The present article is a very brief summary of a monograph I have completed which consists of an analysis of the way ancient manuscripts (rolls and codices) containing hexametric poetry mark book-ends.[1] I have addressed two main questions: 1) if and how these manuscripts mark the end of books and how end-marks change over time, especially with the adoption of the new format of the codex; and 2) how epic poems in rolls and codices were arranged, in particular whether, after the end of one epic book, another book followed and, if so, where it was placed, i.e. in the same column (or page) or in the next one.

    In this paper I will only provide a general overview of the problems presented by such research, of the methodology I followed, and of the results. The reader is referred to the original monograph[2] for the complete analysis, the discussion of the most relevant cases, and the description of the manuscripts I have analyzed.

    Collecting the Data: Papyri Analyzed

    In this work I focus on ancient manuscripts (both rolls and codices) containing hexametric poetry and showing the end of a book. A total of fifty-five manuscripts, ranging from the 3rd century BC to the 6th century AD, have been collected and studied. This sample includes mostly Homeric manuscripts, with forty papyri containing the Iliad and eleven containing the Odyssey. In addition to these, there are also two manuscripts of Hesiod (one containing the Theogony and the other containing Theogony, Works and Days and Shield), one of Eratosthenes' Hermes, and one of Oppian's Halieutica. Most of these fragments come from rolls (forty-four in total) and they range in time up through the 4th century AD; eleven are codices, from the 3rd– 4th to the 6th century AD. A complete list of all fifty-five manuscripts is offered in the table at the end of this paper.

    Evidence Available, Problems and Methodology

    The first problem has been to develop a methodology to assess the diverse and problematic set of evidence collected. In most cases, only fragments of ancient manuscripts are available, so that it is impossible to reconstruct completely the original layout of an ancient book; in the worst case, papyri containing the end of a book must be discarded because they are so damaged that the evidence they offer is inconclusive.[3] In some cases we cannot determine whether a title is placed at the end of a book, and hence is relevant to our study, or at the beginning (and hence is not).[4] Four titles[5] are completely detached from the rest of the text, thus they can only provide evidence for the general "shape" of the title, but not for its position within the roll.


    In many cases margins are missing so it is not possible to determine whether marks, such as coronis and paragraphos, were present. Thirty-one out of a total of the fifty-five manuscripts analyzed have missing margins. In twenty-four of them both coronis and paragraphos cannot be studied because all or most of the left margin has been lost.[6] In the other five cases only the coronis is visible, but the paragraphos, if there was one, is in lacuna.[7] In two cases the paragraphos is visible, but the coronis is in lacuna.[8] In some cases missing lower margins and a missing right-hand side of a column make it difficult to determine the presence of end-titles.[9]

    Book layout

    To understand the internal organization of a papyrus roll, one must determine whether there is any sign of another book following the book-end and, if so, whether the second book is placed either under the end title in the same column, or in the next column. Here the fragmentary status of the evidence available becomes a great problem: small scraps of papyrus offer little help to understand what came next. Even when part of the blank column underneath the end title is visible, it is often impossible to say whether a book followed in the next column, if the next column is lost. As a general rule, one can conclude that a book in a roll is not followed by another one, if the next column (or a considerable part of it) after the end of the book is preserved and is blank. All the other cases where the next column after the end of the book is not preserved must be counted as "uncertain." Twenty-five out of fifty-five cases fall in this category.

    However, even in the uncertain cases, when the fragments are large enough it is still possible to infer something about the position of a possible following book:

    1) The following book cannot possibly be in the same column, if the height of the column is known and the fragment is blank from the end of the book until the end of the column and the column after the end of the book is missing.

    2) The following book cannot possibly be in the next column, if only the first verse of the next book is visible and no following column is preserved. This is a peculiar case, which happens when, in a fragment, the last lines of a book are followed by only the first line of the next book. If the fragment breaks off and nothing else is preserved, it is not possible to determine whether this is a case of a versus reclamans,[10] or whether the end of one book is simply followed by the beginning of the next one without any major interlinear space. Among the manuscripts I studied, this happens in at least five cases.[11] All these cases are uncertain in terms both of the presence of the versus reclamans and of a following book. However, if that verse is not a versus reclamans, its presence indicates that the book could not have started in a different column as the book that ended.

