianus Append. Class. 4. 1. Copy of A. Written by Ephraim Monachus ca. 950.[6] It contains the first 7 tetralogies and part of the 8th, although this may be copied from elsewhere. At the end of the 7th tetralogy is a note indicated the “end of volume 1”; again it must be descended from a two volume medieval Plato. Probably copied from A when it was complete. B and T have some links, possibly because an ancestor of one was corrected from the other. T is online here:

W – Vienna suppl. phil. gr. 7. 12th century? Contains tetralogies 1-3, and then the dialogues of 4 to 7 in a jumbled order. It is independent of B and T. It was probably acquired in Greece or Sicily in the 14th century by Nerio Acciaiauoli, passed in 1478 to the Certose near Florence, and in 1725 to Vienna. W is online here:

D – Venetus 185 (Coll. 576). 12th century. Once belonged to Bessarion. Seems to be independent of A. For the first 4 tetralogies is closely related to B, but not derived from it.[7]

B, A/T, D, W form a family of closely related manuscripts. Manuscript F is from a very different family.

F – Vienna suppl. phil. gr. 39. 13th century. It contains the dialogues from tetralogy VI.3 (Gorgias) to IX.1 (Minos). From a different family to B, A/T and W. Its readings often agree with the quotations in Stobaeus and Eusebius, whether the reading is authentic or corrupt. Some of its errors are explicable if the scribe copied directly from a manuscript written in an uncial hand, i.e. an ancient manuscript, with no word division and limited punctuation. This is confirmed by the papyri which demonstrate that the F text-type goes back at least to the second century AD. This is unique among the mss of Plato. Dodds estimates from the probably dimensions of the exemplar that it may have been a “cheap papyrus code which was manufactured in quantity in and after the third century A.D.” and represents “the ‘commercial’ texts which circulated among the reading public rather than the more scholarly editions,” complete with vulgarisations.

The tradition of the ninth and final tetralogy is somewhat different from the others, and manuscripts of it are less common. All the manuscripts, including the 11th century Armenian translation of its first two dialogues (Minos and the Laws), derive from a manuscript equipped with variants, reproduced rather faithfully. This may be an ancient manuscript, or more likely a Byzantine transliteration of the 9th century.

The Papyri

No ancient copies of any work of Plato have reached us. But small fragments of such copies do survive: little scraps of papyrus found in the ancient rubbish dumps of deserted cities in Egypt. The database lists 95 papyrus fragments, although this is a mere handful compared to the number of papyri of Homer. The oldest four fragments date from the first part of the 3rd century BC: a scrap of the Phaedo, Laches, Sophist, and an epistle. But the vast majority date to the 2nd and 3rd centuries AD, including a long section of the Symposium (P.Oxy. 843, 2nd c.), and the numerous and long fragments of the Phaedrus.

Pasquali wrote, “There was much discussion about the value of those papyri [the 3rd c. BC Phaedo and Laches] immediately after their discovery: now the general opinion is clear. They provide an apparently careless text: there are frequent spelling errors and negligent mistakes, such as arbitrary and impossible shifts of words, none of which is surprising in private copies; nor do they lack small lacunae. All this matters very little if a solid foundation can be glimpsed through the damaged surface. And for the most part they are like this: the Laches papyrus contains only 189d -192a, yet it greatly improves our text.”

The Indirect Tradition

The text of Plato is quoted in a number of ancient authors. These quotations are extensive; between a quarter and a half of some dialogues are quoted. The most important source is Stobaeus Anthology, and then Eusebius Praeparatio Evangelica. Other authors quoting more than a page of the Greek text are Iamblichus, Galen, Theodoret, Theon Smyrnaeus, Clement, Justin Martyr, John Philoponus, and Athenaeus. [8] The quotations are of the greatest value for the transmission of the text. In some cases they preserve the correct reading where the entire direct tradition has been corrupted.[9]

Commentaries on Plato

Another witness to the text is ancient commentaries, in which that text is quoted and discussed. The oldest commentaries on Plato are lost, but a great number of neoplatonist commentaries survive from the 5th century AD, including works by Hermias, Proclus, Olympiodorus and Damascius, together with a 6th century fragment of a commentary on the Parmenides preserved in a palimpsest from Bobbio. The commentaries are often little more than student notes, but each note is often preceded by a lemma, i.e. a word or extract from Plato. While in theory these might have been modified themselves from later copies of the text, it has been shown that the lemmata in Proclus’ Commentary on the Timaeus must be as Proclus saw them, because his comments rely upon them being as they are.

