George Houston's new book, Inside Roman Libraries: Book Collections and Their Management in Antiquity (UNC Press, 2014) offers a valuable overview of what we know about book collections in the Roman World. I limit my remarks to one chapter. My only real concern is the number of assumptions that are made in chapter four, and not with its overall conclusions about intellectual life in Oxyrhynchus.
In Chapter Four, "Books Collections of Oxyrhynchus," states generally "Grenfell and Hunt noticed that the ancient residents of the town had occasionally thrown out not just random materials, but substantial collections of interconnected written materials (1). These concentrations, coherent in date and content (2), usually consisted of official documents or of some individual's letters and accounts. But in the inter of 1905-06 Grenfell and Hunt discovered three concentrations (3) each of which contained numerous literary papyri." About the first of these, the author reports, that they "were confident that they all had been discarded at one time and all came from a single library."
"Interconnected written materials (1)" and : "coherent in date and content (2). When documents are considered, there are clear criteria for determining their interconnectedness: date and prosopography. The archive of the sons of Ptollas, some of which were held together by string (P.Mich. vi 399-407), were found together in a sealed dump at Karanis (*158). That is, structure 158 was built on top of this dump, and it has, in theory, a fixed terminus ante quem. More often, discarded archives were deposited in a group, as was the case with the Archive of Aur. Isidorus, even though it was not excavated, but found during an illegal dig and destined for the antiquities market. A few papyri made it onto the market, and were sold to western collections (Merton and Michigan), but most of the archive was recovered by the authorities, and so remained together.
Connecting literary texts into a group is a trickier affair than Houston lets on. He simply relies on the confidence of Grenfell and Hunt, and says no more about it. Sometimes, as in the case of the Archive of the sons of Ptollas, and in Archive of Leonides, son of Theon, papyri are found tied together. (TM Archives). But for the most part, it is a matter of conjecture. Grenfell and Hunt were pre-scientific excavators, and so give no accurate record of the dumps they dug, as Bingen did for Mons Claudianus ("Dumping and the Ostraca at Mons Claudianus," in D.M. Bailey, Archaeological Research in Roman Egypt, JRA Suppl. 19, 1996), 29-38).
As to the second point (2: "coherent in date and content"), dating is problem, not only because most of the literary papyri the author groups together must be dated palaeographically, that is to say, not historically, but because they are not coherent in their dates. For the "Second Find" at Oxyrhynchus (1906), most have a range of three centuries (first-third AD, pp. 159-62). We have examples of literary texts that from a coherent group, in topic and date, that were deposited together, that is the Tura Find of 1941. Even here, however, we do not really know the date of deposition. One might suppose they were deposited in cave 35 at Tura after Origen and Didymus' work was declared anathema by Justinian in 553, but we cannot know with certainty. These were whole books, not partial, fragmentary or worn out, in contrast to the Oxyrhynchus finds. It is salutary to have some comparanda when evaluating with what we think we know about literary papyri in dumps.
Another problem is the author's assumption that, first, we can know the date these "concentrations" of literary papyri were deposited, and second, we may safely assume they were deposited together, at the same time, not gradually. "As a reasonable and conservative guess, I will assume they were thrown in about 400 AD" (p.162). Perhaps I am misunderstanding here, but if he means there is either a terminus ante quem for the dump (which we can hardly say, because there we do not know all the datable objects that were found there, and, since, I assume, and may be wrong, the dump was not sealed), or can have any good idea about the precise date of deposition. We may be able to ascertain the earliest datable text or object form this layer of the dump, and so arrive at a rough terminus post quem, apart from Grenfell and Hunt's observation that they were excavating a fourth-fifth century layer. Grenfell and Hunt report that "at Oxyrhynchus it is not uncommon to come upon large groups of papyri which have been thrown away simultaneously" ("Excavations at Oxyrhynchus" in Oxyrhynchus: A City and its Texts (edd. A. Bowman et al., EES Greco-Roman Memoirs no. 93, 2007), p. 360. This assumption may be reasonable, but it cannot be demonstrated, and must be understood as conjectural rather than factual.
Second, if, as is assumed, these texts (First Find, Second Find, etc.) were deposited in the dump simultaneously, how can we know if this was a primary or secondary deposit? That is, can we know that discarded bookrolls were taken straight to library to the dump, in a sort of spring cleaning? These rolls, or their detritus, of might just as well been collected in the library first, (garbage often accumulated in spare rooms, or just outside doorways) all at once, or over time, and only later taken to the dump. Therefore the dump would be a secondary deposit, and we are another step away from knowing when these book rolls, or parts of them, were discarded originally.
More needs to be said, about how one may apply Houston's results as to the longevity of books to Christian text, especially in light of the fact that he deals almost exclusively with book rolls, not codices. Given the early Christian use of single-quire codices, notoriously clumsy and easily damaged, his findings cannot be imported without qualification into the context of the longevity of Christian books. But that is for another time.