    Last but not least, one must be aware that papyrus rolls tend to be preserved towards the end more than at the beginning; this means that the end of the books placed at the end of the roll are more likely to survive than the end of the books placed closer to the beginning of the rolls.


    Given all the problems presented by the data, only the unambiguous cases have been used to carry out the analysis and to draw conclusions. The uncertain cases have been used to counterproof the results reached on the basis of the unambiguous data.

    In order to provide a systematic analysis of all possibilities and to deal with the problems described above, all the manuscripts have been analyzed in search of the following elements:

    • 1. Versus Reclamans.
    • 2. Paragraphos.
    • 3. Coronis.
    • 4. End-title
    • 5. Presence of a following book.

    Each papyrus has been classified for each of these five elements either as "yes," "no," or "uncertain." I also considered one last element:

    • 6. Position of the following book.

    For this element, four different results are possible: "in the same column," "in a different column," "not in the same column," "not in a different column."


    The present analysis allows us to identify some clear patterns and to understand the way ancient manuscripts containing epic and above all Homeric poetry marked the end of books and organized their content over a period of time ranging from the 3rd century BC to the 6th century AD.

    Paragraphos & Coronis in Rolls

    The paragraphos was one of the signs used to mark the end of a book and it consisted of a line at the left margin stretching under the first letters of the last line of the book. This line was used to separate visually the last line of a book from the first line of the next book.[12] The distinction between paragraphos and forked paragraphos (or diple obelismene) in hexametric texts is a chronological one: the simple paragraphos of the earliest rolls[13] evolved into a forked paragraphos from the end of the 1st century BC, and from the 1st century AD onwards the forked paragraphos becomes the norm with no exception in the data collected.

    The coronis is, on the contrary, an entirely different sign.[14] When first used[15] it was a more mimetic sign, resembling a bird. In its standard form, the coronis has a sinuous shape similar to § crossed by horizontal strokes of different length, but a large number of variations can be found throughout our sample. Sometimes one of these horizontal strokes is particularly long and forked, and is identical to a paragraphos. When this happens, the coronis can be considered to be combined with a paragraphos. In our sample, the coronis is often strictly connected with a paragraphos which crosses, or is attached to, the coronis and reaches underneath the first letters of the lines.[16]

    The function of the paragraphos and the coronis is identical, but they must be distinguished, even though they might share a common origin.[17] The paragraphos is more rare, and it seems to have been used to mark book-ends in epic poetry before the coronis. Of all the manuscripts that present paragraphoi (twenty, both rolls and codices), only two have a paragraphos but not a coronis and they are Ptolemaic or very early Roman papyri.[18] This evidence, together with the fact that among the nine Ptolemaic and very early Roman papyri only two[19] present a coronis together with a paragraphos, suggests that the paragraphos was the first sign used to mark book-ends. Around the first century BC, the coronis started to be used in place of, or together with, the paragraphos and later it became the standard way to mark the end of an epic book.

    End-titles in Rolls

    End-titles first appear in the Roman period. The first known end-titles are found in P.Oxy. XLII 3000 (#10) and PSI inv. 1914 (#11), dated between the 1st century BC and the 1st century AD. After then, the normal way of marking a book-end in Roman rolls is with an end-title and marginal signs, normally a coronis, often combined with a paragraphos. Among the manuscripts that show unambiguous data, there are eleven rolls where the end of the book is marked by a paragraphos and/or a coronis, and by an end-title.[20] The end-title therefore seems to have been a constant presence in the Roman period; this is also confirmed by the "uncertain" evidence: eleven papyri, dating from the end of the 1st century BC to the 4th century AD,[21] have traces of an end-title but have lost the left margin, making it impossible to ascertain whether a marginal sign was present. There are also two cases[22] which have an end-title but no coronis or paragraphos. There is no case of a manuscript without end-title from the 1st century AD onwards (out of a total of forty-four rolls analyzed) and only two cases where the presence of an end-title is uncertain because of missing margins.[23]