There is also a papyrus of the 2nd century AD containing a commentary on a long stretch of the Theatetus. The work was probably composed not long before.


Further remains of ancient commentaries survive in the scholia in the margins of manuscripts of the BWT family. There are two sets. The first were entered in B by the hand of Arethas of Caesarea, the “Arethae scholia”. These are most abundant for the Gorgias and the Theatetus. The other set of scholia were added later to B in another hand, and also appear in T, and often in W. These have been called “scholia vetera,” although there is no evidence that they are earlier than the others. Neither set is very useful for textual questions, except occasionally.


Plato wrote in Greek, but in antiquity and later translations were made into other languages.

Cicero made a Latin translation of the Timaeus, and elsewhere in his works he quotes and translates many other passages of Plato, often at some length. In the 4th century AD Chalcidius translated into Latin the first part of the Timaeus and commented upon it. He dedicated it to a certain “Osius” who may have been bishop Hosius of Cordova. This translation passed into medieval Latin libraries, and influenced Dante. Both translations are preserved in manuscripts of the 9th century and later.[10]

Translations from Greek were made into Coptic, Middle Persian and Armenian. A fragment of a Coptic translation of the Republic 588b1-589b3 is preserved in codex VI of the Nag Hammadi library of Gnostic texts. The translation is of very poor quality, and initally went unrecognised. Agathias (Hist. II 29, 1-2) tells us that some Greek works were translated into Middle Persian for Chosroes I, and that he was especially interested in Plato and Aristotle, so probably Plato was among them. In Armenian translation the Timaeus, Euthyphro, Apology, Minos, and 12 books of the Laws have been preserved in a manuscript in the Mechitarist monastery in Venice. A translation of the Phaedo is lost. The translations may be the work of a Magister Gregorius (ca. 990-ca.1058), although others have argued for a 6th century date for the translation of the Timaeus. The translation is very literal, and seems to be based on a older text of the A-family.

Some researchers have suggested that Hunain ibn Ishaq translated the Republic into Arabic.[11] Several Arabic authors tell us that the Timaeus and other dialogues were translated into Arabic in the 9-10th century. Knowledge of Plato in medieval Arabic authors seems to derive from summaries made in Arabic or translated into Arabic.[12]

Dodd’s stemma for the Gorgias.

Analysis of the Medieval Manuscripts

The medieval manuscripts share certain characteristics. All of them derive from the collection of tetralogies known in antiquity, and other orders of the text are all secondary. They also share some obvious, and mostly unfixable, corruptions: doubled readings, rare interpolations, even rarer lacunae. It is clear that they all derive from a common ancestor.

But how old was this common ancestor? It must predate the invention of minuscule bookhand ca. 900, because none of the shared errors arise from misreading a minuscule bookhand.

The 2nd century AD commentary on the Theatetus shares two obvious corruptions with the medieval manuscripts. These corruptions must be earlier than the 2nd century. But the commentary also has a better reading than the medieval manuscripts in at least five places. In general the lemmas in the commentary agree much more with manuscript W than with B and T. All of this suggests that the common ancestor of the medieval manuscripts, and the 2nd century commentary, must be earlier still, and divided into two branches before the 2nd century AD; one the ancestor of the medieval codices, the other of the text in the commentary.

The roughly contemporary papyri of the Phaedrus confirm this. P.Oxy.1017 has a number of readings superior to the medieval mss, just as the commentary does. It also contains marginal and interlinear variants in a second hand, which cannot be conjectures to improve the text because in fact they do the opposite. The papyrus differs from the medieval text in 29 places, but in 8 of these places, the medieval reading is given in the marginal variants. This means that our medieval text, and also its errors, already existed in the 2-3rd century AD. P.Oxy.1017 tends to agree more with T than B. In fact P.Oxy. 1016 has similar features, but it also has readings found in inferior medieval manuscripts. So does P.Oxy. 2102 (2nd c.).

From this we can conclude that the medieval tradition has its origins in an ancient exemplar, and that many of the divergences found in the medieval codices are also ancient. Some of the manuscripts seem to continue an ancient family of the text, and presumably derive from a different uncial exemplar to the others. This is certainly true for the text of the Timaeus in F, which also shares errors with Plutarch, Galen, Eusebius, Proclus, Stobaeus and Chalcidius. The same is true for the text of the Republic and the Gorgias.

