1. Introduction
In the third century ce, Mani and his followers expanded his new revelation throughout Mesopotamia, westwards through Egypt and the Roman Empire, and eastwards through the Sasanian Empire and beyond. Mani boasted of his own writings as an advantage to the fledgling community, claiming to have written down his own revelations and encouraging his followers to write, copy, and translate his teachings.2 It is clear that written texts were an important part of the early Manichaean communities’ identity and practice, as well as a target of their opponents.3 More generally, modern scholars tend to think of Manichaeism as a proto-typical ‘religion of the book’ and have given due attention to notions of textuality, scripture, or books in Manichaean discourse and anti-Manichaean polemics.4
We know very little about Mani’s own scribal practices, or the book culture of the religiously diverse Sasanian Mesopotamian milieu in which he was raised and began his mission. Yet over the past 120 years a number of Manichaean writings in a dozen languages have been discovered at disparate sites ranging from North Africa to central China, thus surviving, in part, a history of persecution and book-burnings under various regimes.5 Despite these troves of manuscripts, which total in the thousands, from immense papyrus books to mere fragments, the codicology of Manichaean manuscripts is still relatively undeveloped. In fact, there is essentially only one major study of Manichaean codicology, Zsuzsanna Gulácsi’s analysis of the illuminated manuscript fragments from Turfan. Although her focus is narrow, as only about one-twentieth of the extant fragments from Turfan are illuminated, her careful delineation of principles provides a foundation for the further study of Manichaean codices, especially in Central Asia.6 We are not aware of any comparative study of Manichaean manuscripts from the Mediterranean world. Furthermore, Manichaean book culture has been largely neglected in the otherwise lively academic debates about the history of the book in Late Antiquity, which are mostly based on Christian and Buddhist evidence. Yet Manichaean books are some of the earliest surviving witnesses of the codex format. Likewise, the still-nascent field of comparative manuscript studies would benefit greatly from the broad geographical distribution provided by the Manichaean evidence.7
In this chapter, we seek to establish the foundations for a comparative study of Manichaean manuscripts which integrates them into the broader study of ancient and medieval textual cultures. In so doing, we argue that the textual culture of the Manichaeans cannot be studied in isolation from the textual cultures with which they interacted, but also, and equally importantly, that it is past time for studies of those cultures, such as the early Christian one, to take the Manichaeans into account. In the first section of this chapter, we provide a survey of the surviving Manichaean manuscript corpora, which, uniquely, are evenly represented between the Mediterranean and Central Asia. Our emphasis is not on their content, but rather two aspects of their materiality: first, structural aspects, such as format and writing support; second, paratextual elements employed within this structure. As we describe the various surviving Manichaean texts, we compare them with non-Manichaean manuscripts from the same languages and regions, to evaluate any salient differences or exceptional qualities. Having outlined the evidence, in the second section, we compare these Manichaean texts across Eurasia in an effort to identify features shared over various linguistic cultures and geographic regions. One of these comparative features is structural (quires), three are paratextual (text-division strategies, running headers, and indices); all four of these features are salient parts of the overall construction and organization of a book. An additional goal of this second section is to highlight the importance of paratextual aspects for comparative manuscriptology, and make initial steps towards sketching a general history of the appearance and usage of the three paratextual features under discussion. Finally, in the third section, we discuss how the Manichaean corpora contribute to the debate about the ‘rise of the codex’ vis-à-vis the scroll, looking at both the Mediterranean and Central Asian evidence. Our results suggest that some distinctive aspects of Manichaean book culture found at either end of Afro-Eurasia go back to the scribal practices of the early community in Mesopotamia. More generally, this chapter attempts to chart an approach to the comparison of manuscript cultures through focusing on a transregional community.
2. The Manichaean Corpora
In this section we give an overview of the extant Manichaean textual corpora, focusing on the codicological characteristics of the surviving manuscript remains. We begin with the Syriac evidence, which, though from Egypt, is perhaps closest to the book culture of Mani and his first disciples in Mesopotamia; we then proceed roughly chronologically, and from the Mediterranean towards Central and East Asia. That ‘western’ texts happen to be older than the ‘eastern’ ones is in large part based on the random circumstances of their preservation and contemporary discovery. However, the former are largely found in Egypt, where the climate was conducive to survival; in contrast, no Iranian manuscripts from the Late Antique Mesopotamian heartland of Manichaeism, nor from Sasanian Iran proper, have survived. The extant texts from Turfan and Dunhuang, including translations into Iranian languages, Old Uighur, Chinese, and others, come from areas that were missionized later, in the early medieval period.
