What happened to the great and famed Library of Alexandria? There are many assertions. All are weak tea. The evidence never pans out as those making these assertions imply. So the most honest answer is the most frustrating one of all: “We really don’t know.” The answer could even be boring. It may have simply declined from ever-shrinking funding until it simply wasn’t a library anymore. That was, after all, a common outcome in the Dark Ages—and though Medieval Egypt exists outside that specific narrative, all its metrics of decline are similar. I did try to get to the bottom of this decades ago when studying for my PhD at Columbia University, and only found a lot of really annoying problems with all of the evidence—some of which almost completely unknown to academia.

In particular? The really messed up condition of the text of Epiphanius, a late 4th century Christian encyclopedist, living on the cusp of the looming Middle Ages. Which alone has wider-reaching significance: all too often, scholars are far too trusting of the extant published text of Epiphanius. And this matters to much more than the question at hand today. For example, a better understanding of the poor transmission of this text (and thus the high error rate in its preservation and contents) can both solve and confound problems regarding what Epiphanius said about, say, Christian sects, such as the Nazorians, who evidently believed Jesus was executed by stoning a hundred years before Pontius Pilate (as reported by Babylonian Rabbis in the Talmud). Really, anytime and anywhere you see someone cite “Epiphanius” for evidence of something, you have cause to pause and wonder whether we actually have what Epiphanius originally wrote about that. Probably we do. But confidence cannot be too high. You might need to check.

The following is a work in progress. I will update it as information comes in. It was originally written for an academic conference presentation at UC Berkeley in 2005 (a decade after I graduated there; I was a Columbia PhD candidate then), with some updates since (as I happened upon any), so it’s officially twenty years out of date (even the updates in the ensuing ten tears are incomplete). I have never been able to find the time to update its findings with all the new literature or scholarship that may have come out since. Though honestly, I am not aware of there even being anything new. What few discussions of the Library of Alexandria I’ve noticed coming out since appear to just repeat the same information—and often the same mistakes—of the literature I examined for my original study.

But I will be grateful to anyone who deep dives anything published after that (especially under peer review) regarding the Library of Alexandria and its fate, and finds anything not already covered here. If you do, notify me in comments and I’ll look into it, and update this article if needed. I’ll appreciate even a bibliography (just lists of academic books or articles on point published after 2005, even if you can’t access them). Because I might be able to follow-up on that (I also might already have seen it or have it in my own personal notes, but you never know, so it never hurts to supply the data; also, you will notice I already cite some in the article below, so you needn’t duplicate those).

Okay. That said, here we go.

Scribal Error and the Destruction of the Library of Alexandria

In his biblically-inspired treatise On Weights and Measures the Christian author Epiphanius devotes a chapter to the story of how the Septuagint came to be. This involves him in a digression on the creation of the great Library of Alexandria by Ptolemy Philadelphus, where we find this passage:

After the first Ptolemy, the second who reigned over Alexandria, the Ptolemy called Philadelphus, as has already been said, was a lover of beauty and a lover of learning. He established a library in the same city of Alexander, in what is called the Brouchion. This quarter of the city is deserted today. And he put in charge of the library a certain Demetrius, from Phaleron, commanding him to collect the books that were in every part of the world… (§9)

This is from the oldest manuscript of the treatise, a Syriac translation of the original Greek of Epiphanius. The translation here is from James Dean, Epiphanius’ Treatise on Weights and Measures: The Syriac Version (University of Chicago, 1935), 25. I also personally observed microfilm of the Syriac on folio 52b of Or. Add. 17148 (now in the British Museum), which is internally dated between A.D 648 and 659, and I verified the translation with experts in Syriac at Columbia University. This agrees well enough with another Syriac manuscript, Or. Add. 14620, from the 9th century.

This passage is significant due to the passing remark “this quarter of the city is deserted today,” which is occasionally cited as evidence that the great library no longer existed when Epiphanius wrote those words in A.D. 392 (Dean, Epiphanius’ Treatise, 2, 7-9). But it does not appear he actually wrote those words. I believe the manuscript evidence indicates this remark is an interpolation, which definitely dates before A.D. 659, and likely dates after A.D. 600. I will eventually discuss evidence that this is likely a scholium adapted from the Latin Res Gestae of Ammianus Marcellinus, written between A.D. 378 and 391. Thus its appearance in the text of Epiphanius in the early seventh century might not reflect any actual knowledge of the fate or status of the Library in Alexandria.

The Disputed Fate of the Library of Alexandria

The fate of this library is much debated, as you’ll find across all the expert literature, the best of which (but here in reverse chronological order, not order of merit):

