The editions that begin with Tsutsui in 1992 were written under a double pressure. On one side stood Harnack’s text, still the default citation, and the growing sense, named by Aland, May, and Beyschlag, that a new complete reconstruction was required. On the other side stood Knox, Hoffmann, Tyson, and Klinghardt, who had reopened the literary question, and Schmid, who had shown on the Apostolos what a reconstruction controlled by citation habits looked like. The new gospel editions disagree with one another, and they disagree about what an edition is. Some print a discontinuous text with graded confidence. Some print a continuous narrative that fills silence from Luke or from the Western tradition. One prints English under the title The First New Testament. The disagreement is the old disagreement, now equipped with better source editions and with a century of argument about what silence may be made to say.
What Had Changed in the Sources
No new Marcionite manuscript appeared. What changed was knowledge of the old witnesses. Tertullian’s Against Marcion received Claudio Moreschini’s edition, Ernest Evans’s text and English translation, and René Braun’s Sources chrétiennes volumes. New scriptural indices, including Biblia Patristica, made it possible to compare Tertullian’s Marcion citations with his citations of the same verses elsewhere. The language question, Greek copy or Latin translation, was reopened and, for a growing number of scholars, closed in favor of Greek. Epiphanius received Jürgen Dummer’s revised Holl, an index volume, and Frank Williams’s English Panarion. Carroll D. Osburn studied Epiphanius’s own text of Acts and the letters. The Adamantius dialogue received Vinzenz Buchheit’s edition of Rufinus and Tsutsui’s 2004 commentary on books I–II. The Greek is still read in W. H. van de Sande Bakhuyzen’s 1901 edition, which Harnack already used. Schmid’s 1995 monograph, though written for Paul, became a manual of citation habits for the three main witnesses.
The archive itself has not grown. Tertullian’s fourth book remains the longest witness. Epiphanius’s seventy-eight scholia remain the nearest thing to a collation. The Adamantius dialogue remains later and mixed. A handful of minor witnesses add scattered verses. Schmid’s Apostolos study had already shown how much labor a smaller collection requires once citation habits are taken seriously. What has increased is the possibility of knowing how each witness cites scripture when he is not talking about Marcion.
Claire Clivaz suggested in 2005 that P69 might be a fragment of Marcion’s redaction of Luke. The suggestion rests on silence about verses that are unattested for Marcion, and it has not won general acceptance, though BeDuhn took it up in 2013. The source situation of 1924 is still the source situation. The difference is that an editor in 2015 can know those sources in a way Harnack could not.
Tsutsui 1992
Kenji Tsutsui’s “Das Evangelium Marcions: Ein neuer Versuch der Textrekonstruktion,” published in the Annual of the Japanese Biblical Institute in 1992, is the first complete reconstruction of the gospel since Harnack. Tsutsui printed the text in three rows: Tertullian’s Latin, the references of Epiphanius and Adamantius, and text-critical discussion. The main text is Latin, not Greek, which makes the edition unique among the eight major reconstructions. Tsutsui differed from Harnack at a list of verses he was careful to enumerate. He also wrote that Harnack’s name would live forever and that Harnack’s reconstruction was not a sanctuary.
After nearly seventy years a new edition needed a new method as well as a new text. Tsutsui treated Tertullian’s “direct quotations,” introduced by a verb of saying or kept in an independent sentence, as especially reliable and therefore as matter that could be discussed more fully than the rest. René Braun pointed out that this did not take sufficient account of Tertullian as a citer, and that Tsutsui did not even apply his own rule consistently. Clabeaux had already shown, for the Pauline corpus, how little a verb of saying guarantees. Tsutsui did not use Clabeaux. Schmid’s judgment was that the approach must be counted, in method, as a step backward behind a problem-consciousness already reached. Klinghardt later said that the same negative evaluation he offered of Harnack applied to Tsutsui as well. Bilby’s normalized count, converting Tsutsui’s Latin by a Vulgate ratio, puts the 1992 edition at roughly 3,300 Greek-equivalent words, the smallest of the eight. The first new reconstruction was a minimal, Latin, discontinuous text, written in Japan, published in a Japanese annual, and still often skipped in English narratives that jump from Harnack to BeDuhn. It is the edition that broke Harnack’s monopoly.
BeDuhn 2013
Jason BeDuhn’s The First New Testament: Marcion’s Scriptural Canon (Polebridge, 2013) gave general readers and scholars an English Evangelion and an English Apostolikon in one book. BeDuhn introduced Marcion, reconstructed both texts, and argued for the consequences of that reconstruction for Luke–Acts, the two-source theory, and Q. The English edition occupies a middle ground, in Bilby’s description, between a continuous and a discontinuous approach. The 2021 census papers convert that English to 7,419 Greek-equivalent words. Bilby’s later Greek dataset, made with BeDuhn’s revision of a reverse translation, counts 7,206 tokens. The conversion figure is the one that first set the eight editions on one scale; the later count is of a Greek retroversion, not of a different edition. Either number sits between Roth’s sparsity and Klinghardt’s fullness. BeDuhn’s is the first gospel reconstruction since Harnack’s to treat the Apostolos as a full partner in the same volume, and the first to do so in English.