2.1 Syriac Fragments
The Syriac language and manuscript tradition has a special connection to the textual culture of the earliest Manichaean communities. Mani spoke a dialect of Aramaic close to Syriac, and used an Aramaic script closely related to the Palmyrene script. This script variety came to be used by the Manichaeans to write not only Aramaic, but also Iranian languages including Middle Persian, Parthian, and Sogdian, and, in Turfan, Old Uighur as well—it has hence become known as the ‘Manichaean script’ and attributed to Mani himself, although there is no evidence that he actually invented it and the extent to which the Manichaean script was limited in use to Manichaean communities, at least in Late Antique Mesopotamia, is also unclear.8 In addition to Syriac authors such as Ephrem and Theodore bar Konai, who may preserve direct quotations of Manichaean texts in their original Syriac, there have survived a small number of extremely fragmentary manuscripts in Syriac (though there is not enough text preserved to assess dialectical variation) in Manichaean script from the Roman Empire. There are four groups of Manichaean Syriac manuscript fragments, all parchment. Nils Pedersen and John Moller Larsen, who have collected, re-edited, and thoroughly analyzed these, argues persuasively that they treat Manichaean topics. While their exact provenance is unknown, they were all apparently acquired from Egypt in the late 1800s and early 1900s.
These manuscripts are of great interest for the early history of Manichaean codicology, as they may have been composed in Egypt by Mesopotamian scribes, or brought there from the community’s Sasanian heartland.9 The Allberry Fragments (Cambridge University Library OR. 2252–2553), named after the first editor of the Manichaean Psalmbook, are fragmentary. According to their first editor, F.C. Burkitt, they are ‘from Middle Egypt, and appear to have been used to bind some ancient Coptic MSS.’10 The fact that they have writing on both sides suggests the codex format. The Ashmunain Fragment (BL OR. 6201 C(1)), reportedly from Middle Egypt as well, is also a parchment codex, apparently a miniature one.11 The nine Berlin fragments (Berlin P 22364) are rather small, and difficult to reconstruct. According to conservator Myriam Krutzsch, they could be either from a small codex (albeit one with three columns) or a roll.12 Pedersen notes that, based on their content, ‘it is almost certain that these fragments are Manichaean,’’ and therefore ‘it seems probable that they come from a codex.’13 There is so far no concrete reason to assume that the Manichaeans originally preferred the codex, especially in the earliest period of their missionary activity, as will be argued in the following section; but the fact that there is Syriac text on both sides, and apparently in the same hand, does suggest the codex format. The same cannot be said of the Oxyrhynchus fragments (Bodleian MSS. Syr. D.13(P) and D.14(P)), which have writing on both sides: the Syriac (on the recto) is apparently earlier, while the Greek (on the verso) is later.14 Although this feature of its codicology is not discussed by Pedersen, the writing of a different language and text on the blank verso of a sheet strongly suggests that the original format was the roll. Given that the Greek text is tentatively dated to the fourth century, the Syriac text may well go back to the earliest period of the Manichaean mission in Egypt. Finally, the Heidelberg Fragment (P.Heid.Syr. 1), has the same Syriac hand on both sides, and was thus likely a codex.15
In conclusion, all of the fragmentary Syriac Manichaean manuscripts are made of parchment; two are likely from codices, one is likely from a roll, and one might be from either a codex or a roll. This evidence is consistent with the hypothesis that the early Manichaeans used the roll format likely known to Mani from his Mesopotamian background. Some of these would have made their way to Egypt, perhaps in the context of early missionary activity there, where they were preserved with other late antique textual remains. By the fourth century, Manichaeans were also experimenting with the codex format, possibly in both Mesopotamia and Egypt.
2.2 Egypt
2.2.1 Kellis
The earliest securely dated Manichaean manuscripts are from Kellis (modern Ismant el-Kharab), an Upper Egyptian town in the Dakhleh Oasis, within the western desert.16 In Late Antiquity, Kellis had between 1,000 and 1,500 inhabitants, but was close to the larger oasis cities of Trimithis (modern Amheida) and Mothis (modern Mut el-Kharab), the nome capital. Excavations at Kellis in the 1980s and 1990s under the direction of Colin Hope revealed a number of manuscript remains in a private domestic context: Houses 1–3, for example, were the findspots for various writings in Coptic (in several dialects), Greek, and Syriac, many with specifically Manichaean themes, dating in large part to the second half of the fourth century. As a group, they attest to a robust community which apparently lived openly in this remote oasis town, without direct threat of persecution, as its members and associates, including a ‘teacher’ who might have been one of the Elect, traveled throughout Egypt.17
The evidence for Manichaean manuscripts at Kellis is striking in its diversity, not only in language but also in format and material. Alongside documents and private letters on papyrus, there are papyrus amulets, wooden tablets, and codices on both papyrus and wood. Indeed, because these remains were discovered in a known archaeological context, rather than an antiquities market, it is possible to appreciate how a single community’s patterns of manuscript usage could embrace a variety of forms and contexts; this fact has been largely ignored in studies of early Christian book history, which have focused on the most expensive manuscript productions, the codex and the roll, in developing a narrative of ‘triumph’ and ‘replacement’ rather than co-existence.18
The Kellis Manichaean evidence demonstrates the importance of copying and sharing texts within the community. Among the private letters associated with a large, multi-generational Manichaean family (including Makarios and his wife Maria), one, from Makarios to Maria, mentions that their child Piene is receiving instruction to be a reader (P.Kell.Copt. 25), and another that he is traveling to Alexandria with the ‘teacher,’ presumably a member of the Manichaean hierarchy (P.Kell.Copt. 29).19 In another letter (P.Kell.Copt. 19), Makarios encourages Matthaios, Piene’s older brother, to study a number of texts, including the Judgment of Peter the Apostle, the Great Prayers, the (Greek) Psalms, Sayings, and Prostrations, and further to ‘Write small selections now and then, more and more. Write a copy (typos) daily, because I need you to write some books here’ (P.Kell.Copt. 19.15–19, my translation). The letter to Matthaios suggests commonly known, authoritative titles of texts, of use for study and worship. Although none of them are specifically identifiable, another letter (P.Kell.Copt.120) from the same context, House 3, mentions the ‘Acts’ and the ‘Gospel,’ which might refer to the Acts of the Apostles (or the 5 apocryphal apostolic acts) and the Diatessaron, but more likely to Mani’s Living Gospel and narratives about the missionary efforts of his disciples.
