Lost Books of Antiquity: Summaries That Outlived the Books

How a lost book leaves evidence

The evidence for a lost book is sorted here into five kinds of trace. A summary keeps the outline but not the words, while a quotation keeps a few exact words. A scrap of an old copy keeps part of a page, a catalogue keeps a title and often a length, and a passing mention may keep only the name. The chart below sorts the 12 works by these traces.

In this set, quotation is the commonest trace: 10 of the 12 works are quoted somewhere in these sources. The table gives each work's main witness and its latest dated witness. Here a latest dated witness means the latest ancient or medieval text in these sources that copies, quotes, sums up, lists or names the work, other than to say it is gone.

Dates come from the editions cited or, where those give none, from English Wikipedia's summaries. The page on how the site follows a paper trail explains how records are dated. The CSV names a source for every row.

Table data: lost-books-antiquity-catalogue.csv, with the sources and links for every row.

Known from a summary

Livy wrote his History of Rome between 27 and 9 BC, and it once ran to 142 books. Of those, 35 survive: books 1–10 and 21–45, four of them with gaps. The Periochae, short summaries made in the 4th century AD according to English Wikipedia, cover every book except 136 and 137.

The summaries are terse. In the Latin text on Wikisource, the one for book 138 has three short sentences: the Raeti are subdued by Tiberius and Drusus, Agrippa dies, and Drusus holds a census.

The numbering needs care. W. A. McDevitte's English translation of 1850 numbers the summaries only to 140, and its book 136 matches book 138 in the Latin text. The status of every book is in lost-books-antiquity-livy-books.csv.

Ctesias of Cnidus was a court physician to the Persian king Artaxerxes Mnemon from 401 to 384 BC, as J. H. Freese notes in his 1920 translation of Photius. Ctesias wrote a Persica in 23 books, now lost.

Photius, patriarch of Constantinople in the 9th century, still read it. Codex 72 of his Bibliotheca opens, "Read the Persica of Ctesias of Cnidus in twenty-three books," and then sums it up (Freese). The Persica ended its story in 398 BC, by English Wikipedia's dates, so Photius read it more than 1,100 years later.

The Cypria was an epic in 11 books on the Trojan War up to the point where the Iliad begins, known chiefly from a plot summary by one Proclus in a handbook called the Chrestomathy. H. G. Evelyn-White's 1914 translation names him Eutychius Proclus of Sicca, while English Wikipedia calls that link uncertain. A later reader did see the poem itself: Pausanias, who lived about AD 110–180, wrote, "I know, because I have read it in the epic Cypria" (Evelyn-White's translation). In the 12th century, by English Wikipedia's dating, John Tzetzes still named its author: "Stasinus composed the Cypria" (Evelyn-White).

Known from quotations and scraps

Sappho's poems filled at least eight books in their ancient Alexandrian edition, English Wikipedia says. H. T. Wharton, writing before his death in 1895, says only that nine books "are said to have existed." Only one poem, the Ode to Aphrodite, is certainly complete. The whole of it survives only because Dionysius of Halicarnassus quoted it.

Dionysius quoted it at Rome as a model of smooth, finished style, about 25 BC by Wharton's date or between 20 and 10 BC by the conjecture in W. Rhys Roberts's 1910 edition. Papyrus adds more. P.Oxy. 7, a 3rd-century AD copy found at Oxyrhynchus in Egypt, appeared in print in 1898 with part of another poem.

Euripides first staged his Andromeda in 412 BC, and Aristophanes mocked it in the Thesmophoriazusae a year later, as English Wikipedia summarizes. More than five centuries on, in the 2nd century AD, Lucian retold a story of a fever at Abdera whose victims went about reciting the play. He quotes the start of Perseus's speech: "O Love, who lord'st it over Gods and men" (H. W. and F. G. Fowler's translation, 1905). The myth behind it, with Andromeda chained to a rock, is traced in the Cassiopeia entry.

Cicero's Hortensius (45 BC), a dialogue urging readers toward philosophy, survives only in fragments that English Wikipedia finds in five later authors. Among them is Augustine. His Confessions (AD 397–400) say he came on the book in his nineteenth year: "This book of his contains an exhortation to philosophy, and is called 'Hortensius'" (E. B. Pusey's translation).

