The second-century physician and bibliophile Galen (c. 129–216 ce) tells a story about a ‘man of letters’ (‘τις ἀνὴρ τῶν φιλολόγων’) who encounters a bookroll titled ‘The Doctor by Galen’ at a bookseller's stall in Rome. Yet the work has been misattributed. Although it addresses a medical topic, it does not match Galen's style (‘λέξις’) and is falsely titled (‘ψευδῶς ἐπιγέγραπται’). In Galen's self-aggrandising account, the title (‘ἐπιγραφή’) is a dishonest bookseller's clever ploy to pass off an inferior work by connecting it to a renowned medical writer. But a discerning reader can tell the difference. Upon reading only the first two lines, the educated man recognises the deception and corrects the misattribution by ripping up the book tag (or the whole book: ‘ἀπέρριψε τὸ γράμμα’) with the attribution to Galen.Footnote 1 This vignette reveals how, for Galen and other elite readers in the Roman Mediterranean, correctly identifying literary works was a mark of παιδεία, vital for maintaining their status as cultural arbiters.Footnote 2 Debates about authenticity, attribution and textual transmission appear again and again in the self-fashioning of Roman elites.Footnote 3
Early Christian thinkers participated in these bibliographic debates about titles and authenticity. Correct attribution was part of shaping the practices of a reading community and of asserting and maintaining privileged literary corpora: ‘Scripture’, ‘the Gospels’. Arbiters of these corpora positioned themselves as tastemakers for others.Footnote 4 The present article analyses one complex set of relationships between texts, titles and works in late antiquity. The relationship between the Gospel according to Matthew and the Gospel according to the Hebrews reflects a late ancient project of bibliographic categorisation that continues to shape how modern scholars read evidence about Gospel books and readers in the first several centuries ce.
Late ancient figures, Christian and otherwise, exhibit a remarkable preoccupation with bibliography – the practice of organising knowledge about books and their readers – as a way of knowing the world.Footnote 5 Bibliography is seldom, even never, just about cataloguing the library. Again and again, bibliographic thinking provides a way for people to organise wide vistas of knowledge and experience – including phenomena that are not bookish in and of themselves. Talking about books is a way of talking about other things: ethnography, cosmology, theology and so forth. Inversely, other ways of thinking about the world often intervene in the practice of bibliography. Organising the world and organising the library go hand in hand.
Ancient bibliographic thinking affords a revisionist analysis of the texts known in modern scholarship as ‘Jewish Christian’ Gospels.Footnote 6 Early Christian thinkers leveraged distinctions between texts, titles and works in order to categorise ‘Jewish’ and ‘Christian’ books and readers. This process of organising texts and readers emerges as a heresiological strategy within the broader developments often described as the ‘parting of the ways’. The fact that modern scholars or ancient heresiologists have imagined a Gospel according to the Hebrews distinct from a Gospel according to Matthew is the result of the late ancient practice of organising the world in bibliographic terms. That practice was deployed for particular theological ends, specifically the effort to identify what books and what readers were Christian and to distinguish them from other books and other readers defined as Jewish. This bibliographic development reflects the role of Gospel reading in late ancient constructions of Judaism and Christianity. Heresiologists’ shifting categorisations of books and readers both illuminate late ancient textual practices and continue to influence modern scholarship.
In what follows, it is first demonstrated that the material which modern scholars associate with one or more ‘Jewish Christian’ Gospels – and which late ancient writers associate with a Gospel according to the Hebrews – reveals a substantial textual relationship to the Gospel according to Matthew. Then it is shown that Christian heresiologists from the second to fifth centuries describe several individuals or groups who use only Matthew and not other Gospels. Critics characterise these groups as observing Torah in particular ways and sometimes associate them with Jewish ethnicity. A change in description occurs, however, as fourth- and fifth-century critics of these same groups characterise their Gospel-reading practices differently. The library has been reorganised: the Gospel that these people read is not Matthew; instead, it is a Gospel according to the Hebrews. Finally, the article analyses the implications of this bibliographic recategorisation for Christian thinking with and about books – and thereby about Jewishness – in late antiquity.Footnote 7
This article advances no claims about the religious demography of the Roman Mediterranean. This argument is not about who went to synagogue, about who participated in Christian liturgies or about who observed Torah and how. Such questions are part of conversations about the parting of the ways, but the evidence here is ill-suited to answer them. Instead, this article analyses an intriguing early Christian conversation about books and about the reported or imagined readers of those books.