The efforts of Piene and Matthaios in reading and writing reflect the strong literary culture of Manichaeans, even in a small oasis community such as Kellis, which is itself connected to the metropolis Alexandria through an in-group network in which the ‘great teacher’ must have played a central role. The surviving literary texts from Kellis confirm that Manichaean literary texts were kept there, in an intriguing variety of formats, lengths, and languages, including: two papyrus codices of Mani’s Letters, in Coptic (P.Kell.Copt. 53 and 54=11–83 and 84–93);20 an apocryphal papyrus fragment in Greek related to the Acts of John (P.Kell.Gr. 97=94–110); two biblical papyrus fragments in Coptic: Romans 2:6–29 (P.Kell.Copt. 6=94–97) and Hebrews 12:4–13 (P.Kell.Copt. 9=98–100); a Kephalaia-like composition (P.Kell.Copt. 8=94–77); two Syriac-Coptic bilingual glosses on wooden boards (P.Kell.Syr./Copt. 1=105–111 and P.Kell.Syr./Copt. 2=112–126); and a fragmentary page from a double-sided parchment codex in four columns, apparently two in Syriac and two in Greek (the content is too fragmentary to assess, but mention of ‘spirit,’ ‘pity,’ and ‘mercy’ suggests it could be Manichaean (P.Kell.Syr./Gr. 1=129–131).21
Although it is not certain whether the literary manuscripts from Kellis were actually copied there, the numerous texts on wooden boards, sometimes bound in codex form, were shorter and ephemeral, with substantial evidence for reuse; these were almost certainly produced in the oasis. These include school exercises in both Greek (P.Kell.Gr. 90) and Coptic (P.Kell.Copt. 10); another Kephalaia-like composition (T. Kell.Copt. 1), and, especially, various Psalm manuscripts, presumably for personal use. Of the twenty-one manuscripts with Manichaean Psalms at Kellis, most of them very fragmentary, five are on wooden tablets: P.Kell.Gr. 98; T.Kell.Copt. 2, 4, 6, 7.22 P.Kell.Gr. 98, containing the Prayer of the Emanations, a daily prayer for catechumens, is written on a single board, perhaps to aid in recitation, whether in a private or communal context. T.Kell.Copt. 2 is a wooden codex constructed of five boards, with only one containing legible text, in this case a psalm; as the editors note, this seems to have been ‘recycled’ from an earlier codex in the construction of a new one, as revealed by two sets of drilled holes.23 There is writing in two columns on both sides, consistent with the codex format. However, only the beginnings of Psalm verses are recorded; they are cut off at the end of the line, and the next line starts the new verse.24 T.Kell.Copt. 4 is a wooden board that was also originally part of a codex, with writing on both sides, containing psalm 222 from the Medinet Madi Psalmbook (part 2), and Psalm 109, from the unpublished Psalmbook part 1.25 In this case, verses are written out in full. T.Kell.Copt. 6 is another board with Psalms (this one has Psalm 261 from part 1 of the Psalmbook), but the text is fragmentary and it is not possible to reconstruct the layout of lines and verses fully.26 Three leaves and related fragments (P.Kell.Copt. 1–3) may have belonged to a codex (or codices), but their multiple, unpracticed hands have led the editor to hypothesize that they ‘belong together and are the remains of what might best be termed a “scrap-book” of Manichaean psalms.’27 Finally, two Greek hymns (P.Kell.Gr. 91 and 92) are written on both sides of a papyrus leaf with no signs of being part of a codex; the editors suggest that they were amulets.28
2.2.2 Medinet Madi
The Medinet Madi corpus of Manichaean codices consists of seven Coptic papyrus codices and several associated wooden covers, which appeared on the Cairo antiquities market in 1929–1930. They represent important works of Manichaean literature, attributed to Mani himself and his early followers: the Acts, related to early missionary journeys; the Epistles of Mani; Homilies related to his death and the subsequent persecutions; two volumes of Kephalaia, that is, doctrinal chapters, the first in Berlin, the second in Dublin; the Psalmbook; and the Synaxeis.29 These were purchased separately by Chester Beatty and Berlin institutions, and in the subsequent decades underwent a slow, tortuous path of conservation and study, interrupted by various events, including World War II. As a result, despite being of comparable importance to the Nag Hammadi Library, and ongoing intensive conservation efforts, they remain only partly edited.30 Given this situation, it is perhaps not surprising that there is no overview of the codicology or paleography of the corpus. This section thus can provide only a provisional assessment, to be expanded in a later study, since various codicological and paratextual elements cannot be assessed for the unpublished sections.
