Context

The history of scholarly reconstructions of the Gospel of Marcion is thoroughly outlined in the introductory sections of various monographs (BeDuhn, 2013; Gianotto, 2019; Klinghardt, 2015/2020; Klinghardt, 2021; Roth, 2015). More succinct overviews may be found in previous Journal of Open Humanities Data data papers for the von Harnack (Bilby, 2021b), Roth (Bilby, 2021c), Hahn and Zahn (Bilby, 2021d), and Klinghardt and Nicolotti (Bilby, 2021e) datasets. In the last two centuries, six complete Greek texts have been published (Gianotto & Nicolotti, 2019; Hahn, 1832; von Harnack, 1924; Klinghardt, 2015/2020; Klinghardt, 2021; Roth, 2015; Zahn, 1888/1892). The Greek reconstruction presented here is the seventh. Another major reconstruction (Bilby, 2023), utilizing an iterative open science approach to publishing, is nearing completion.

BeDuhn’s work has been widely reviewed in a consistently positive manner (Baudoin, 2016; Frankfurter, 2015; Jefford, 2014; Le Boulluec, 2016; Muir, 2016). Growing out of a 2014 Society of Biblical Literature panel on the book, several more positive reviews (Pervo, 2015; Rothschild, 2015; Scherbenske, 2015; Winninge, 2015) were published in the journal Early Christianity, which also gave the author a chance to respond (BeDuhn, 2015). Some reviewers found the case against Marcionite omissions persuasive (Frankfurter, 2015; Pervo, 2015), while others remained convinced of the early orthodox hypothesis of Marcion as a textual abridger (Le Boulluec, 2016; Winninge, 2015), yet both the research and translation were praised.

Both the monograph (BeDuhn, 2013) and responsive review (BeDuhn, 2015) explore the thinking behind the choice to reconstruct the Evangelion in English. The ambiguity and multilingual nature of patristic attestations and manuscript variants made it difficult to establish precise wording in Greek, but a coarse-grained English translation could accept this ambiguity while still achieving greater clarity and moving beyond the limitations of previous editions (Tsutsui, 1992; von Harnack, 1924). Subsequent Greek reconstructions (Bilby, 2023; Klinghardt, 2015/2020; Klinghardt, 2021; Nicolotti, 2019; Roth, 2015), along with five critical translations (Bilby, 2023, English; Gramaglia, 2017, Italian; Klinghardt, 2015/2020, German; Klinghardt, 2021, English; Nicolotti, 2019, Italian) have since shifted the discourse, meticulously organizing, detailing, and engaging the patristic evidence and manuscript variants, and increasingly clarifying philological patterns. While editors and translators have expressed widely varying degrees of confidence about the reconstructions—whether their own or those of others—, the international scholarly discourse has collectively become more scientifically-minded. Gramaglia (2017) thoroughly critiqued Klinghardt’s work by means of word frequencies; Bilby (2023) has endeavored to shift the discourse and methods to integrate them with currents in open access, data science, and computational linguistics. The rapid evolution of scholarly engagements with Marcion’s scriptures led BeDuhn to reevaluate his initial position and collaborate with Bilby on a Greek reconstruction and corresponding datasets under his editorial control.

The born-digital critical edition presented here for the first time is not only a companion volume to the English presentation (BeDuhn, 2013), but also a major improvement over its print-based counterparts. As an open-access text, it is accessible to a far broader audience than other recent editions. The simple typography in the PDF—using bold font only to offset wording attested in Greek witnesses, rather than Latin, Syriac, etc.—leaves aside quasi-scientific preoccupations with subjective editorial confidence and instead facilitates quantitative comparisons and statistical validation of linguistic patterns across data derived from diverse and complex attestations. Its rigorous, consistent segmentation and clear layout accommodates a wide variety of use cases, both human- and machine-based. Its critical apparatus at many points is more fully conversant with other recent scholarly editions than those editions are with each other and often contains a more thorough accounting of variants than found even in the copious works of Klinghardt and Nicolotti. In summary, it is a significant contribution to the scholarly discourse of the last decade and a bridge toward future scientific and computational research.