The ancient authors discussed in this article were influential in their own day and have continued to be so, to varying degrees, for later Christians and for the history of scholarship. They include familiar names: Irenaeus, Origen, Eusebius, Epiphanius and Jerome. Yet these figures are not representative of Christ-followers in late antiquity. Rather, they have, through a confluence of elite education, influence in their own day and later reception, secured a place in histories of early Christianity. When Irenaeus or Epiphanius tells us about Ebionites or Nazoreans, historians should not assume that the heresiologists accurately describe social groups and their practices, or even that the groups that they describe existed as groups at all.Footnote 8 The present argument does not require such transparency from the late ancient sources. Rather, late ancient writing about books illuminates the ways in which these writers thought about what it meant to be Christian and what it meant to be Jewish. What were these early Christian figures doing when they wrote about a Gospel according to the Hebrews?
Matthew and related textual traditions
The text or texts which early Christian writers describe and cite as a Gospel according to the Hebrews was related to the text that modern readers know as Matthew.Footnote 9 They were sufficiently similar that the Gospel according to the Hebrews sometimes circulated as an alternate edition under the title of Matthew.Footnote 10
This argument encounters two inescapable complexities. First, no extensive texts survive from what late ancient Christian writers called a Gospel according to the Hebrews. Instead, historians have various short descriptions and citations, a constellation of fragments embedded in varied literary and argumentative contexts.Footnote 11 The scant evidence is often contradictory. The fourth-century bishop and heresiologist Epiphanius of Salamis (d. 404 ce) is worryingly unreliable, but he is the source for much of what scholars have. Epiphanius’ contemporary Jerome of Stridon (c. 347–420 ce) is, if anything, less trustworthy.Footnote 12 Historians must look for the model that best explains the evidence, but questions remain. Second, debate continues over how many ‘Jewish Christian’ Gospels existed.Footnote 13 In a recent monograph, Andrew Gregory posits two: a Gospel according to the Hebrews and a Gospel according to the Ebionites. This Gospel according to the Ebionites is a way of treating Epiphanius’ citations as a distinct text.Footnote 14 Other modern scholars argue that Jerome has two separate Gospels, yielding a total of three ‘Jewish Christian’ Gospels. This third text is termed the Gospel of the Nazoreans. Yet the titles ‘Ebionites’ and ‘Nazoreans’ are modern fictions – and the works, as such, might be too. At best the titles offer a convenient shorthand, but they often create confusion about the ancient evidence. Only two titles appear in the sources: the Gospel according to Matthew and the Gospel according to the Hebrews. Numerous late ancient figures treat these two titles as related or interchangeable.
The current debate about the number of ‘Jewish Christian’ Gospels is thus misleading. The fiction of modern nomenclature leads scholars to think that they are dealing with distinct works and with distinct groups of readers, but matters are more complicated. The Gospels that Eusebius of Caesarea (c. 260–339/340 ce), Epiphanius or Jerome called ‘according to the Hebrews’ were not identical to one another or to what modern readers know as the Gospel according to Matthew; the available evidence indicates at least minor variations. Yet recent work in material philology and reception history has demonstrated that differing textual forms or titles need not indicate distinct works, much less separate reading communities.Footnote 15 These are questions of bibliographic reception. What texts do different readers choose to read? How much do they care about variations in title or text?
Three arguments demonstrate the similarities between the Gospel according to the Hebrews and the Gospel according to Matthew. First, Epiphanius and Jerome both present Matthew and Hebrews as interchangeable designations for the same text (Epiphanius, Panarion xxx. 3.7; Jerome, Adversus Pelagianos iii.2; Commentariorum in Matthaeum libri IV xii. 13). Epiphanius, for example, writes that the Ebionites ‘accept the Gospel according to Matthew … They call it, according to the Hebrews’.Footnote 16 Given these authors’ efforts to demonstrate that the text is not really Matthew, historians should take seriously their tacit admission that the texts were frequently interchangeable.