• Christophe Rico and Anca Dan, eds. The Library of Alexandria: A Cultural Crossroads of the Ancient World (Polis Institute Conference Proceedings, 2017).
• Mostafa El-Abbadi and Omnia Mounir Fathallah, eds., What Happened to the Ancient Library of Alexandria? (Brill, 2008).
• Edward Watts, City and School in Late Antique Athens and Alexandria (University of California, 2008), 149–52.
• Lionel Casson, Libraries in the Ancient World (Yale University, 2002), 45–48.
• Roger Bagnall, “Alexandria: Library of Dreams,” Proceedings of the American Philosophical Society 146.4 (December 2002), 348–62. 
• Daniel Heller-Roazen, “Tradition’s Destruction: On the Library of Alexandria,” October 100 (Spring 2002), 133–53.
• Robert Barnes, “Cloistered Bookworms in the Chicken-Coop of the Muses: The Ancient Library of Alexandria,” in Roy MacLeod, ed., The Library of Alexandria: Centre of Learning in the Ancient World (London: I. B. Taurus, 2000), 70–75.
• Diana Delia, “From Romance to Rhetoric: The Alexandrian Library in Classical and Islamic Traditions,” The American Historical Review 97.5 (December 1992), 1449–67.
• Mostafa El-Abbadi, Life and Fate of the Ancient Library of Alexandria, 2nd. ed. (UNESCO, 1992).
• Luciano Canfora, The Vanished Library: A Wonder of the Ancient World (University of California at Berkeley, 1990), 66–99.
• Edward Parsons, The Alexandrian Library, Glory of the Hellenic World: Its Rise, Antiquities, and Destructions (Elsevier, 1952), 288ff.
• See also the early analyses of Rice Holmes, The Roman Republic and the Founder of the Empire, vol. 3 (Russell & Russell, 1967; orig. 1923), 487–89, and John White, The Scholia on the Aves of Aristophanes (Georg Olms Verlag, 1974; orig. 1914), xxx–xxxvii.

Was It Julius Caesar?

The most common claim is that Julius Caesar destroyed the library by accident, as several ancient authors do report. But this has been discredited by Barnes, Canfora, Holmes, and Parsons (and for context, see MacLeod, Library, 1–15): only book warehouses on the docks were burned. Canfora presents the clearest analysis to this conclusion. These warehouses were well-known. A commercial book trade flourished in result of the library’s presence. And it was custom (and once law) that any book passing through Alexandria had to be copied into its library. Writing around the turn of the 3rd century A.D., in his Commentary on the Epidemics of Hippocrates (at Kühn §17.607–08), Galen describes this custom, and reports that the library’s catalog even marked such acquisitions as “from the ships” (Heller-Roazen, 141; Galen even writes as if he has seen these catalog marks and considers books from Alexandria more reliable for this reason). The result would be that thousands of books just imported or about to be exported in the book trade, and thousands of books waiting to be copied or catalogued into the library, would be in these storehouses (ἀποτιθεμένοιϲ, as Galen describes them).

The original source for what happened was a now-lost account by Caesar’s contemporary Livy, whom the 5th century Orosius (§6.15.31) would later paraphrase as saying the fire hit thousands of books proximis forte aedibus condita, “just by chance near the housings,” meaning residences (and thus very distinctly not in a library), presumably meaning the residences of the ships (which had “perchance been docked,” classis forte subducta, suggesting a parallel structure in his sentence), or else imagining there were also homes or apartments along the docks—since we can’t tell if this is Livy’s wording or Orosius’s presumptive paraphrasing. But in any case, no mention of a library or even temple. Those would be behind the fortified walls of the royal quarter anyway, not adjacent to docked ships.

The deciding clue is that this matches the wording two centuries before Orosius in Cassius Dio (§42.38.2), which is often mistranslated, but actually says τε καὶ τὸ νεώριον τάς τε ἀποθήκας καὶ τοῦ σίτου καὶ τῶν βίβλων, i.e. what was destroyed were “the dockyards, including warehouses of both grain and books,” not a library (much less Museum or Brouchion). Orosius’s text mistakenly makes the number of books “four hundred thousand,” but from sources closer to Livy (like Seneca, On Tranquility of Mind 9.5) we know it was only forty thousand. This was simply telephone-gamed through several ancient accounts into the erroneous tale that the library was destroyed, misleading even Orosius. Caesar and his troops were actually stationed within the walls of the Brouchion quarter at the time, so there is no plausible way it burned; and a comparison of sources makes clear that what was originally a tragic destruction of thousands of books warehoused on the docks spiraled in the retelling into the whole library being burned. In reality, evidence from literature and archaeology confirm that the library remained all the way into the 6th century.

Even Ammianus Marcellinus (that 4th century historian whose account we’ll look at) attests to major Alexandrian scholarship (requiring a substantial library) in the Brouchion district during the Roman period, naming Didymus Chalcenterus (1st century A.D.), Aelius Herodianus (2nd century A.D.), and Ammonius Saccas (3rd century A.D.). And the Medieval Suda entry on “Theon” likewise reports that he was head of the same Museum that would have housed the Library—and that’s the late 4th century, well after Christians took over the empire. He was alive when Christians burned the Serapeum, Alexandria’s annex library—which was nowhere near the great library. So, no, the Christians did not burn the Library of Alexandria. Although in burning the Serapeum their target wasn’t the books, accounts do still suggest they cared little for their destruction (see my discussion of this point elsewhere).

We also also have Synesius, who discusses paintings he saw hanging in the Museum of Alexandria in the 5th century (Calvitii Encomium 6), and an extant documentary papyrus of the 6th century recording that a certain Asclepiades “worked all his life in the Museum” in “the great city of Alexander” as “a teacher of philosophy” (P. Cair. Masp. 3.67295). One might try to quibble about what “the Museum” means here, but the continuation of Alexandria’s fame as a center of scholarship well into the 4th century, with medical, mathematical, and astronomical schools, producing numerous commentaries and compendia, would not have been possible without some substantial library; and since the famed library would be in the Museum, it is not possible for that museum to still exist but not its library. We get this same sense from Strabo (§13.1.54 and §17.1.8), who confirms the Museum was there and well intact (and not “destroyed”) a mere decade after Caesar’s campaign. Yes, he does not explicitly mention the library, but that would be in the Museum, and thus cannot have been destroyed apart from it—and he could hardly have not remarked upon its loss if it were. Yet Strabo appears unaware any such thing happened.