    In rolls with Homeric poetry, end-titles are normally written in two lines: the first line has the name of the poem in the genitive (ΙΛΙΑΔΟΣ or ΟΔΥΣΣΕΙΑΣ), and the second has the letter corresponding to the book which has come to an end. There are, however, cases where the name of the poem and the letter of the book are in the same line.[24] The genitive of the title is a short-hand for the full formula: ΤΕΛΟΣ ΕΧΕΙ ΙΛΙΑΔΟΣ (or ΟΔΥΣΣΕΙΑΣ) followed by the book number indicated by the letter of the Ionic alphabet.[25]

    PSI inv. 1914 (#11) is the only case where the name of the episode's title seems to be present along with the end-title. This fragment contains Iliad 5.905–6.2, and after Iliad 6.1–2, which serve as reclamantes, it preserves half of the end-title and of half of what is likely to be the title of the Iliadic episode: [ΙΛΙ]ΑΔΟΣ | [Ε ] ̣ς | [Διο]μ̣η̣δους | [αριστ]ε̣ι̣α̣. Since this papyrus is quite early (it has been dated between the 1st century BC and the 1st century AD), it is possible that it still preserves the system, commonly used in classical times, of designating Homeric text by episode's titles.[26]

    The end-titles of other authors are different, because we have the name of the author in the genitive followed by the nominative of the title: for example ΗΣΙΟΔΟΥ | ΘΕΟΓΟΝΙΑ in PSI 11.1191 (#36).[27] The reverse order is attested in P.Oxy. XLII 3000 (#10), which has ΕΡΜΗΣ | ΕΡΑΤΟΣΘΕΝΟ[ΥΣ].

    Titles are normally written in a larger size, and placed underneath the last line of the book, after some major interlinear space. However, the dimensions of this interlinear space, as well as the size of the script, are unpredictable. There is even one example of title placed in a fresh column, in P.Lond.Lit. 6 (#13).

    Ptolemaic and Very Early Roman Papyri

    The papyri in our data set prove that the common claim that Ptolemaic papyri used to have the entire poems written continuously in one, very long roll[28] without distinguishing one book from another is false. Out of nine papyri from the Ptolemaic and very early Roman ages in our sample (3rd century–1st century BC), only one, P.Gen. inv. 90 (#2), does not mark the transition from one book to the other. Even though none of them has an end-title, four Ptolemaic and very early Roman papyri have at least one marginal sign marking the transition from one book to the other. They mark the end of a book with either a paragraphos or a coronis, or with both, placed on the left margin of the text, and this appears to be the most common method during the Hellenistic period. In particular, P.Sorb. inv. 2245A (#4) and P.Mert. II 52 (#9) have only a paragraphos, whereas P.Berol. inv. 16985 (#5) and P.Mil.Vogl. II 36 (#6) combine the paragraphos with a coronis. The paragraphos seems thus the most ancient way to indicate the end of a book. The other remaining four Ptolemaic papyri[29] are uncertain because the left margin is missing.

    To conclude, four papyri against one unambiguously show a separation between Homeric books. On the contrary, in none of the Ptolemaic papyri are end-titles present. [30]


    In addition to the use of marginal signs such as the paragraphos and/or the coronides, there is also the so-called versus reclamans. Among Ptolemaic and very early Roman papyri, only P.Berol. inv. 16985 (#5) and P.Mert. II 52 (#9), dated to the 1st century BC, offer unambiguous evidence of reclamantes. The versus reclamans in papyri without end-title can be a way of recognizing the end of the book; however, the reclamans is also used in later papyri with an end-title.[31] In these cases, the function of the reclamans is probably different. The most likely hypothesis is that reclamantes were used as signs to identify the relative position of different rolls. This means that the versus reclamans occurs only at the end of a roll, and this certainly happens in P.Berol. inv. 16985 (#5), P.Lond.Lit. 11 (#14), and most likely in P.Mert. II 52 (#9).