Second, Jerome asserts that the Gospel according to the Hebrews – which he claims to know in Hebrew or Aramaic but uses in Greek – was related to an original Hebrew Matthew.Footnote 17 This does not offer evidence for an actual Hebrew (or Aramaic) Matthew, but it does indicate that someone looking at a Greek text that they called ‘according to the Hebrews’ could see it as another version of the material that they associated with the title ‘according to Matthew’.Footnote 18
Third, excerpts attributed to a Gospel according to the Hebrews often reflect close relationship with the text of familiar Matthew. Some material that early Christian writers cite from a Gospel according to the Hebrews is not attested in other forms of Matthew.Footnote 19 This is to be expected; various late ancient scholars quarried a Gospel according to the Hebrews to find ‘extra’ material not preserved in the four Gospels that they regarded as canonical. Yet the material that these early Christian writers associate with a Gospel according to the Hebrews often does intersect with the Matthean textual tradition. This is evident in the material from Epiphanius. Compare, for example, the baptism account with those of other Synoptic Gospels.Footnote 20 As part of the Ebionites’ Gospel according to Matthew – which Epiphanius also describes as ‘according to the Hebrews’ – Epiphanius cites an expanded text that incorporates recognisable Lukan material. Yet one can read it, like Epiphanius does, as an expanded form of Matthew. Moreover, the textual tradition of familiar Matthew includes details that parallel Epiphanius’ Ebionite Gospel. For example, Epiphanius mentions a light at Jesus’ baptism (Panarion xxx. 13.7). The same detail appears in two Old Latin codices at Matthew iii.16.Footnote 21 This suggests that the variation fits within the spectrum of textual variation in the text of Matthew. Elsewhere, Epiphanius says that the Ebionites ‘chop off’ (‘παρακόψαντες’) the genealogies of Matthew and begin with John's baptism.Footnote 22 Epiphanius’ claim indicates that he understood the Ebionites’ Gospel as a modified Matthew.
Epiphanius is not the only writer to present a Gospel according to the Hebrews as a form of Matthew. In his Commentary on Matthew, Origen of Alexandria (c. 184–254 ce) discusses parallel versions of the story of the rich young ruler.Footnote 23 He analyses versions from Matthew, Mark and Luke (Matt. xix.16–24//Mark x.17–25//Luke xviii.18–25) and then an expanded narrative in a Gospel according to the Hebrews. A passage attributed to Eusebius’ Theophany recounts an alternate version of Matthew's parable of the talents (Matt. xxv.14–30; cf. Luke xix.12–27).Footnote 24 Instead of two industrious enslaved persons and one timid enslaved person, this version of the parable involves a profligate enslaved person, an industrious enslaved person and a timid enslaved person.
Jerome's engagement with an alternate Gospel text occurs primarily in his Commentary on Matthew. He reports an alternate version of the healing of a man with a withered hand (cf. Matt. xii.13).Footnote 25 He states that the Gospel according to the Hebrews interprets the name Barabbas.Footnote 26 He asserts that this same text (‘the Gospel we have often referred to’) mentioned the shattering of the lintel of the Jerusalem Temple in addition to the tearing of the Temple curtain.Footnote 27 He also reports variant readings which make sense only if the text resembles Jerome's Matthew. Examples include the Hebrew מחר (maḥar) for crastinum (ἐπιούσιον) or ‘daily’ in Matt. vi.11Footnote 28 and the reading ‘son of Jehoiada’ for ‘son of Barachiah’ at Matt. xxiii.35.Footnote 29 Finally, several medieval manuscripts of Matthew attribute marginalia to τὸ ἰουδαϊκόν.Footnote 30 In ancient and late ancient textual scholarship, the neuter substantive adjective invites one to supply the noun ἀντίγραφον (‘copy’). These marginalia attest an alternate ‘Jewish’ version that a late ancient scholar collated into the margins of familiar Matthew.