The page sizes for the seven Medinet Madi codices are difficult to determine in their present state, as many pages seem to have been damaged and diminished during the conservation process. According to the eminent papyrus conservator Hugo Ibscher, who did the initial work of separating and stabilizing the pages, six of the books had an original page size of 31.5 x 18 cm, with the slightly smaller Psalmbook measuring 27 x 17.5 cm.31 The conserved leaves are usually smaller, but the likely explanation is that papyrus fibres were lost as they were removed from the book blocks. Extant page and column size, and estimated original number of pages are given for the seven codices in Table 1 below.
| Text | Page (Column) Size | Total Original Pages |
|---|---|---|
Psalmbook (CBL Pma 3/4) |
27 x 17.5 cm (16.5–17.5 x 10–11 cm) |
~ 560 pages (estimated) |
Synaxeis (CBL Pma 5) |
30 x 19 cm (21.5 x 12–12.5 cm) |
> 500 pages |
Kephalaia 1 (P. Berol. 15998) |
31.5 x 18 cm (22 x 12–13 cm) |
> 514 pages |
Kephalaia 2 (CBL Pma 1) |
32 x 18–19 cm (uncertain) |
~ 496 pages |
Epistles (P. Berol. 15999) |
30 x 17.5 cm (uncertain) |
~500 pages |
Homilies (CBL Pma 2) |
26.7 x 14 cm (20 x 13 cm) |
~ 96 pages in present state (an unrestorable bookblock is now lost) |
Acts (P. Berol. 15997) |
31.5 x 18 cm (19 x 11.4 cm) |
only a handful surviving leaves, original length uncertain |
These dimensions correspond to ‘Group 5’ in Turner’s typology of papyrus codices (approximately 30 x 18 cm), which did not include the Medinet Madi corpus. Codices from ‘Group 5’ are attested from the third to the seventh century, with the most occurring in the fifth century.32 While this is consistent with the suggested date range for the Medinet Madi texts from radiocarbon analysis, what is perhaps most striking is that the Group 5 dimensions are ‘well paralleled in parchment.’33 In the case of Manichaean texts, this might have been an attempt to reproduce the size of source books from the Syro-Mesopotamian region, which would likely have been made of parchment rather than papyrus.
The Medinet Madi corpus has some important additional shared features. Although the size of individual leaves is not particularly large, in respect to the number of leaves, several of the Medinet Madi codices are among the very largest surviving papyrus codices from antiquity, with at least three originally containing 500 or more pages, and a fourth at around 496. Not surprisingly, all have multiple quires, of consistent size throughout the manuscript, either quaterniones or seniones, and all feature quire numbers, even if these are only occasionally still extant (for a fuller discussion, see section 3.4 below). Although a comparative paleography of the corpus has not been carried out, the preliminary assessment of Paul Dilley suggests multiple hands between (and sometimes within) the codices, but a relatively consistent style. With regard to page layout, all the Medinet Madi manuscripts, with the exception of the Psalmbook index, are written with a single column per page (which, it must be said, is in contrast to most early Syriac examples). All but the Psalmbook codex have running headers, a rarity in Late Antique codices from Egypt, and otherwise unattested for Coptic ones. Only one, the first volume of the Kephalaia, has page numbers. Instead, the codices are organized by numbered headings, breaking the text up into chapters, psalms, epistles, and readings.
In contrast to the Nag Hammadi corpus, as well as most of the Bodmer papyri, each of the Medinet Madi codices has a clear thematic unity, rather than being a miscellany of various texts, even if most have more than one work.34 Paola Buzi suggests that, at least until the sixth century, single-text manuscripts ‘usually transmit “orthodox” biblical texts;’ the Medinet Madi codices, though far larger than surviving Coptic biblical manuscripts from this period, must have played a similarly authoritative role for the Manichaean community.35 It is likely that they constituted part of a reference library, which wandering Electi took from one Egyptian Manichaean community to another, where they would have been copied as needed for local use. Indeed, the relative linguistic, codicological, and paleographical uniformity of the Medinet Madi corpus suggests that they were not only composed within the same community, probably in middle Egypt around Lycopolis, but also that they traveled together to their modern findspot, allegedly the Fayyum, where a different Coptic dialect was used. Despite their size, they could be transported, and indeed Manichaean texts would have constituted the Elect’s most important material possession.