    The use of the reclamans decreased with time, and from the 2nd century AD onwards none can be found. This is probably due to the growing use of end-titles together with beginning-titles which, if used together in rolls, were more effective than a versus reclamans at indicating the relative order of different rolls, because it is easier to understand which book you are reading from the title than from the quotation of the first line.

    Paragraphos & Coronis in Codices

    Codices have coronis either with or without paragraphos, but there are no examples of paragraphos without coronis. However, there is one codex[32] which certainly does not have any marginal sign in the fully preserved left margin. Codices in particular tend to change the original shape of the coronis: the Harris Homer Codex (#42) and P.Ryl. I 53 (#47) display quite traditional coronides; the Morgan Homer (#43) and P.Vind. inv. 19815 (#51) instead present a more complex pattern. Also, the position of the coronis within the page changes. They start to be used in larger numbers and at different positions: around the titles in couples, or at the center of the page underneath the title as in Morgan Homer (#43), or placed as a frame to the title as in P.Vind. inv. 19815 (#51).

    End-titles in Codices

    All the codices available to us have an end-title. This is normally in the full form of the genitive of the name of the poem, followed by the letter designative of the book which has reached the end. In only two codices is the end-title uncertain because parts of the page are missing,[33] but there is none where the end-title is certainly absent. This is due to the fact that by the time of the appearance of the codex (3rd cent AD), the use of end-titles was well established. In some codices the end-titles are combined with beginning-titles,[34] and they are often written in a much reduced form: only the letter to indicate the book, without any genitive of the work.[35] In particular, in P.Ryl. I 53 (#47), the full title (ΟΔΥΣΣΕΙΑΣ Ω) is to be found only at the end of the codex, when the poem reaches its end, whereas in the other instances the codex uses only the capital of the letter corresponding to the book number.

    Books Organization in Rolls and Codices

    It is difficult to determine how many books of epic poetry an average roll could contain. The fact that many rolls taken into account (twenty-seven out of forty-four) show remnants of only one book is not conclusive, since these might be only small portions of the original roll. Only three papyri of our sample certainly contained only one book: P.Lond.Lit. 25 (#16), P.Bodm. I 1 (#44) and P.Bodm. I 2 (#45). On the other hand, at least six papyri (three Ptolemaic papyri[36] and three Roman papyri[37]) of the sample show remnants of at least two books, because the end of a book is followed by the beginning of the next in the same fragment. This evidence, together with the fact that rolls were long enough to contain an entire tragedy (ca. 1200 to 1700 lines), leads to the conclusion that rolls, also in the Roman period, could contain more than one book, although there were editions also with only one epic book per roll.[38]

    The analysis of the book number of books certainly placed at the end of a roll has shown that most rolls reach an end with an even number of epic book. This suggest that the most common way to organize an epic poem like the Iliad or the Odyssey in a roll was to have two books per roll, for a total of twelve rolls for a complete edition of the Iliad or Odyssey. This is consistent with the average length of a roll, which could contain a drama and thus have an average number of ca. 1000–1500 stichoi. Editions with three books per roll, hence in eight volumes, might have also been used, especially for the Odyssey, which has, on average, shorter books. As for the position of the books within the roll, it seems that epic books were put one following the other in the Ptolemaic period, and then, in Roman times, rolls began to have the new book in a new column, but also the older Ptolemaic system of having one book after the other in the same column was still used.

    Codices always have more than one book; in our sample nine cases out of eleven show at least two books.[39] Codices could contain even the entire epic poem.[40] In terms of layout, of the nine codices that have a following book, six place it on the same page of the previous one. This evidence suggests that codices tend to have one book following the end of the previous one, and thus that they tend to occupy all the available space.