Reading Preferences

Reconstructing a Greek text of the Evangelion requires adjusting the principles used when distilling its semantic content from all available attestations in multiple languages (BeDuhn, 2013). Tertullian of Carthage’s early third century Latin treatise Adversus Marcionem provides the most attestations; but given the ambiguities of Latin translation in relation to Greek and Tertullian’s indirect citation habits, it should not be given preference over Greek sources. Instead, as a general rule, Epiphanius of Salamis’s late fourth century work against heresies, Panarion, should be preferred for the Greek text wherever he preserves it, despite its significantly later date, not only because he wrote in Greek, but also due to his procedure of copying out passages from the Evangelion verbatim before commenting critically upon them. The approach taken by Epiphanius avoided the fragmentation and syntactical reworkings of Evangelion content characteristic of Tertullian’s rhetorical rebuttal. This distinction applies only to Epiphanius’s Scholia, since his Elenchoi were composed later when he no longer had the Evangelion for reference and employ a rhetorical approach similar to Tertullian’s, often with significant paraphrasing. Epiphanius is generally to be preferred to the other major Greek source, the fourth century dialogue of Adamantius (falsely attributed to Origen of Alexandria), due to uncertainty over how strictly the latter adhered to the Evangelion text in specific instances and inconsistencies between its Greek and Latin versions. Nevertheless, correlation of the evidence of Epiphanius, Tertullian, and the Greek and Latin versions of Adamantius has proven its overall value.

Challenges and Resolutions

The published data paper for the von Harnack datasets (Bilby, 2021b) sets forth the normalization standards developed to transform von Harnack’s noisy reconstruction into a singular, clean, and consistent transcript (or text-as-dataset) whose words can be tokenized and thus quantitatively analyzed and compared by humans and machines. By comparison, BeDuhn’s indications are much clearer, simpler, and more consistently implemented.

1. plain text for clearly attested content
2. italics for wording attested for the Marcionite text for which no corresponding variant(s) are identified for the canonical version
3. ellipses … to indicate gaps within attestations and possible, though unknown wording
4. [square brackets] for “connective content” necessary for the intelligibility of adjacent attested content
5. (words within parentheses) for English words not explicitly part of the underlying Greek text, but necessary or helpful to the intelligibility of the translation

BeDuhn’s text proved easier to normalize as positive data than did most other Evangelion versions, in part because possible variants are detailed within the extensive endnotes, rather than in the main running text itself, and passages that are attested as not present or are unattested (excepting “connective content”) are simply omitted rather than detailed within the main text. To normalize the given wording, wording corresponding to 1 and 2 is rendered in normal font, 3 replaced with empty parentheses, 4 kept in square brackets, and 5 ignored. After the Greek text was established, variants were indicated in the apparatus, rather than—in the habit of several other Evangelion editors—noting them within the main running text. In anticipation of future syntactical tagging, Bilby updated his 2021 normalization standards to include Greek colons and periods, and also to separate fields in the datasets with commas, while still omitting commas within the main text in all files. New column headers were also added to the first row, and quotation marks were added as wrappers around each field to facilitate computational parsing. Consistent with previous Evangelion data papers and datasets, the BibleWorks Greek Morphology (BGM) encoding was again used for part-of-speech tagging (Bilby, 2021a).

Quality and Version Control

Quality control is quite different for this born-digital text of the Evangelion compared to that used for its print-born counterparts, since transcription and normalization processes were largely replaced by the collaborative translation and reconstruction of a digital text. Bilby commenced this process first by creating a normalized English text of BeDuhn’s edition, and then, with BeDuhn’s permission, creating a preliminary Greek version as a reverse-translation of BeDuhn’s published English reconstruction. To do this, Bilby considered several options for each word choice in light of past Greek editions, research for his own Greek reconstruction, and canonical manuscript variants. Some of these options and choices filtered into the beginnings of a critical apparatus. BeDuhn then revised this retroversion and apparatus in consultation with his research notes for the 2013 edition and after a fresh look at the patristic and manuscript evidence, by turns confirming and updating the word choices to become a reliable reflection of his sense of the Greek text behind his English presentation. BeDuhn additionally made corrections to the text in several instances where his thinking had changed, indicating these clearly in the apparatus. The refinement process involved several more back and forth collaboration cycles, repeated until the main text attained a stable state. The apparatus gradually expanded around portions of the text attested in Greek, but remained more selective for portions attested only in other languages, partly in anticipation of future, larger-scale collaborative critical editions.

It is hoped that this edition and the related datasets will be a blueprint and template for future digital critical editions of Marcion’s Evangelion, and also prove to be the new gold standard for Humanities scholars and especially for computational linguistics research on this text. It is simultaneously the most recent reconstruction and dataset published in full, but also achieves a reasonable, moderate standard that avoids both minimalistic and maximalistic extremes. Amidst word counts that fluctuate wildly from edition to edition (14,442 in Hahn; 10,572 Zahn; 4338 von Harnack; 4169 Roth; 12,850 Klinghardt; 10,870 Nicolotti), BeDuhn’s 7206 tokens should be considered the new via media standard for all future reconstructions and as a solid baseline for quantitative research.

Dataset Creators

Mark G. Bilby (PhD, University of Virginia) and Jason D. BeDuhn (Northern Arizona University) created these datasets.