Other sources and documents confirm this. After Julius Caesar we have documents and accounts confirming the library indeed thrived there: P. Merton 19 attests that in 173 A.D. Valerius Diodorus was “ex-vice librarian and member of the Museum”; a monument to Tiberius Claudius Balbillus, Egyptian Prefect under Nero, evinces he was president of the library (sometime around 56 A.D.); BGU 3.729 (144 A.D.) and P. Ryl. 2.143 (38 A.D.) provide examples of men granted the right to dine for free at the Museum for life; P. Kron. 4 (135 A.D.) discusses certificates of membership at the library in Alexandria. And the continuance of the Museum and Library is attested in several literary sources in the know: Suetonius, Life of Claudius 42 (which suggests Claudius added a wing to the library to house his books) and Life of Domitian 20 (which attests that in the late first century the library at Alexandria was still “the” place to go to replace the books of a library that burned in Rome); likewise, again, Galen, Commentary on the ‘Epidemics’ of Hippocrates (per above); and remarks in Philostratus, Vitae Sophistarum 1.22.524, 1.25.533, and 2.21.604; Athenaeus, Dinnersages 15.15.673d and 15.21.677e; and Scriptores Historiae Augustae, “Life of Hadrian” 20.2.

Even less plausible is the now oft-heard claim that the library was destroyed in a siege of either Aurelian or Diocletian during the chaos of the 3rd century. That is not in any ancient source, even erroneously. Not only does this have no basis in any evidence, it is contradicted by the evidence of the Museum and Library’s continuance in the 4th, 5th, and 6th centuries, as just noted. But that leaves two questions: the passage in Epiphanius (and its relationship to Marcellinus), and the Medieval claim that Muslims are the ones who destroyed the library (or, rather, its books). Let’s take those in reverse order.

The Arab Destruction Theory

The Arab destruction is doubted by many scholars, though for insecure reasons. The best case against it is made by El-Abbadi, but it is easy to detect logical fallacies in his approach, which appears more apologetical than skeptical. One of the oldest accounts of this tale is found in a mid-13th century Arabic work the History of Learned Men by Ibn Al-Qifti. There he reports that “a man named John the Grammarian of Alexandria in Egypt, a pupil of Severus, was a Coptic priest who was defrocked by a council in Babylon for some heresy concerning the Trinity” and when the Arabs took the city this man asked that the books in the library of Ptolemy Philadelphus “which had been guarded and preserved by kings and their successors to this day” be given to the city’s scholars. The commander of the army wrote to Caliph Umar about this, who replied: “Concerning the books you mentioned, if what is written in them agrees with the Book of God, they are not required; if it disagrees, they are not desired. Therefore, destroy them.” So the books were burned as fuel for the city baths over the next six months, and the library was no more. Did this happen?

The objections to this account are not very weighty. 

First, it is argued that this source is nearly six hundred years late. But that’s a weak argument here. We lack a great many works from the intervening period, and those that do survive are brief and fragmentary with regard to the capture of Alexandria, and thus it is not improbable that no earlier report would be extant even if it existed. Which makes this too weak as an argument from silence. We accept statements of such an age in other cases and thus it is not a weighty objection in and of itself. For example Arrian is in many cases the sole preserver of certain early accounts of Alexander the Great, yet he also wrote over five hundred years after the facts. More importantly, that the library was burned by order of Umar is corroborated by an independent Arab source who actually wrote a decade or more before Al-Qifti, a certain Ibd Al-Latif of Bagdad, who toured the ruins himself in the early 13th century: see Isya Joseph, “Bar Hebraeus and the Alexandrian Library,” The American Journal of Semitic Languages 27.4 (July 1911), 335–38. It is apparent, then, that this story does derive from earlier Arabic sources now lost to us.

Second, it is argued that the “John the Grammarian” is John Philopon, who was long dead by 642 A.D., so “the whole account” must be a legend. However, that identification is not secure. First of all, it does not appear to be describing that John. Philopon was not “a defrocked Coptic priest,” in fact he wasn’t even declared a heretic until a century after he died, and by a council in Constantinople, not Babylon. Philopon was also not a pupil of any Severus—he studied under Ammonius and Proclus. The Severus meant is probably the founder of the Monophysite movement that Philopon did sympathize with later in life, so “pupil” might mean simply a student of Severus’s teachings, not the man himself. But that can describe any number of people, even named John, for several centuries.

So Al-Qifti (or his source) might be referring to a different John, or someone less famous who was later replaced in the “story” by the more famous one, albeit with a completely garbled biography. We thus cannot reject the entire account on this one transmission error. John was an exceptionally common name, and “Grammarian” an exceptionally common epithet. So Al-Qifti could have mistook the John of his source as John Philopon. That would not discredit the core account, which is corroborated by Al-Latif without reference to anyone named John. In fact, El-Abbadi reports that the whole passage describing this John is almost a verbatim copy from a 10th century work by Ibn Al-Nadim, which is likewise ambiguous as to whether Philopon was meant, yet which does not mention his connection to the library’s destruction (El-Abbadi, 172). Thus, it looks like Al-Qifti added this material about John to a different story he acquired from some other source that lacked this detail. So this is not a strong argument against the historicity of the event.