    Manuscripts Considered
    # Papyrus Name MP3 Date Format Material Content
    1 P.Grenf. II 4 + P.Hib. I 22 979 3rd BC R P Iliad 21–23
    2 P.Gen. inv. 90 890 2nd half 3rd BC R P Iliad 11–12
    3 P.Lefort 1 1145 2nd half 3rd BC R P Odyssey 21–22.1
    4 P.Sorb. inv. 2245 1081 2nd half 3rd BC R P Odyssey 9–10
    5 P.Berol. inv. 16985 980 1st BC R P Iliad 21–22 + 23.1–2
    6 P.Mil.Vogl. II 36 815 1st BC R P Iliad 7, 8.1
    7 P.Mil.Vogl. II 37 1103 1st BC R P Odyssey 11, 12.1
    8 P.Oslo III 68 705 1st BC R P Iliad 3, 4.1
    9 P.Mert. II 52 1033 2nd half 1st BC R P Odyssey 2, 3.1
    10 P.Oxy. XLII 3000 364.2 1st BC – 1st AD R P Eratosthenes, Hermes
    11 PSI inv. 1914 769.11 1st BC – 1st AD R P Iliad 5, 6.1–2
    12 P.Lond.Lit. 27 998 1st half 1st AD R P Iliad 23–24
    13 P.Lond.Lit. 6 + P.Ryl. III 540 643 1st AD R P Iliad 2
    14 P.Lond.Lit. 11 697 1st AD R P Iliad 3–4, 5.1
    15 P.Lond.Lit. 22 899 1st AD R P Iliad 13–14
    16 P.Lond.Lit. 25 ("Harris Homer") 953 1st AD R P Iliad 18
    17 P.Lond.Lit. 30 1039 1st AD R P Odyssey 3
    18 P.Hamb. II 157 702 1st – 2nd AD R P Iliad 3
    19 P.Louvre inv. AF 12809 571 1st –2nd AD R P Iliad 1
    20 P.Paris 1.3 ter 772 1st –2nd AD R P Iliad 6
    21 PSI Od. 5 1052.2 1st –2nd AD R P Odyssey 4, 5.1
    22 P.Harr. I 120 892 2nd AD R P Iliad 12
    23 P.Köln IV 182 867.1 2nd AD R P Iliad 10
    24 P.Lond.Lit. 24 952 2nd AD R P Iliad 18
    25 P.Lond.Lit. 28 ("Bankes Homer") 1013 2nd AD R P Iliad 24
    26 P.Ross.Georg. 1.5 1057 2nd AD R P Odyssey 4
    27 PSI XII 1275 1011 2nd AD R P Iliad 23
    28 Hawara Homer 616 2nd half 2nd AD R P Iliad 1–2
    29 P.Lond.Lit. 8 676 2nd –3rd AD R P Iliad 2
    30 P.Mich. inv. 2 953.1 2nd –3rd AD R P Iliad 18
    31 P.Oxy. III 445 = P.Lond.Lit. 14 778 2nd –3rd AD R P Iliad 6
    32 P.Oxy. III 563 1028 2nd –3rd AD R P Odyssey 1
    33 P.Oxy. IV 771 929 2nd –3rd AD R P Iliad 15
    34 PSI XI 1185 795 2nd –3rd AD R P Iliad 6
    35 PSI XI 1188 852.02 2nd –3rd AD R P Iliad 10
    36 PSI XI 1191 493.2 2nd –3rd AD R P Hesiod, Theogony
    37 P.Oxy. inv. 19 2B. 79 / C (1–2) a NA 1st half 3rd AD R? P Oppian Halieutica 4
    38 P.Cair. inv. 3675 601 3rd AD R P Iliad 1
    39 P.Mich. inv. 5760d 1113.1 3rd AD R P Odyssey 14–15
    40 P.Mil.Vogl. inv. 1225 686.1 3rd AD C P Iliad 3– 4
    41 P.Ross.Georg. I 4 941 3rd AD R P Iliad 17
    42 P.Lond.Lit. 5 ("Harris Homer Codex") 634 2nd half 3rd AD C P Iliad 2–4
    43 P.Amh. inv. G 202 ("Morgan Homer") 870 3rd –4th AD C P Iliad 11–16
    44 P.Bodm. 1.1 736 3rd –4th AD R P Iliad 5
    45 P.Bodm. I 2 736 3rd –4th AD R P Iliad 6
    46 P.Köln I 40 1033.3 3rd –4th AD R P Odyssey 3–4
    47 P.Ryl. I 53 1106 3rd –4th AD C V Odyssey 12–15, 18–24
    48 P.Amh. II 159 985 4th AD C P Iliad 21–22
    49 P.Stras. inv. Gr. 2675 789 4th AD C P Iliad 6, 9–14
    50 PSI inv. 1210 621.1 4th AD R P Iliad 1
    51 P.Vind. inv. 19815 499 2nd half 4th AD C P Hes., Th., WD, Shield
    52 Bodl.Libr. MS. Gr. class. g. 49(P) 961 4th –5th AD C P Iliad 18
    53 BL Add. MS 17210 ("Cureton Homer") 897.1 5th –6th AD C V Iliad 12–16, 18–24
    54 Ilias Ambrosiana 554.1 5th –6th AD C V Iliad 1–2, 4–17, 21–24
    55 P.Oxy. XV 1817 948 6th AD C P Iliad 17–18