One or more varying forms of Matthew best explain this data. Some examples could be expansions or alternate versions of material from any Synoptic Gospel. But others depend on distinctive Matthean material or redactional features.Footnote 31 Moreover, this textual relationship to Matthew applies to material that scholars have associated with all three hypothesised ‘Jewish Christian’ Gospels – Hebrews, Ebionites and Nazoreans.
This Matthean textual fluidity is not out of the ordinary for early Christian Gospels. Mark has multiple variant endings. The D-text (or ‘Western’ text) of Luke includes additional material in a work that is still known as Luke (for example, at vi.4). Not only does John come to include the pericope adulterae (vii.53–viii.11), but several significant shorter plusses appear in the first few centuries. What early Christians cite from a Gospel according to the Hebrews is no more extensive or dramatic than these variations in the textual traditions of other Gospels. In each case, additional material finds its way into a Gospel text and multiple differing textual forms of a work circulate under the same title.Footnote 32
The Gospel according to Matthew becomes a distinct Gospel according to the Hebrews through a process of bibliographic differentiation. This does not occur for Mark, John or the D-text of Luke. But historians might compare the emergence – the invention, even – of the Gospel according to the Hebrews with an earlier moment of bibliographic differentiation: the distinction between Marcion's Gospel and the Gospel according to Luke.Footnote 33 In both cases, bibliographic distinction maps textual difference. But differentiation goes beyond textual criticism. Distinctions between books and titles reflect not simply different texts but a division between reading communities. Historians must therefore inquire not about the ontology of the text, but about its sociology.Footnote 34 In the case of Marcion, in the second century ce, the differentiation of books and readers was mutual. Marcion wished to distinguish his Gospel from related texts that were read as the Gospel according to Luke; Marcion's critics, from Irenaeus onward, were happy to identify Marcion's Gospel as different and defective. It is unlikely that the distinction between Gospels ‘according to Matthew’ and ‘according to the Hebrews’ was similarly mutual. None the less, in the relationship between these two Gospels, a bibliographic parting of the ways occurs. It is a recategorisation of books and of readers, motivated by questions of heresiology and ethnicity – by questions of Jewishness.
‘Only the Gospel according to Matthew’
A number of early Christian writers describe ‘heretical’ groups who use only Matthew. Heresiologists described these groups as sharing ideas, practices and texts.Footnote 35 Ebionites or their imagined founder Ebion appear in several second- and third-century texts which ascribe a Jewish profile to the group.Footnote 36 In fourth- and fifth-century texts, Ebionites are joined by Cerinthians and Nazoreans as heretics who practice ‘Jewishly’. For heresiological writers, these figures form a cluster, marked by Jewish ethnicity or practice.Footnote 37
Several heresiologists discuss Gospel-reading practices. In his five-volume treatise Against heresies, the second-century bishop Irenaeus of Lyon (fl. c. 180 ce) asserts that heretics known as Ebionites ‘use only the Gospel according to Matthew’. Irenaeus describes other Ebionite practices: they are said to circumcise, observe the law and revere Jerusalem as God's house. They were, according to Irenaeus, ‘Jewish in their way of life’ (‘et iudaico charactere uitae’).Footnote 38 In a second passage, Irenaeus again states that ‘the Ebionites use only the Gospel according to Matthew’ (‘Ebionei etenim eo euangelio quod est secundum Matthaeum solo utentes’).Footnote 39 He describes several other groups who use only forms of one or another Gospel. Four particular Gospels are so well established, on Irenaeus’ account, that even heretics appeal to them: Ebionites use only Matthew; Marcion's followers use only a form of Luke; Valentinus’ followers use John, but not other canonical Gospels; unnamed individuals who distinguish Jesus from the Christ prefer Mark. Although Irenaeus complains that Marcion ‘maims’ (‘circumcidens’, literally ‘circumcises’) Luke's Gospel, he makes no such claim about the Matthew that Ebionites use. If there are textual differences, they do not yet pose a problem.