Third, El-Abbadi suggests that since the description of John (and also some material cribbed from The Letter of Aristeas, an early Greek source about the origin of the library) can be found in earlier extant sources, but not so for the account of the destruction, we should assume the latter was invented by Al-Qifti. But this is not secure reasoning. We may have simply lost his source for it. True histories often used multiple sources to fill out a description. And since the report of the burning is also heard from Al-Latif, an earlier independent scholar, Al-Qifti clearly did not invent it himself. There was certainly an even earlier common source shared by both. And due to the scarcity of extant texts and the fragmentary and sketchy nature of those that do survive, even for a true story it is unreasonable to expect more than we have.

Several weaker arguments can be readily dismissed, such as that all the books would have been of vellum (or parchment; paper vellum did not yet exist, but calfskin vellum did), which El-Abbadi claims doesn’t burn. In fact, the vast majority of books there would still have been of papyrus, especially in an old, declining library, and most especially in Egypt where papyrus was far cheaper than vellum. And vellum certainly does burn (it is literally animal skin). Likewise, that there could not possibly have been enough books to fuel Alexandria’s baths for six months is no argument, either, because it is not stated that they did, only that it took six months to dispose of them in this way. Given the enormity of the library and certainly the limited manpower that would be sequestered on the task, it may well have taken six months to clear out the inventory. But exaggerations of numbers and scale are commonplace in ancient histories anyway. They do not discredit the core of a report. As to the argument that the use of the books to fuel the baths sounds too economical to be true, this actually weights for authenticity. It would be implausible that the library itself would be set afire, for the risk to the whole city would be too great, the building itself would remain of great value and utility, and the fact that safe fires were already regularly maintained to heat the baths would be an obvious opportunity for disposing of the books. In fact, as heated baths were no longer extant in the centuries after the Arab conquest, this detail sooner suggests historicity. How likely is it that Al-Latif would know Alexandria had heated public baths then?

It is also possible that the Arabs actually destroyed the library by accident, an event which inspired the more damning stories now extant. Yet those stories, even if exaggerating or erroneously elaborating the details, do not describe the improbable. Arab interest in Greek scholarship would not begin for another century at least, and an illiterate, fanatically religious army would have little respect for heathen books—or probably little interest in even absorbing the expense of maintaining them. Moreover, such book burning appears to have been a common practice of the Muslim armies of that day, as it is recorded on many other occasions by Arab authors, even in official chronicles, and the story fully agrees with the earliest Muslim sacred belief that the Koran had superseded all earlier books and thus rendered them obsolete (Joseph, “Bar Hebraeus,” 337; Zaydan, Tarikh, 45; El-Abbadi, 221n58). 

In comparison with this evidence, El-Abbadi’s attempt to locate a motive for the invention of such a story in the late 12th century is wholly born of his imagination and quite far-fetched (El-Abbadi, 172–9). He claims it was a parable defending Saladin’s sale of the Fatimid library, saying it is better to sell books than to burn them, in possible allusion to a legend of that library also being deliberately burned. But there is no basis for such an interpretation. A sale of the books is not mentioned as an alternative in the Alexandrian story. The only polemical comment is that one should wonder at how many books were destroyed. There is no commentary on how they should have been saved or sold off, or any reference to the example of Saladin or the Fatimid library—yet we would expect this if such were the intent of telling this story. This implausible thesis also hinges on Al-Qifti’s association with Saladin—yet this fails to explain how Al-Latif came by and reported the same story decades earlier. Finally, the very motive El-Abbadi alleges could just as easily have inspired including a true story as a false one. Indeed, the legend of burning the Fatimid books could be a legend inspired by the actual burning of the Alexandrian books. So even if there was such a motive to include the story, we cannot then assume the story was made up—rather than used because it was true.

The reasonable conclusion is that the library’s destruction by Muslims in 642 is plausible. The arguments against it do not carry much weight, and the evidence for it is not incredible, being corroborated by the fact that scholarship in Alexandria suspiciously ended around the very same time. But we have nothing better than that to confirm it, either. It’s thus the most likely fate of the library, but only in comparison to all the others. It is still not an event we can be at all certain happened.

Which gets us back to Epiphanius and Marcellinus.

Epiphanius

Regardless of any other evidence we could examine regarding the fate of the library, the remark attributed to Epiphanius would entail, if true, that the entire Brouchion district (also known as the Royal Quarter) was deserted at the end of the fourth century, and thus even the famous Museum (and its even more famous Library) was no longer in existence—that in fact the area wasn’t even inhabited. This was certainly untrue when Epiphanius was alive. The Brouchion district and the Museum and its Library were all still well intact as late as the fifth century, and even if it ever came to any harm before that, archaeology confirms it was not only inhabited but thriving when Epiphanius wrote.