      1. A previous study on book conventions for papyri containing Homeric text was carried out by Lameere 1960. On book rolls see also Johnson 2004. return to text

      2. Τὸ μέγα βιβλίον: Book-Ends, End-Titles, Coronides in Papyri with Hexametric Poetry. Am.Stud.Pap. XLVIII (Durham, NC 2010).return to text

      3. This happens with P.Köln I 40 (#48), a codex. Precisely where one book ends and the following one begins both papyri are missing, so that it is impossible to verify whether they had a title and/or other signs to mark the book-end. If these manuscripts do not offer any firm evidence as far as the layout of the end of the book is concerned, they are, however, to be taken into account in the analysis of the number of books in rolls and codices since they contain at least two books.return to text

      4. Examples of such cases are P.Paris 3 (#20) and P.Harr. I 120 (#22). On beginning titles see now Caroli 2007.return to text

      5. P.Lond.Lit. 31, P.Oxy. LXVIII 4663, P.Oxy. XI 1399, and P.Achm. 3 i. These papyri are not included in the main database but collected in a separate Appendix in the monograph.return to text

      6. They are: P.Grenf. II 4 + P.Hib. I 22 (#1); P.Lefort 1 (#3); P.Mil.Vogl. II 37 (#7); P.Oslo III 68 (#8); PSI inv. 1914 (#11); P.Lond.Lit. 27 (#12); P.Lond.Lit. 6 (#13); P.Lond.Lit. 30 (#17); P.Paris 3 (#20); P.Harr. I 120 (#22); P.Köln IV 182 (#23); P.Lond.Lit. 24 (#24); P.Ross.Georg. I 5 (#26); PSI XII 1275 (#27); P.Lond.Lit. 8 (#29); P.Oxy. III 563 (#32); PSI XI 1191 (#36); P.Mil.Vogl. inv. 1225 (#40); P.Köln I 40 (#46); P.Amh. II 159 (#48); PSI inv. 1210 (#50); Bodleian Libr. Gr. clas. g. 49(P) (#52); Cureton Homer (#53), and Ilias Ambrosiana (#54). return to text

      7. They are: P.Louvre inv. AF 12809 (#19); P.Mich. inv. 5760d (#39); P.Bodm. I 1 (#44); P.Vind. inv. 19815 (#51) and P.Oxy. XV 1817 (#55).return to text

      8. They are: P.Lit.Lond. 22 (#15) and P.Oxy. III 445 (#31).return to text

      9. As in PSI XII 1275 (#27), P.Oxy. inv. 19 2B. 79 / C (1–2) a (#37), and P.Stras. inv. gr. 2675 (#49). In PSI XI 1185 (#34) only half of the column is preserved (the left hand side), and therefore it is still possible that a colophon was present in the missing part.return to text

      10. Versus reclamans: the first, or first two, lines of the following book which are put immediately after the end of a book. Cf. West 1963.return to text

      11. P.Hib. I 22 (#1), P.Lefort 1 (#3), P.Mil.Vogl. II 36 (#6), P.Mil.Vogl. II 37 (#7), and P.Oslo III 68 (#8).return to text