Other heresiologists likewise associate certain readers with a preference for Matthew. The fourth-century bishop Epiphanius claimed that Cerinthians used only Matthew.Footnote 40 He wrote that the Ebionites ‘accept the Gospel according to Matthew. Like the Cerinthians and Merinthians, they too use it alone. They call it, according to the Hebrews’.Footnote 41 The fourth-century heresiologist Filastrius, who relies on both Irenaeus and Epiphanius, likewise asserts that Cerinthus, the eponymous founder of the Cerinthians, ‘accepts only the Gospel according to Matthew. He spurns the [other] three Gospels’.Footnote 42 As demonstrated above, Epiphanius and Jerome attest a relationship between Gospels according to the Hebrews and according to Matthew. Yet fourth- and fifth-century writers identify these varied ‘Jewish’ groups as readers of a different Gospel, one differentiated from Matthew.
A ‘Gospel according to the Hebrews’
Starting in the fourth century, several heresiological writers describe the reading habits of Cerinthians, Ebionites and Nazoreans differently. These groups are now said to use only a Gospel according to the Hebrews. In other words, these heresiologists attribute the Gospel according to the Hebrews to the same figures who had previously been described as using only Matthew. One description replaces the other.
Eusebius of Caesarea describes the Gospel used by Ebionites as a Gospel according to the Hebrews. He writes, ‘[the Ebionites] used only what is called the Gospel according to the Hebrews; the rest they gave short shrift’.Footnote 43 This shift in description is even more significant because Eusebius employs Irenaeus as one of his main sources. Elsewhere in his History, Eusebius categorises this Gospel according to the Hebrews among the νόθα, books of mixed parentage.Footnote 44
Epiphanius makes a similar move in his Panarion.Footnote 45 Although he acknowledges a relationship between the Gospel according to Matthew and the Ebionites’ Gospel according to the Hebrews, Epiphanius emphasises differences between the two, arguing that the Ebionites’ Gospel is not properly Matthew. Epiphanius asserts that ‘the Gospel that [the Ebionites] call according to Matthew … is not at all complete but is illegitimate and mutilated’ (‘νενοθευμένῳ καὶ ἠκρωτηριασμένῳ’: Panarion xxx. 13.2). Epiphanius describes this text by the title ‘according to the Hebrews’ except in contexts where associating it with Matthew aids his complaints that the Ebionites’ irresponsible practices damage the text (thereby changing it from Matthew to Hebrews). Ironically, Epiphanius’ efforts to distinguish the two texts provide rich evidence for the relationship between the Ebionites’ Gospel and other forms of Matthew. Yet Epiphanius emphasises that the Ebionites’ Gospel is ‘according to the Hebrews’ and not ‘according to Matthew’. This is a way of removing apostolic authority and canonical status from the Gospel that these other Christ-followers employ.Footnote 46
There is a similar impulse in Jerome's engagement with a Hebrew or Jewish Gospel. Jerome is no more welcoming of Ebionites or Nazoreans. As he writes, ‘What shall I say of the Ebionites who claim to be Christians? … since they want to be both Jews and Christians, they are neither Jews nor Christians.’Footnote 47 Throughout his corpus, Jerome criticises Ebionites and Nazoreans as heretics. Like Eusebius and Epiphanius, Jerome distinguishes the Gospel according to the Hebrews from the Greek Matthew that he deems canonical. None the less, he leverages its connection (real or imagined) with Jews, Judaism and the Hebrew language to advertise his own erudition.Footnote 48
Bibliography as a parting of the ways
In the fourth and fifth centuries, a process of bibliographic recategorisation occurs.Footnote 49 Groups that were once described as using only Matthew are now said to read a different text, the Gospel according to the Hebrews. This form of Matthew has been catalogued under a different title and relocated to another part of the library. This Gospel according to the Hebrews is presented as Jewish, not Christian. Under its new shelf-mark, the text remains available to heresiological writers, but it affords different uses. It is demarcated from the emergent canonical Gospel tradition, but – because it is catalogued as a Jewish text – scholarly readers can appeal to it for linguistic and historical information.