2.2.3 Cologne Mani Codex
The Cologne Mani Codex (CMC) is a miniature parchment codex containing a Greek version of a biography of Mani. It appeared on the antiquities market in the 1960s, of unspecified Egyptian provenance.36 Like the Syriac fragments, it is not certain that the CMC was produced in Egypt; the use of parchment could suggest a different origin, such as Syria. The Greek text itself, likely translated from Aramaic,37 has had major ramifications for the scholarly understanding of Mani’s early life and mission. As a physical object it is likewise exceptional, being one of the smallest codices from antiquity, with a page size of only 38 x 45 mm. With 23 text-lines plus a header per page, letters of about 1mm in height, and originally about 200 or more pages, it has rightfully been described as ‘a masterpiece of ancient book production’.38
The CMC’s dating has been much debated, centering almost exclusively on paleography: the editors dated the script to the fifth century but a subsequent study argued for an eighth-century date.39 Most recently, Orsini has dated the manuscript’s two scribal hands differently: an earlier hand B was responsible for quires 2–7 and the first part of quire 8 can be compared with hands of sixth-century manuscripts, while a later hand A (who also restored damaged text from the beginning and end of the codex) was responsible for the first quire and the end of quire 8. Hand A is much later, being contemporary with eighth-ninth century manuscripts.40 However, the recent radiocarbon dating of the codex suggests a range between the mid-third century to the mid-sixth century.41 Indeed, an eighth or ninth-century copy would be unlikely given the lack of other evidence for Manichaeaens in Egypt at such a late date. Whatever its date, the CMC shares the prevalent codex form, but its small size makes it exceptional.
2.3 North Africa
Although the existence of Latin Manichaean texts, probably codices, from North Africa in Latin is both implied and explicitly mentioned in the writings of Augustine, original Manichaean Latin manuscripts have, with one exception, not survived.42 That lone exception is the Tebessa Codex (BnF Nouv. acq. lat. 1114), a fragmentary codex discovered near its namesake (ancient Theveste, modern Tebessa) in Algeria in 1918.43 The extant text preserves parts of at least one work, possibly two, which discusses the status of Manichaean elect and hearers. The codex is on parchment, written two columns to the page, in a half uncial script and dated to the fifth-sixth centuries. Because of its fragmentary nature, little else can be said regarding its codicology, as the quire structure and page format are quite unclear. The most recent studies argue that it is indeed an original Latin composition, that it may draw from the Diatessaron for Gospel quotations, and that in other places it shares readings with both the Vulgate and Vetus Latina.44 It follows the standard format and material of other surviving Late Antique Latin books from North Africa, namely the parchment codex.
2.4 Turfan
The Manichaean manuscript remains from sites in the Turfan oasis (modern-day Xinjiang, China) comprise the largest, though also the most fragmentary, find of Manichaean materials from antiquity. Several thousand fragments in Middle Persian, Parthian, Sogdian, and Old Uighur, as well as some in Bactrian, Early New Persian, and Tocharian, were discovered by expeditions to the Turfan region in the early 1900s.45 Most came from what seems to have been a major Manichaean communal complex in Qocho, with other fragments from various smaller sites in the region. It is also clear that some texts were produced elsewhere, both at local sites which have not yielded Manichaean remains in situ, as well as places farther afield, including as far west as Samarkand, and later brought or sent to Qocho.46 Many of these manuscript fragments, including the most impressive illuminated ones, were produced after Manichaeism became the official religion of the Uighur kingdom around 760 ce, and began to benefit from state patronage.47 While only one text bears a clear date in the text (M1 / MIK III 203, mentioning both 762 and 832 ce), several others can be dated to the late eighth century and some even as late as about 1000 ce through comparative means.48 The texts thus attest to original compositions from the late eighth to early 11th centuries, as well as translations and copies of much older works, including parts of the canonical works of Mani. Moreover, it has also been speculated that original works composed in Iran or Khorasan, and preceding the eighth century, are preserved among the Turfan remains.
The Manichaean manuscript remains from Turfan are not only linguistically diverse, but also show a wide range of manuscript formats and materials.49 There are numerous examples of codices, with a huge variety of page sizes attested, ranging from 5.8 x 3.4 cm to over 50 x 30 cm.50 However, no complete codex, or even complete quire, survives. The vast majority of codex fragments are written on paper, both ramie/hemp-based rag paper as well as mulberry-fibre paper, which may indicate differing places of production, but there are a handful of parchment fragments well. The codex pages are laid out with 1 to 4 text-columns per page, although the most common layout is 1 or 2 columns. There are also a number of paper scrolls. These are largely a secondary book format—Manichaean texts written on the blank versos of re-used Chinese Buddhist scrolls.51 There are also a few official letters which were written on new scrolls, and Gulácsi argues that fragments of Mani’s Picture Book from Turfan come from both scrolls and codices.52 In addition, there are a number of standalone texts on paper slips and sheets either produced to the size needed, or trimmed from Chinese Buddhist scrolls. There are even a few examples of Manichaean pustaka-format books, and texts of uncertain format on silk cloth.53 The dominant book format, however, is certainly the paper codex.
The Turfan fragments themselves, however, provide few references to the context of book production and use in the Manichaean community there. Even the documentary texts connected to the Manichaean sites mention books or writing only infrequently, such as in Bezeklik B, a letter between high-ranking Manichaean officials, which reports that various Elect read from various writings (Sogdian npyk, a generic term for a written document) during the month of fasting.54 Fortunately, there are two illuminated fragments of codex pages which depict books and writing. A depiction of the Bema celebration on MIK III 4979 verso shows an Elect holding a codex aloft, suggesting that the codex was the book form used in liturgical contexts.55 And the illumination on MIK III 6368 shows multiple Elect writing with reed pens on rectangular sheets of paper.56 These may depict loose folia which would later be assembled into quires of codices.