The Chinese Sanwu liji (三五曆紀), credited to Xu Zheng of the 3rd century AD, is lost, and Chinese Wikipedia says only passages survive, quoted in later encyclopedias such as the Yiwen leiju, finished in 624. Its passage on the world's beginning opens "天地混沌如雞子．盤古生其中", which reads "Heaven and earth were one formless mass like a hen's egg, and Pangu was born inside it" (Translation: Where Tales Began). A note in the Wikisource text used here says this chapter is missing from the Song printing and was filled in from a Ming one.

Known from a list

Lists keep titles even when the books are gone. Diogenes Laertius, who wrote in the first half of the 3rd century AD, gives a catalogue of Aristotle's works that opens with On Justice, On Poets and On Philosophy. It closes, "In all 445,270 lines" (R. D. Hicks's 1925 translation, Perseus Digital Library).

Elsewhere Diogenes calls On Philosophy a dialogue. English Wikipedia's article on the Aristotelian corpus says the dialogues survive only in fragments.

A book list can itself be lost. Callimachus's Pinakes ("Tables"), from about 245 BC, listed the authors held at the Library of Alexandria in 120 books, as English Wikipedia summarizes it. Its form shows through later citations.

Athenaeus, writing about AD 200, says Callimachus listed books on the art of making cheesecakes. In a list of laws, he adds, Callimachus "quoted the first words" of a code of rules for banquets. In C. D. Yonge's 1854 translation, that quotation ends, "it is written in three hundred and twenty-three verses."

A list can show that a book existed without saving a line of it. The same question, what a record of books can prove once the books are gone, comes up at a royal library of the 7th century BC at Nineveh. It is followed in the Library of Ashurbanipal entry.

Known by little more than a title

Aristotle's Poetics promises more than survives. Chapter 6 says, "Of the poetry which imitates in hexameter verse, and of Comedy, we will speak hereafter" (S. H. Butcher's translation). Yet the part that survives covers tragedy and epic, as English Wikipedia notes.

The Rhetoric twice points readers to a discussion of jokes in the Poetics, and J. H. Freese's 1926 note on the second passage says, "The chapters are lost." Diogenes Laertius lists "A Treatise on the Art of Poetry, two books."

The Margites was a comic poem about a fool. Aristotle, who credits it to Homer in Poetics chapter 4, writes that it "bears the same relation to Comedy that the Iliad and Odyssey do to Tragedy" (Butcher).

Evelyn-White's 1914 collection prints five fragments, only a handful of lines, among them "He knew many things but knew all badly." The Suda, a 10th-century encyclopedia, gives the poem to Pigres of Halicarnassus instead.

Claudius, emperor from AD 41 to 54, wrote in Greek a history of the Etruscans in 20 books and one of Carthage in 8. The source is Suetonius's Life of Claudius, written around 121.

Suetonius also says that a second museum at Alexandria, named after Claudius, was added to the old one. On set days each year, the Etruscan history was to be read through in one museum and the Carthaginian in the other, with readers taking turns. English Wikipedia's article on Claudius says none of his works survived.

Last seen

The timeline below sets each work's date of writing beside its latest dated witness in the sources checked. The dates are in lost-books-antiquity-witnesses.csv. Read each latest witness as a floor, not a date of loss, since a later copy or quotation may exist outside these sources.

This entry assigns no cause of loss to any of the 12 works; it only marks where each trail goes quiet.

Those latest witnesses run from the 2nd century AD to the 12th. Suetonius named Claudius's histories around 121, and Lucian quoted the Andromeda later in the same century. Augustine wrote about the Hortensius around 400, Photius read the Persica in the 9th century, and John Tzetzes named the author of the Cypria in the 12th.

The last ancient copies of Sappho that survive are parchment pages of the 6th and 7th centuries. By the 12th century, English Wikipedia reports, John Tzetzes wrote that time had destroyed Sappho and her works.

What became of the Library of Alexandria is a separate question. This entry leaves it aside. The table shows something narrower: the latest witnesses for the Persica, the Margites and the Cypria are Byzantine writings of the 9th, 10th and 12th centuries.

Some lost text has come back. A set of papyri published in 2014 holds parts of ten consecutive poems by Sappho, according to English Wikipedia. The Homeric Hymn to Demeter survives in only one manuscript, found at Moscow, as Evelyn-White notes. Its story is followed in the entry on Persephone's pomegranate.

One open case is a set of more than 1,800 numbered papyri from the Villa of the Papyri at Herculaneum. English Wikipedia (read October 2026) says ways to read them are still being worked out; see the Herculaneum scrolls.