For example, there is evidence a fire collapsed some buildings in the Brouchion in the third century, but that same evidence confirms they were fully rebuilt and in service well before the end of the fourth (El-Abbadi and Fathallah, What Happened?, 80–86). Other evidence abundantly confirms this. Besides the scholarship listed above, see:

• Konstantinos Staikos, The History of the Library in Western Civilization, vol. 1 (Kotinos, 2004), 1.157–248, 1.283–88 and The Great Libraries: From Antiquity to the Renaissance (3000 B.C. to A.D. 1600) (Oak Knoll, 2000), 57–90.
• Paul Chapman, “The Alexandrian Library: Crucible of a Renaissance,” Neurosurgery 49.1 (July 2001), 1–13.
• Judith McKenzie, The Architecture of Alexandria and Egypt, c. 300 B.C. to A.D. 700 (Yale University, 2007), 50.

Which is confirmed by all the evidence above, demonstrating the Museum’s scholarship (and hence library) continued unabated.

So where did that comment in Epiphanius come from? The Weights and Measures happens to also survive in 10th and 13th century Georgian translations, which lack this remark (M.-J. van Esbroeck, Corpus Scriptorum Christianorum Orientalium 460–461 (1984) = Scriptores Iberici 19–20). They read simply: “He,” that is, Ptolemy Philadelphus “established a library and collected the books that were in every part of the world,” omitting the entire reference to Demetrius of Phaleron as well as everything else, which reading is further confirmed by an Armenian translation, which dates at least to the 10th century. These translations may be related, but as such suggest an older tradition of the text: see Michael Stone, “Concerning the Seventy-Two Translators: Armenian Fragments of Epiphanius, On Weights and Measures,” Harvard Theological Review 73.1/2 (January–April 1980), 331–36, and Michael Stone and Roberta Ervine, The Armenian Texts of Epiphanius of Salamis De Mensuris et Ponderibus (Peeters, 2000) = Corpus Scriptorum Orientalium v. 583 t. 105 (for the passage of our concern, 82, 92, and 99, with introduction, pp. 1–5ff.).

This raises the likelihood that the remark in question was a scribal interpolation occurring in a separate ancestral line of the Greek text (ending up in our early Syriac), probably of a marginal or interlinear note, a common error in textual transmission. I provide a full discussion and bibliography of the scholarship regarding this common scribal phenomenon in Richard Carrier, “On the Accidental Interpolation of ‘Who Was Called Christ’ in Josephus, Jewish Antiquities 20.200,” Journal of Early Christian Studies 20.4 (Winter 2012), 489–514 (affordably reproduced in Hitler Homer Bible Christ). In the entire Syriac text of Weights and Measures we find scores of words and phrases that appear there but not in any extant Greek manuscripts corresponding (we can even see one unmistakable case of interpolation in §21), proving that this form of error was indeed prevalent here. Likewise, in the marginalia of this manuscript we find both kinds of material on almost every page: a grammatical note and a gloss on a rare word are both added on folio 52b, while a full sentence that was omitted by accident was added in the margin on folio 52d, both encompassing the very same section that concerns us, and either one can be easily confused for the other, since no special markings exist to distinguish them.

Which is how such interpolations proliferated: a marginal note would be confused for accidentally omitted text, and thus duly “reinserted” when a copy was made. In fact, in encyclopedic texts like the Weights and Measures (anything we would call a “reference book,” like a manual or a dictionary), there is also a known tendency to do this deliberately, as scribes might treat such texts as something merely useful and thus open to expansion and improvement (a phenomenon that I already noted creates problems for understanding textbooks by Heron of Alexandria).

We see something like this even next to our passage of concern, where the text reads “in the same city of Alexander.” Here the scribe, perhaps sensing something odd about this, recommended “Alexandria” as an alternative reading in the margin. Another notable example is that in §11 the Syriac reads “the first library, which was built in the Brouchion, as I have already said,” whereas all other manuscripts (Greek and Georgian) omit the words “as I have already said,” which suggests that this may have been added after the Brouchion remark was interpolated in §9—and therefore §11 appears to have really been the first time Epiphanius mentioned it, so it wasn’t in §9 originally.

There are much later Greek manuscripts containing the original language of the text. On which, see Karl Holl, “Die handschriftliche Überlieferung des Epiphanius,” Texte und Untersuchungen zur Geschichte der altchristlichen Literatur 36.2 (1910). A more recent survey of the entire manuscript evidence is available in Elias Moutsoulas, “La tradition manuscrite de l’oeuvre d’Epiphane de Salamine De mensuris et ponderibus,” in Jürgen Dummer, ed., Texte und Textkritik: Eine Aufsatzsammlung (Akademie Verlag, 1987 = Texte und Untersuchungen zur Geschischte der altchristlichen Literatur Bd. 133), 429–40, and “L’oeuvre d’Epiphane de Salamine «De mensuris et ponderibus» et son unité littéraire,” Papers Presented to the Sixth International Conference on Patristic Studies Held in Oxford 1971: Part I (Akademie Verlag, 1975 = Studia Patristica 12 = Texte und Untersuchungen zur Geschischte der altchristlichen Literatur = Bd. 115), 119–22. Although it is still important to compare O. Viedebantt, Quaestiones Epiphanianae Metrologicae et Criticae (Lipsiae, 1908).