The figures described as Ebionites, Nazoreans and so forth might have modified their habits of Gospel-reading over time, whether by using a different Gospel or by naming the same Gospel differently. Either of these possibilities would attest that such readers sought to distinguish themselves and their books from other Christ-followers and their books. If these Christ-followers began to call their own Gospel text ‘according to the Hebrews’, this decision about nomenclature would reflect bibliographic work as a process of self-definition between groups of Christ-followers. After all, a mutual – although not amicable – process of bibliographic distinction occurred in the case of Marcion's Gospel. But Epiphanius and Jerome uneasily attest an interchangeability between Gospels ‘according to Matthew’ and ‘according to the Hebrews’. Both acknowledge that (some) readers of this text call it ‘according to Matthew’. These observations suggest that mutual separation is not an adequate account for the change in title and categorisation of this Gospel.
The evidence reflects a heresiological project of bibliographic recategorisation. Heresiological writers reclassify the Matthew used by (ostensibly) Jewish Christ-followers as a different Gospel. Insofar as Gospel texts and liturgical reading were fundamental to Christian practice and identity in late antiquity, describing the Gospel used by Ebionites or Nazoreans as not Matthew, as not canonical, was a way of excluding such Christ-followers from being Christians at all. As Chris Keith writes, ‘reading events in assembly would eventually become a litmus test for canonicity’.Footnote 50 The inverse also is true: the texts that are read would eventually determine the validity of a reading event and a reading community. Differentiating books is a way of differentiating readers.
This reconstruction reveals heresiological writers in the fourth and fifth centuries addressing questions of ‘heresy’ and ‘Jewishness’ in bibliographic terms, as about what books one reads and how one reads them.Footnote 51 These figures are demarcating Christians from Jews; this bibliographic distinction is one way of drawing these lines. This is not simply a parting, then, but ‘an imposed partitioning of what was once a territory without border lines’.Footnote 52 This early Christian project of redescription advances a bibliographic parting of the ways that occurs, as it were, in the library stacks.
This recategorisation is not a necessary response to textual plurality. There were differing textual forms of Matthew. Yet there were also differing forms of Mark, Luke and John, and the same late ancient Christian thinkers discuss those differences. The separation between Matthew and Hebrews is retrospective, an imaginative attempt to divide Christians from others by separating Christian books from the books of others: Christians and Jews cannot share books. Bibliographic recategorisation is motivated by worries about overlapping libraries and intersecting communities of readers. Defining the limits of a textual work is a way of defining the limits of a reading community.
This heresiological move enables late ancient Christian writers to use the text that they call the Gospel according to the Hebrews in new ways. They employ it as a parallel form of Matthew and as a source of scholarly detail. Different authors exhibit more or less anxiety about this text. Epiphanius is especially critical. He describes the Gospel used by the Ebionites in harsh terms as ‘illegitimate and mutilated’ (Panarion xxx. 13.2). Others are more positive. Jerome differentiates this Gospel text from Matthew and describes it as a Hebrew and Jewish source of knowledge, but he puts it to work as a supplemental Gospel. Even in texts less motivated by heresiological polemic, this Gospel is framed as Hebrew or Jewish. We find both grudging respect and wariness, both visible from early on, such as in (pseudo-)Origen's emphasis that the Gospel according to the Hebrews is not authoritative, although it provides a parallel to material in Matthew (Commentarium in evangelium Matthaei xv. 14). Appeals to a Gospel according to the Hebrews present the exegete as having access to special, ethnically coded knowledge.
Titles, citations and bibliographic descriptions illuminate broader developments in late ancient social and intellectual history. Christian thinkers in the fourth and fifth centuries addressed questions of heresy and Jewishness in bibliographic terms, differentiating various forms of the Gospel according to Matthew along heresiological lines. Like Galen's educated reader, distinguishing between rightly attributed books and fraudulent knockoffs, these early Christian readers displayed their expertise – and sought to control the boundaries of a reading community – by identifying the Gospel used by Ebionites and Nazoreans as a distinct Jewish work, a Gospel according to the Hebrews. Ongoing use of this material is inflected by an imaginative recategorisation of Gospel books and Gospel-readers. Bibliography is a way of organising identities, between Jewish and Christian, and thereby advances the late ancient rhetorical project that we know as the parting of the ways.