The Turfan manuscript remains, like those from Kellis, discovered in the ruins of Manichaean sites, show that the community made use of different types of written materials, and different book formats in particular, for different purposes. Unfortunately, though, the archaeological context of the Turfan discoveries was extremely poorly documented, and the reason for the variation in writing supports is not yet well understood, but there are relationships between the purpose of a given text, the availability of local resources for book production, and the local cultural contexts in need of further exploration. Unlike the Manichaean textual remains from Egypt, however, those from Central Asia stand out from contemporaneous neighboring manuscript traditions in several additional ways: the dominant use of the codex form is only shared with the Christian remains from Turfan (although it is not clear to what extent Christian and Manichaean communities may have overlapped or interacted there) but not with the Buddhist or non-religious administrative text remains. Moreover, the use of the Manichaean script to write several languages (in addition to the use of their native scripts), was limited to the Manichaean community. In Turfan, then, it is certainly possible to speak of a locally distinctive Manichaean book format.
2.5 Dunhuang
When Manichaeism was introduced to China, not only were many of its writings translated from Iranian languages to Chinese, but they were also incorporated into Chinese manuscript culture.57 Three Chinese Manichaean texts, all well-preserved, and one text in Old Uighur were discovered in the famed Mogao Caves near Dunhuang. In contrast to Turfan, where the majority of texts are in codex format, all of the Manichaean texts at Dunhuang, both doctrinal and liturgical, were written on paper scrolls. Unfortunately, most discussions, and even editions, of these texts do not discuss their codicological details, and so this section mostly consists of exploratory notes situating the Chinese texts within the larger orbit of Manichaean scriptures and non-Manichaean texts from Dunhuang. The sizes of the Chinese scrolls from Dunhuang are presented in Table 2 below.
| Text | Total Size | Sheet Size |
|---|---|---|
| Hymnscroll (BL Stein S2659) | 27.6 cm x 730.7 cm | 27.6 x 38.2–40.2 cm; the first (19 cm) and last (7.2 cm) differ |
| Traité/Tractatus (bei 8470) | 27 cm x 639 cm | 27 x 37–38 cm |
| Compendium (BL Stein S3969) | 26 cm x 150 cm | (no data available) |
The Hymnscroll, Traité/Tractatus, and Compendium have all been dated to the Tang period (618–907 ce), largely by their content and special characters; none are datable by colophon. Their codicology confirms their Tang-era composition, although it has not been previously discussed for the purpose of dating, either individually or as a group. All three manuscripts have measurements consistent with the Tang period according to the codicological typology for Dunhuang Buddhist scrolls proposed by Akira Fujieda, which have individual paper panels 26–28 cm in height and 40 cm in length.58 Thus the Manichaeans who wrote their texts in Chinese, whose manuscripts have been recovered from Dunhuang, adopted the typical book format of Buddhist texts, just as they made significant use of Buddhist (and Taoist) terminology; notably, the Manichaeans did so rather than adopt a codex-like structure such as the booklet, which would have been available. In short, it would seem that the Manichaean attachment to the codex, which is still evident at Turfan, had faded away at one more remove from Western Eurasia.
The Old Uighur text from Dunhuang is an example of the Xwāstwānīft, a confession manual for Elect, which is one of the best attested eastern Manichaean writings, as it is preserved in two nearly complete paper scrolls, one from Dunhuang (BL Or. 8212), and one from Turfan, currently held in St. Petersburg; there are, as well, numerous codex and scroll fragments of it from Turfan, which are currently held in Berlin.59 The Dunhuang Xwāstwānīft is quite different from the standard sutra format of the Chinese Manichaean texts: it measures 10 cm in height and 430 cm long; it was presumably brought to Dunhuang from a Turkic-speaking region, perhaps Turfan itself. The St. Petersburg Xwāstwānīft manuscript, from Turfan, is slightly larger than the Dunhuang Manichaean scrolls, being approximately 30 cm in height and 256 cm in length, with individual paper sheets of 46 cm in length. Because, as Larry Clark has argued, the two appear to have been copied from the same base text, it seems plausible that this text was first copied in Dunhuang, using the paper available there; and then brought to Turfan, where it was copied using the slightly larger format.60
3. Comparative Manuscriptology: Four Case Studies
In this section we examine in-depth four features of comparative manuscriptology related to surviving Manichaean texts, comparing the Western and Eastern Manichaean corpora with each other and with as much contemporary evidence as possible, especially Late Antique manuscripts in Syriac, Greek, Coptic, and Latin, as well as the relevant Turfan material, despite its generally fragmentary condition. In doing so, we hope to illustrate several points. First, that the detailed comparison of the physical remains of Manichaean manuscripts, despite their diverse geographical and chronological contexts, is essential for tracing the development and interconnection of Manichaean book practices across Afro-Eurasia. Second, that the comparison of Manichaean evidence with contemporary non-Manichaean evidence productively enriches the study of both. And seen more broadly, this examination of manuscriptological case studies is one way of approaching the larger question of the connectedness and continuity of the Manichaean community over great times and distances.