The Greek manuscripts vary at precisely this point in the text, and modern critical editions try to render something better. But none of the actual manuscripts themselves produce a grammatically correct sentence for the passage of our concern. The outdated Migne’s edition, based on the 17th century edition of Petavius, i.e. J.-P. Migne, Patrologia Cursus Completus: Series Graeca 43:249C-252A (1857), reads as follows (transliterated and then translated as literally as possible):

ho gar meta ton Ptolemaion deuteros basileusas Alexandreias Ptolemaios, ho epiklêtheis philadelphos, hôs proeirêtai, philokalos tis anêr kai philologos gegenêtai. hostis bibliothêkên kataskeuasas epi tês autês Alexandrou poleôs en tô brouchiô kaloumenô klimati (kai esti touto tês autês poleôs, erêmon tanun hyparchon), enecheirise Dêmêtriô tini tô Phalarênô tên autên bibliothêkên, prostaxas synagagein tas pantachou gês biblous…

For after this Ptolemy there was a second Ptolemy ruling Alexandria, who was called Philadelphos, for as was said earlier he became a man who loved beauty and literature. This is the one who, having furnished a library in the same city of Alexander in the quarter called the Brouchion (and it is this [area] of the same city that is now deserted), entrusted the same library to a certain Demetrius the Phalarene, ordering [him] to collect books from every part of the world….

This cannot be correct, for the remark “and it is this [area] of the same city that is now deserted” is grammatically out of place. The subject of the sentence is Ptolemy [Philadelphus], yet the verb (“entrusted”) does not appear until after this aside, which introduces an entirely new verb and subject (“this” = the quarter), without any relative pronoun or other appropriate construction. This would grammatically make “the quarter” the subject of the following verb “entrusted,” which is impossible. No competent author of Greek would have made such a mistake. Even as an interpolation it is inept.

Petavius, whose edition Migne reproduces, was transcribing the 16th century Codex Parisinus Graecus 835, which is a copy of a copy of the Codex Jenensis (ms. Bose 1), transcribed in A.D. 1304 and now in the Universitätsbibliothek in Jena, Germany (Dean, Epiphanius’ Treatise, 4–5). But Jenensis does not read en tô brouchiô kaloumenô klimati kai esti touto, “in the quarter called the Brouchion, and it is this…,” but en tô brouchiô kaloumenô klima kai esti touto, “a quarter in what is called the Brouchion, and it is this…,” which is also grammatically incorrect, and even less logical (you then get “…who, having furnished a library in the same city of Alexander a quarter called the Brouchion and it is this of the same city that is now deserted, entrusted it…”). For fans of Monty Python’s Life of Brian, this is akin to “People called Romanes they go the house.” This was already noted in Dindorf’s edition of the text, Epiphanius: Opera 4.1 (Leipzig, 1859–62), 12 (166B). But I confirmed this and other details myself by viewing a microfilm of the Jena manuscript at Columbia University. The relevant passage appears at the top of folio 164^a. The obvious intent was to make it read “in the so-called Brouchion (this quarter is also…)” but, without a proper connecting word or construction, klima would have to be in the same case as brouchiô yet is not, and this is exactly what we see the later correctors tried to fix in various ways.

There is a more important difference. Jenensis has as the main verb enecheirêse (from egcheireô) instead of enecheirise (from egcheirizô), which means “put his hand to” (as in “attempted” or “began”), not “he entrusted,” which suggests the text has changed—the material about Demetrius wasn’t originally there, and adding it created a conflict with the verb that later scribes would then try to fix. Indeed, Codex Vaticanus Graecus 1196, a 16th century copy of Codex Vat. Gr. 1142, even has enecheirêsan, “they began,” which makes even less sense, but illustrates what a mess the transmission of this sentence had become. What the oldest surviving manuscript of the original Greek text actually says is that Ptolemy “began” or “put his hand” to Demetrius, which is also unintelligible.

But the earlier verb makes sense if we reconstruct the Greek from the Georgian text, using only the Greek words in Jenensis: hostis bibliothêkên kataskeuasas […] enecheirêse synagagein tas pantachou gês biblous. Then it says Ptolemy, “having established a library, began collecting the books that were in every part of the world,” without any of the intervening digressions (about Demetrius or the Brouchion).

This is confirmed by another copy of the same text used by the Jenensis scribe but only surviving in a manuscript of a later date: Codex Vindobonensis Suppl. Gr. 91, from the 14th or 15th century, which also reads enecheirêse (this manuscript also corroborates Jenensis in reading klima kai esti). So an independent witness, combined with two of the oldest Greek manuscripts, affords additional evidence confirming that almost the entire passage of our concern is an interpolation (or several interpolations) of marginal glosses, which were originally stand-alone remarks in the margin, but were hastily inserted into the text without close attention to the grammar.

In support of this conclusion, an independent attestation of this passage in Greek—in Codex Parisinus Graecus 146, which contains an excerpt of just this chapter about the creation of the Septuagint—completely omits the words “this quarter of the same city is now deserted” and says only klima de esti tês autês poleôs, “and [this] is a quarter of the same city,” which is also awkward but betrays the absence of any reference to the quarter being vacant. This looks like a corruption of the original marginal interpolation, with some scribe emending klima kai esti to klima de esti and then dropping the touto as superfluous. Not only is this likely given the authority of Codex Jenensis, but the scribe of Codex Parisinus Gr. 146 seems to have been prone to such meddling, since he also carelessly emended Brouchion (a word he evidently didn’t understand) to archeiô, “headquarters,” and made several other changes to this passage.