      12. On paragraphos, see Turner, GMAW, 8.12–13; Barbis Lupi 1994; Del Mastro 2001; Johnson 1994. On paragraphos in school exercises, see Cribiore 1996, 81–82. return to text

      13. P.Berol. inv. 16985 (#5), P.Mil.Vogl. II 36 (#6), and P.Oxy. XLII 3000 (#10).return to text

      14. On the coronis see Tanzi-Mira 1920; Stephen 1959; Turner, GMAW, 12 and n. 59. On coronides in school exercises, see Cribiore 1996, 83.return to text

      15. In the Timotheus papyrus, of the 4th century BC, and in a papyrus with Euripides' Phaethon (P.Berol. inv. 9771, BKT V.2, 79–84.return to text

      16. This happens in P.Lit.Lond. XXV (#16), P.Hamb. II 157 (#18), the Hawara Homer (#28), P.Mich. inv. 2 (#30), PSI XI 1188 (#35), P.Oxy. inv. 19 2B. 79 / C (1–2) a (#37), P.Cair. inv. 3675 (#38), Harris Homer Codex (#42) and P.Ryl. I 53 (#47) in folio 78 recto.return to text

      17. Cf. Stephen 1959, 4.return to text

      18. P.Sorb. inv. 2245A (#4) and P.Mert. II 52 (#9).return to text

      19. P.Berol. inv. 16985 (#5) and P.Mil.Vogl. II 36 (#6).return to text

      20. P.Oxy. XLII 3000 (#10); P.Lond.Lit. 11 (#14); P.Lond.Lit. 25 (#16); P.Hamb. II 157 (#18); PSI Od. 5 (#21); Hawara Homer (#28); P.Mich. inv. 2 (#30); P.Oxy. IV 771 (#33); PSI XI 1188 (#35); P.Cair. inv. 3675 (#38); P.Bodm. I 2 (#45).return to text

      21. PSI inv. 1914 (#11); P.Lond.Lit. 27 (#12); P.Lond.Lit. 6 (#13); P.Lond.Lit. 30 (#17); P.Köln IV 182 (#23); P.Lond.Lit. 24 (#24); P.Ross.Georg. I 5 (#26); P.Lond.Lit. 8 (#29); P.Oxy. III 563 (#32); PSI XI 1191 (#36); PSI inv. 1210 (#50). return to text

      22. P.Lond.Lit. 28 (#25) and P.Ross.Georg. I 4 (#41).return to text

      23. In PSI XI 1185 (#34) and P.Oxy. inv. 19 2B. 79 / C (1–2) a (#37).return to text

      24. P.Bodm. I 2 (#45).return to text

      25. This formula is found only in one manuscript, a codex, the Harris Homer Codex (#42) which has ΤΕΛΟΣ ΕΧΕΙ | ΙΛΙΑΔΟΣ [Β] at the end of Iliad 2 and ΤΕΛΟΣ ΕΧΕΙ | ΙΛΙΑΔΟΣ | Γ at the end of Iliad 3.return to text

      26. Cf. Hdt. 2.116. return to text

      27. Similar ΗΣΙΟΔΟ[Υ ΕΡΓΑ] ΚΑΙ | [ΗΜΕ]ΡΑΙ in P.Vind. inv. 19815 (#51).return to text

      28. The "single monster roll" as defined by West 1967, 20. On these big rolls see also Lameere 1960, 129, and Birt 1882, 439, 445.return to text

      29. P.Grenf. II 4 + P.Hib. I 22 (#1), P.Lefort 1 (#3), P.Mil.Vogl. II 37 (#7), and P.Oslo III 68 (#8).return to text

      30. West 1967, 20–25, reached similar conclusions. return to text

      31. PSI inv. 1914 (#11), 1st cent. BC–1st cent. AD, P.Lond.Lit. 11 (#14), 1st cent. AD, and PSI Od. 5 (#21), 1st–2nd century AD.return to text