Date of collection and ordering

At what date did the works of Plato come into the form of a collection of tetralogies, in which they now are? Most likely during the early Hellenistic period. Pasquali argues that the collection contains an authentic but unfinished dialogue, the Critias; a dialogue only complete in its externals, the Laws, and, as an appendix to the Laws, it contains a work by Plato’s secretary, Philip of Opuntus under Plato’s name. This must mean that the collection itself dates back to a circle that had Plato’s work at its disposal and that felt obliged to continue it, i.e. the Academy. It cannot have been compiled by Plato’s immediate successors, who would have known very well what he wrote, because it contains a lot of spuria. So it must have been compiled at least a few generations after his death. One of the spurious dialogues, the Alcibiades II, seems Hellenistic rather than Attic. So perhaps the collection dates to the Academy of Arcesilas and Lacydes, of the first half of the 3rd century BC, at which date corruptions and interpolations may already have crept in.

What about the ordering? Diogenes Laertius tells us (III, 61) that “some, including Aristophanes the Grammarian” of Byzantium (fl. ca. 200 BC) classified the dialogues into groups of three; comprising only 15 dialogues, followed by an unordered mass of single dialogues. He also explains at length (III, 65-6) the use of critical signs in ancient copies of Plato, some of which signs have been preserved in medieval copies. But Diogenes Laertius also tells us (III, 56) that it was Thrasyllus the court astrologer of Tiberius who divided the dialogues into tetralogies, which seems far too late. Albinus ca. 150 AD in his introduction to the works of Plato (6) tells us that an otherwise unknown Dercyllides also arranged them thus.[13] The issues are discussed by Philip.[14] Pasquali declines to decide which came first, and is inclined to believe that both arrangements reflect only a bibliographical list, rather than the arrangement of any physical copies.

  1. [1] How they reached us is summarised in quite a lot of detail in some twenty pages of G. Pasquali, Storia della tradizione e critica del testo, 2nd ed., Firenze (1934, repr. 1988), pp. 247-269, from which most of the following material is taken. Useful list of the tetralogies at Wikipedia:
  2. [2]R. Brumbaugh, R. Wells, “Completing Yale’s Plato Microfilm Project”, in: Yale University Library Gazette 64 (1989), 73-5. JSTOR:
  3. [3]Query for author: Plato philosophus, exported the results to CSV, imported this into Microsoft Access as a table “Pinakes”, renamed the first 5 columns, and ran an SQL query: “SELECT country, town, library, collection, shelfmark FROM pinakes AS query GROUP BY country, town, library, collection, shelfmark;”
  4. [4]E.R. Dodds, Gorgias: A revised text, OUP (1959), p.34.
  5. [5]This material mainly from Dodds, Gorgias.
  6. [6]M. Joyal, “The Textual Tradition of [Plato] Theages”, in: Revue d’histoire des textes, 28 (1998), 1-54, p.8, n.30. Persee:
  7. [7]Boter.
  8. [8]Boter, p.285.
  9. [9]G. Jonkers, The Textual Tradition of Plato’s Timaeus and Critias, Brill (2017), p.387.
  10. [10]A list of manuscripts appears in the Wikipedia article,
  11. [11]Boter, The Textual Tradition of Plato’s Republic, Brill (1989), 279-80.
  12. [12]G. Jonkers, p.393-4.
  13. [13]Albinus, Eisagogue c. 6, online in English as “The introduction of Albinus to the Dialogues of Plato” here, p.315:
  14. [14]J. A. Philip, “The Platonic Corpus”, Phoenix 24 (1970), 296-308. JSTOR:

Bodleian Library manuscripts can now be downloaded as PDFs!!

I was looking at the online copy of the Bodleian manuscript of Plato, the “Clarkianus” 39 (here), when I discovered something wonderful.  We can now download the whole thing as a PDF!

This is just so amazing!  It also means that any cyber-attack can only do so much damage, if you have offline copies.

Here’s the screen grabs of what to do:

  1.  Go to the manuscript online:

2.  Click on the “Download” icon and you get this.

3.  Click on the download for the whole item.

Note that if you select a page range, it has to assemble that offline and email you, so it takes longer.

That’s it!  It’s actually the best user interface for downloads that I’ve yet seen.  Nice!

The only downside is resolution.  The download of this manuscript (871 pages) is a pretty massive 800mb.  If you look at folio 1r, the scholia are a bit fuzzy.  So for these you still need to use the website.  It would be good to have an “ultra-high res, kiss your disk space goodbye” option.  But it’s still a huge step forward.