We have selected three paratextual features and one codicological feature for comparison. The paratextual features are text-division strategies (such as the formulas beginning and ending texts within a book), running headers, and indices to hymnbooks. The codicological feature is quire structure (including size and numbering). These four features are present to one extent or another in nearly all Christian/Eastern Mediterranean manuscript traditions, but have hardly been studied comparatively, even across early Christian corpora.61 Quire structure information is not available for many individual manuscripts. While improving existing documentation of these and other features remains a desideratum, this initial foray assembles the available evidence for Manichaean texts, comparing the surviving evidence from the Mediterranean and Central Asia.
3.1 Text Division
The early Christian and Manichaean manuscripts have various strategies for dividing texts within a manuscript, that is, indicating the end of one text, be it a section, chapter, or standalone work, and/or the beginning of another. Some manuscripts simply provide a superscriptio: the title of the work precedes the body text and is set off from it in some way, whether by means of spacing, decorative marks, or another technique. For example, Papyrus 46 (early 200s ce) precedes the text of Corinthians with the simple title ΠΡΟΣ ΚΟΡΙΝΘΙΟΥΣ Α ‘To the Corinthians (Book) 1’ in larger letters set off slightly from the main text.62 Other manuscripts make use only of a subscriptio, such as in Codex Sinaiticus (~350 ce) which ends the Gospel of John with the text ΕΥΑΓΓΕΛΙΟΝ ΚΑΤΑ ΙΩΑΝΝΗΝ ‘Gospel according to John’.63 Both these layouts — title at the beginning or at the end — occur in Greek and Coptic codices dated from the early third century through to the fifth century. The Nag Hammadi codices, for example, vary, using at times either the superscriptio or the subscriptio paratext, and, in a very few places, both together.64
Two additional text-division layouts are also in play by the fourth century, though, both of which involve paratext which brackets or frames the entire work. One is to precede a work with a superscriptio consisting of the title and follow it with an ending formula. The other is to frame the work with both a beginning and an ending formula, which will be referred to here as incipits and explicits after their Latin versions.65 Both of these layouts will be discussed in depth in this section. Strangely, however, these rather fundamental aspects of the organization of text, and indeed of the story of the development of manuscripts in general, hardly merit more than passing mention in any of the standard works on the ancient book or text criticism.66 In addition this section aims to highlight the importance of paratextual strategies for manuscript studies and sketch a general history of the appearance and usage of these strategies.
3.1.1 Title + Explicit
In the Syriac tradition, the oldest datable example of the title + explicit text-division layout occurs in BL Add. 12150, itself the oldest dated Syriac codex, with a colophon stating that it was copied in 411 ce (Fig. 1). Undated examples from about the same time include the Curetonian Gospels67 (Add. 14451; LDAB 116124) and Syriac Sinaiticus68 (ms. Sin. Syr. 30; LDAB 117850), both dated on codicological grounds to the early fifth century ce. In these codices, separate works, or separate books of the Gospel, are demarcated at their beginning by the title of the work or book set off from the main text as a superscriptio. The explicit is composed of the text title preceded by the word ܫܠܡ šlem ‘ended’. Both are typically written with red ink.
|
ܐܘܢܓܠܝܘܢ ܕܠܘܩܐ
The Evangelion of Luke
|
ܫܠܡ ܐܘܢܓܠܝܘܢ ܕܠܘܩܐ
Ended, the Evangelion of Luke |
|
ܡܐܡܪܐ ܕܬܬܘܣ ܕܠܘܩܒܠ ܡܢܝܢܝܐ
The discourse of Titus ‘Against the Manichaeans’
|
ܫܠܡ ܡܐܡܪܐ ܐܪܒܥܐ ܕܬܬܘܣ ܕܠܘܩܒܠ ܡܢܝܢܝܐ ܕܡܦܩܝܢ ܡܢ ܐܝܘܢܝܐ ܠܐܪܡܝܐܣ
Ended, the four discourses of Titus ‘Against the Manichaeans’ translated from Greek into Aramaic
|
The use of the title + explicit strategy continues to appear in Syriac manuscripts from the fifth century onwards, but little mention is made of it in works of Syriac codicology.69 It is not entirely standardized, though; for example Add. 14425, a copy of the Peshitta with a colophon dated 463/4 ce, uses only an explicit with seemingly no title preceding individual texts.70