      32. P.Stras. inv. gr. 2675 (#49).return to text

      33. P.Amh. II 159 (#48) and P.Stras. inv. gr. 2675 (#49).return to text

      34. P.Mil.Vogl. inv. 1225 (#40), the Morgan Homer (#43), P.Ryl. I 53 (#47), P.Vind. inv. 19815 (#51), and the Cureton Homer (#53).return to text

      35. This reduced form of the title is found in only one roll, P.Louvre inv. AF 12809 (#19).return to text

      36. P.Gen. inv. 90 (#2), P.Sorb. inv. 2245A (#4), and P.Berol. inv. 16985 (#5).return to text

      37. P.Lond.Lit. 27 (#12), P.Mich. inv. 5760d (#39) and P.Köln I 40 (#46).return to text

      38. This conclusion demonstrates that the common assumption that Roman rolls contained only one epic book (as claimed, for example, by Lameere 1951, 184–187; eund. 1960, 11; 39; 131; 241–43; Martin 1954, 10–16; Van Rengen 1990, 207–208; Gallazzi 1996, 387, n. 1) is no longer tenable.return to text

      39. The other two codices, Bodleian Libr. Gr. clas. g. 49(P) (#52) and P.Oxy. XV 1817 (#55), are not meaningful since they are only small fragments.return to text

      40. As happens in P.Ryl. I 53 (#47).return to text

      Works Cited

      Barbis Lupi 1994    R. Barbis Lupi, "La paragraphos: analisi di un segno di lettura," in A. Bülow-Jacobsen (ed.), Proceedings of the 20th International Congress of Papyrologists (Copenhagen 1994) 414–417.
      Birt 1882    Th. Birt, Das antike Buchwesen (Berlin 1882).
      Caroli 2007    M. Caroli, Il titolo iniziale nel rotolo librario greco-egizio (Bari 2007).
      Cribiore 1996    R. Cribiore, Writing, Teachers, and Students in Graeco-Roman Egypt. Am.Stud.Pap. XXXVI (Atlanta 1996).
      Del Mastro 2001    G. Del Mastro, "La paragraphos nei PHerc. 1425 e 1538," CronErc 31 (2001) 107–131.
      Gallazzi 1996    C. Gallazzi, "Un rotolo con Iliade Σ diviso tra Londra e Parigi (P. Lond. Inv. 127 A + P. Paris 3 Bis)," RFIC 124 (1996) 385–407.
      Johnson 1994    W.A. Johnson, "The Function of the Paragraphos in Greek Literary Prose Texts," ZPE 100 (1994) 65–68.
      Johnson 2004    W.A. Johnson, Bookrolls and Scribes in Oxyrhynchus (Toronto-Buffalo-London 2004).
      Lameere 1951    W. Lameere, "Pour un recueil de fac-similés des principaux papyrus de l'Iliade et de l'Odyssée," Scriptorium 5 (1951) 177–194.
      Lameere 1960    W. Lameere, Aperçus de paléographie Homérique; à propos des papyrus de l'Iliade et de l'Odyssée des collections de Gand, de Bruxelles et de Louvain (Brussels-Antwerp 1960).
      Martin 1954    V. Martin, Papyrus Bodmer I, Iliade, chants 5 et 6 (Cologne-Geneva 1954).
      Stephen 1959    G.M. Stephen, "The Coronis," Scriptorium 13 (1959) 3–14.
      Tanzi-Mira 1920    G. Tanzi-Mira, "Paragraphoi ornate in papiri letterari greco-egizi," Aegyptus 1 (1920) 224–227.
      Turner, GMAW    E. G. Turner and P. J. Parsons, Greek Manuscripts of the Ancient World. BICS Suppl. 47 (London 19872).
      Van Rengen 1990    W. Van Rengen, "P.Cair. inv. 3675" in C. Saerens, R. De Smet, and H. Malaerts (eds.), Studia Varia Bruxellensia (Leuven 1990) II 207–215.
      West 1963    S. West, "Reclamantes in Greek Papyri," Scriptorium 17 (1963) 314–315.
      West 1967    S. West, The Ptolemaic Papyri of Homer. Pap.Col. III (Cologne 1967).