The most recent critical edition of the surviving Greek (and not the earlier translations of it) is by Elias Moutsoulas, “TO «PERI METRÔN KAI STATHMÔN» ERGON EPIPHANIOU TOU SALAMINOS,” Theologia 43 (1972), 631–70 (635–36), and 44 (1973), 157–209 (168), but it doesn’t account for all of this evidence. Moutsoulas reconstructs a plausible text out of all of the Greek manuscripts available, generating klima de esti touto tês autês poleôs erêmon tanun hyparchon, but that is still awkward grammatically. It gets us “…who, having established a library in the same city of Alexander, in what is called the Brouchion, and this is a quarter of the same city now standing empty, entrusted…” This does not seem likely for Epiphanius to write. The clause “and this is a quarter of the same city now standing empty” looks inserted, and awkwardly. It interrupts the sense, and it does not anticipate that the verb of the main clause follows this subordinate clause, so the conjunction (de) is incorrect here (that is supposed to indicate the start of a new sentence).

It is clear that something very much like the Codex Jenensis text was used by the Syriac translator, since all the same material appears there nearly verbatim. In particular, both use the awkward phrase “city of Alexander,” unlike the Codex Parisinus Gr. 146, which has “the city Alexandria” (and notably, “Alexandria” is a marginal gloss in the Syriac mss. and thus might indicate a known variant or correction). Moreover, the Greek letters Phalarênô are written in the margin of the Syriac manuscript near where Demetrius of Phaleron is mentioned, thus preserving the exact case of the word as it appears in the Greek of Jenensis, as well as the same unorthodox spelling (i.e., though it is a correct reference to Demetrius’ home city of Phaleron, both manuscripts have “the Phal-a-rene,” substituting a for ê).

In contrast, Codex Parisinus has the incorrect Phalêrei, from phalaris, or “coot,” a bird known for its bald white head. This is telling because the Syriac manuscript contains a marginal gloss explaining the Greek word Phalarênô as meaning “bald white head.” Though that is not what the word is doing there (it’s a reference to the city of Phaleron, not the bird; although the city may have been named after that bird, this would have no relevance to Demetrius). But this still evinces the very same confusion with the word phalaris. So this gloss probably existed in the Syriac’s Greek exemplar, and this probably, through a different textual tradition, influenced the text that finally appears in Parisinus Gr. 146. These facts also argue against the possibility that Jenensis is a back-translation into Greek from the Syriac, which is also confirmed by the fact that Jenensis preserves what must have been the original Greek verb (enecheirêse).

I struggle to make sense of all the confusions and variants here, and have experimented with various hypotheses. But I’ve given up trying to reconstruct a transmission history for this garbled text. All I can say with some confidence is that the one line about the quarter being empty does not belong. Indeed, when viewing a microfilm of Codex Jenensis myself I observed that its scribe placed a dot between klima and kai (as well as before the en that begins the whole comment, which is likewise odd if this remark was originally in the text), a practice that elsewhere in this manuscript indicates a break between independent sentences or thoughts (hence where a modern editor would place a period, comma, or other punctuation). So this scribe clearly understood en tô brouchiô kaloumenô klima, “an area in the so-called Bruchion,” as a complete thought (contrary to the reconstruction of Moutsoulas), which looks exactly like a marginal note: it is brief and digressive but informative, yet fails to fit grammatically in the text (because it was not originally meant to be there). The entire gloss may even have included the observation of its supposed desertion (the whole digression deriving from Marcellinus, as I’ll discuss shortly).

We have other clues. According to van Esbroeck (“Une forme inédite de la lettre du roi Ptolémée pour la traduction des LXX,” Biblica 57 (1976), 542–49), the Georgian text (which also lacks this aside) is attested at the beginning of the 7th century, which suggests that remark may have been added, or at least interpolated, sometime between A.D. 600 (when it doesn’t appear in the Georgian) and 659 (when it appears in the Syriac). This can’t be certain, since the Syriac may have been translated or copied from a different Greek or Syriac textual tradition than the Georgian, or the Georgian translator may have ignored the marginal note even if it appeared in his exemplar. Either way, van Esbroeck observes that there are many similarities between the Georgian and the Syriac texts, strongly suggesting that they stem from a common tradition.

Marcellinus

Another clue is that the clauses “area called the Bruchion” and “lies now deserted” are both peculiar phrasing, yet correspond oddly well to what is written in the Latin Rerum Gestarum of Ammianus Marcellinus (§22.16.15), suggesting that this was a gloss added by a scholar familiar with this work. Marcellinus reads:

Sed Alexandria ipsa…Aureliano imperium agente civilibus iurgiis ad certamina interneciva prolapsis dirutisque moenibus amisit regionum maximam partem quae Bruchion appellabatur, diuturnum praestantium hominum domicilium.

…when Aurelian was emperor, after civil quarrels collapsed into murderous conflict and the walls were pulled down, [Alexandria] lost the best part of her districts: the one called the Bruchion, for a long time the home of outstanding men.