In Coptic manuscripts, however, the use of an explicit is quite rare, which is unsurprising if one considers that Coptic texts typically had Greek antecedents.71 Of the Manichaean Coptic manuscripts, only one codex has a preserved title + explicit strategy, the Homilies codex.72 The beginning of each homily — there are four extant in the codex — is preceded by a title set off from the body text with dashed lines. The explicit is similar, but adds a verbal phrase with ϫ ‘end’ or ‘complete’ in the past affirmative conjugation—perhaps a translation of Greek ἐπληρώθη, discussed below. For clarity all examples from the Homilies codex are given below; those on page 7 are illustrated in Fig. 2.73
| [unattested] |
ⲁⲡⲗⲟⲅⲟⲥ ⲙⲡⲥⲁⲡⲥⲡ ϫ[ⲱⲕ]
The Sermon of Worship is ended (p7)
|
|
ⲡⲗⲟⲅⲟⲥ ⲙⲡⲛⲁϭ ⲙⲡⲟⲗⲉⲙⲟⲥ
The Sermon of the Great War (p7)
|
[ⲁϥϫ]ⲱⲕ ϫⲓ ⲡⲗⲟⲅⲟⲥ [ⲡⲛⲁϭ ]ⲡⲟⲗⲉⲙⲟⲥ
Ended, the Sermon of the Great War (p42)
|
|
ⲡⲙⲉⲣⲟ[ⲥ] ⲙⲡⲧⲉⲟⲩⲟ ϩⲁ ⲧ[ⲥⲧⲁⲩⲣⲱⲥⲓⲥ]
The Section of the Narrative on the Crucifixion (p42)
|
ϩⲁⲙⲉⲛ ⲁϥϫⲱⲕ ⲛϫⲓ ⲡⲙⲉ[ⲣⲟⲥ] ϩⲁⲧⲥⲧⲁⲩⲣⲱⲥⲓⲥ [ⲡⲫⲱⲥ]ⲧⲏⲣ ⲡⲁⲡⲟⲥⲧⲟ[ⲗⲟⲥ ⲙⲙⲏⲉ]
Amen. Ended, the Section on the Crucifixion of the Enlightener, the True Apostle (p85)
|
This is an exact parallel with the Syriac examples just discussed. As the title + explicit strategy is uncommon in Coptic manuscripts in general, and only attested in this way in a Manichaean text, which likely has a Syriac ancestor, even if through a Greek intermediary, it is reasonable to postulate that the strategy was carried over from Syriac book formatting by Manichaeans for the preparation of some of their Coptic (and probably also Greek) books. However, it must be noted that text-division strategies in the Medinet Madi codices vary: for example, the Epistles codex uses a superscriptio (sometimes with a simple ϩ ‘amen’ as subscriptio), while the Kephalaia codices have only superscriptiones (including chapter numbers), though the latter are related to internal text divisions.
Significantly, at least one of the Manichaean texts from Kellis, where we have clearer evidence for the use of Syriac by the Manichaeans,74 also contains a title + explicit. In this case, however, it is a wood tablet (T. Kellis 22) containing a text known as the ‘Prayer to the Emanations’ in Greek.75 The exact same strategy as the Coptic and Syriac is used, but in Greek.
|
ἡ ἐυχὴ τῶν προβολῶν
The Prayer to the Emanations
|
ἐπληρώθη ἡ τῶν προβολῶν ἐυχή
Ended, the Prayer to the Emanations
|
The Cologne Mani Codex, the longest surviving Manichaean text in Greek, does not have any extant text division strategy; it appears to have been a monographic codex (though the beginning and end are not preserved), and sections are only indicated by a superscriptio usually consisting solely of the name of a disciple.76 As mentioned previously, its dating is debated, with estimates for its original production ranging from the fifth to eighth – ninth centuries. It may thus be a slightly later witness to Egyptian Manichaean book production tradition than the Medinet Madi texts, although it is also possible that its layout reflects an ‘archaizing’ style, just as paleographers have observed for scripts, perhaps especially if the work was considered to be an early and authoritative one by the copyists.
To sum up: the title + explicit layout is not only common in the Christian Syriac book tradition, but also appears in the Manichaean Coptic and Greek. These thus agree in this paratextual strategy against the Christian Coptic and Greek book traditions, which tend to use only a superscriptio or subscriptio.77
3.1.2 Incipit + Explicit
The strategy of demarcating a work or simply framing an entire monographic text with incipit and explicit formulas seems to appear for the first time in Latin manuscripts of the Bible. We are not aware of early Latin manuscripts with other text division strategies. Although the relevant parts of many early manuscripts are not preserved, we see the incipit + explicit strategy in the earliest dated Vetus Latina versions of the Gospels. For example, in the fourth/fifth century Codex Bobbiensis, which contains only parts of Mark and Matthew, an entire page is dedicated to demarcating the transition between Mark and Matthew, containing the explicit to Mark and the incipit to Matthew in red ink (Fig. 3).78
|
euangelium cata marcum exp.
Ends, the Gospel according to Mark
|
incip. cata mattheum feliciter
Begins, (the Gospel) according to Matthew, happily
|
Of roughly similar date to Codex Bobbiensis is the Greek/Latin bilingual Codex Bezae. It uses the incipit + explicit strategy for the Latin text of the right-hand pages, but also, and uniquely among Greek manuscripts of this period, for the facing Greek text of the left-hand pages (Fig. 4).
|
ευαγγελιον κατα μαθθαιον ετελεσθη
euangelium sec. mattheum explicit
Ends, the Gospel according to Matthew
|
αρχεται ευαγγελιον κατα ιωαννην
incipit euangelium sec. iohannen
Begins, the Gospel according to John
|