The Latin phrase Bruchion appellabatur is identical to the Greek Brouchiô kaloumenô now in Epiphanius (differing grammatically but not semantically); and klima is an expected Greek translation of the Latin regio; and most notably, erêmon tanun hyparchon (“stands now empty”) looks like a scholarly paraphrase of amisit (“lost,” a valence of the Greek word erêmon). Someone who read that could then easily scribble their own remark in the margin of his copy of Epiphanius that “this is now gone,” not realizing that that isn’t what Marcellinus meant. His remark concerned a time when Emperor Aurelian put down the revolt of Firmus in A.D. 273, but contrary to modern interpreters (like Wikipedia), Marcellinus never says anything about “burning down” or “razing” the Bruchion. He only mentions “tearing down the walls” of the Brouchion (i.e. removing its fortifications), which does not mean “razing” or “destroying” the district. By this choice of words, in fact, Marcellinus was saying only that the district was no longer separated from the rest of the city and thus its “identity” was lost in 273, not that it actually became uninhabited (or even that it was ruined). In any case, it didn’t—as we know from all the literary and archaeological evidence already cited above.

It should be noted that Marcellinus is also quite unreliable. He completely garbles the history of the libraries of Alexandria. Contrary to modern texts that have him say there were ‘two libraries’ (which would be correct), in fact he wrote only ‘priceless libraries’, and confuses them all as the same library. He also gets wrong what happened to them, claiming to have confirmed “they” were in the Serapeum, and that was destroyed by Caesar. But the Serapeum is nowhere near Caesar’s fire, and was not the location of the great library but only a small annex on the other side of the city. And we know from other sources that Caesar’s fire did not destroy even the great library but only book warehouses on the docks. And contrary to Marcellinus’s assertion that “all” sources agree with him, in fact he simply paraphrases a single unreliable source, the 2nd century Latin author Aulus Gellius. Not only is the number of books identical in both sources, but a key part of the Latin is nearly identical—only the bracketed words are absent from Ammianus: bello [priore] Alexandrino dum diripitur [ea] civitas (Attic Nights 7.17.3), an impossible coincidence. See: J. den Boeft, et al., Philological and Historical Commentary on Ammianus Marcellinus XXII (Egbert Forsten, 1995), §16.12–22.

But even with these errors we can confirm Marcellinus was not talking about Aurelian destroying the library, because when he does discuss the “libraries,” he is not even aware that the Brouchion is the location of one, and erroneously believes they were destroyed under Caesar and thus were not even around for Aurelian to destroy. This also means we know how Marcellinus describes destroying libraries—with words conspicuously absent from his unrelated account of Aurelian removing the walls of the Brouchion. So if he had meant libraries and the Museum were destroyed, that’s what he would say. Instead, he only mentions the walls being removed and the Brouchion losing its distinction—there was no longer a Brouchion because of a mere administrative deletion, not a physical deletion of its contents. It’s like saying San Francisco’s “China Town” ceased to exist because it was no longer called that and the arches advertising its borders were removed; which is not saying it was destroyed. It’s still there. It’s just now not a distinct neighborhood.

In the end, I think it’s more probable than not that in the original Weights and Measures §9 Epiphanius most likely wrote that Ptolemy Philadelphus “established a library and began collecting the books that were in every part of the world.” Everything else inserted into this sentence is an interpolation of a marginal gloss (either one or several, or with scribal emendations as well) sometime between A.D. 600 and 659. It could be earlier, depending on when the Greek text used by the Georgian and Armenian translators diverged from the Greek text used by the Syriac translator, but in any case, well after Epiphanius. And even the gloss itself was a telephone-gamed misunderstanding of the text of Marcellinus, and thus not a correct account of the state of the Alexandrian library.

Conclusion

No source reliably says the Library of Alexandria was destroyed, and abundant evidence confirms it continuously survived all the way into the Middle Ages. The 500s are the last century we can confirm it existed. All the evidence cited to try and argue it was destroyed before that all falls apart on inspection, being demonstrably erroneous, and contradicted by actual archeology, and by other sources more reliable. Indeed many of the claims of its destruction don’t exist in ancient sources at all, even erroneously. No ancient source says Aurelian or Diocletian had any effect on it. The only story that is hard to disprove is the Muslim destruction. The library finally does disappear from the record precisely in that century. And though there could be errors in the subsequent legend of it, that isn’t sufficient to argue it didn’t happen. But neither can we confirm with confidence that it did.

Which leaves one important point to make. We can still infer that the great library of Alexandria was in decline since the 3rd century—not only did funding for all libraries decline in that century (a century of civil and economic decay), and not only was that followed by less means and enthusiasm for restoring them in the 4th century (than one would have found in the 2nd), but also by the 6th century the library had been controlled by Christians for probably hundreds of years. Its holdings no doubt decayed in favor of Christian literature, and literature favorable to Christianity, as its dwindling funding was simply shifted away from preserving everything else. This is a more pernicious kind of destruction. Nothing so shocking as simply burning the books, but merely letting them rot away from disinterest over many decades—tossing them in the trash once no longer worth keeping and replacing their shelf space with more preferable texts, or even erasing them and writing preferable texts right over them. A cultural cannibalism that typified the age.

As funding dwindled into non-existence, it could well be that the once-great library of Alexandria simply petered out from disinterest. Its last remaining books may even have been used as fuel for heaters and furnaces—given that their decay had rendered them unfit for anything else. It is also possible the library burned in an accidental fire, a fate that has erased many a library before and since, and this event simply went unrecorded. As a population still resides on the site, archaeology cannot conclusively prove this never happened, particularly if it did in, say, the 6th or 7th century, when whatever is there now started to be built atop whatever was there then. So no one might be to blame for its fate but Mrs. O’Leary’s cow.