The book is both a reconstruction of wording, with notes on patristic attestations and corresponding manuscript variants, and an argument about literary history: that Marcion’s canon preserves an earlier form of the gospel and of Paul than the Catholic editions. Later discussion sometimes pries those tasks apart. One can use the English text and refuse the argument, or accept the argument and still want a Greek edition with a fuller apparatus. BeDuhn’s 2017 review of Roth drew the same line through Roth’s book. He praised Roth’s caution and charged that the caution, in practice, biases toward the canonical form of Luke. The newest English reconstruction and the newest technical Greek reconstruction were already reading each other as opponents in a fight older than either book.
Roth 2015
Roth’s The Text of Marcion’s Gospel (Brill, 2015) prints a discontinuous Greek gospel with graded confidence. The method follows Schmid. One must study the citation habits of Tertullian, Epiphanius, and the Adamantius dialogue before one prints Marcion’s words. Roth lists verses attested as present, verses attested as absent, and verses unattested. He walks through the gospel first under Tertullian’s multiply cited verses, then under verses cited only in Against Marcion, then under Epiphanius, then under Adamantius, then under the minor witnesses: Irenaeus, Clement, Origen, the Refutation of All Heresies (which Roth cites under the name of Hippolytus), Pseudo-Tertullian, (Pseudo-)Ephrem, Jerome, Filastrius, and Eznik. The reconstruction itself grades confidence. Unattested material is not converted into omission by theology, and silence is left as silence.
The result is a discontinuous Greek text, in the line of Zahn, Harnack, and Tsutsui rather than of Hahn. Bilby’s normalized count is about 4,200 words. Reviews by Judith Lieu, Christophe Guignard, Elijah Hixson, Simon Gathercole, and Michael Holmes made this the reconstruction most often recommended as a base text. Daniel A. Smith based a statistical analysis on it. BeDuhn’s critique, already noted, is the serious opposing review: Roth’s confidence system, by rewarding alignment with Luke and with a father’s ordinary citation habits, may reconstruct a Marcion who looks more like Luke than the sources require. The charge is the old charge against Hahn, now inverted. Hahn filled silence to protect a cut Luke. Roth, on this view, declines to fill silence and thereby protects a Luke-shaped Marcion. Roth’s own claim is that previous reconstructions, including Harnack’s, failed precisely by not distinguishing levels of certainty. The edition is a method book as much as a text.
Roth does not, in the 2015 volume, claim to settle the literary relation the way Harnack did at page 240*. That restraint is one reason the edition could be used by people who disagree about priority. It is also why it could not end the production of rival texts.
Klinghardt 2015–2021
Matthias Klinghardt’s Das älteste Evangelium und die Entstehung der kanonischen Evangelien appeared in 2015, in a second German edition in 2020, and in English in 2021 as The Oldest Gospel and the Formation of the Canonical Gospels. The reconstruction fills the second volume. Klinghardt’s siglum *Ev names not only the text attested by the heresiologists but the oldest gospel in a larger tradition-historical model. In that model *Ev is a precanonical version of Luke and the source for precanonical versions of Mark, Matthew, and John; those precanonical gospels were then edited into their canonical forms. The reconstructed *Ev is, by design, more expansive than the wording actually recoverable from Tertullian, Epiphanius, and Adamantius. Bilby’s normalized count, 12,850 words, approaches Hahn’s maximalism and far exceeds Roth.
Klinghardt returned to a continuous text and supplied a robust critical apparatus. He also returned to the literary thesis of the early Ritschl and the early Baur, now extended beyond Luke to the Fourfold Gospel. Christopher M. Hays had already answered Klinghardt’s 2006 “Plädoyer” in 2008. Michael Wolter’s Luke commentary rejected it. After the edition came critiques by Thomas Johann Bauer, Ulrich Schmid, and Roth, and Pier Angelo Gramaglia’s 2017 Italian translation and critical commentary, Marcione e il Vangelo (di Luca). Gramaglia’s book is not a peer Greek reconstruction, but it is one of the most detailed engagements *Ev has received. Gramaglia granted that the gospel Marcion used is earlier than canonical Luke, and then refused Klinghardt’s larger model: the earlier book is, on his reading, a first edition by the same author who later wrote the Luke we have; Marcion used that book rather than writing it; it is not the source of Matthew, Mark, and John; and the two-source theory does not fall with *Ev. That is a third way between simple priority and simple posteriority, reached in Italian, on a German edition, without waiting for an English translation. Klinghardt’s edition is the most ambitious reconstruction since Hahn. The ambition is the point of dispute. An apparatus that restores unattested wording from Western witnesses and from a synoptic theory will look, to a follower of Schmid, like the creation of evidence. A sparse edition that refuses those restorations will look, to Klinghardt, like an edition of the fathers’ silence rather than of Marcion’s book.
Nicolotti 2019 and the Bertinoro Conversation
Andrea Nicolotti’s reconstruction appeared in Claudio Gianotto and Andrea Nicolotti, Il Vangelo di Marcione (Einaudi, 2019). It is the eighth text in Bilby’s census of major published reconstructions. Nicolotti prints a continuous Greek text with an Italian translation and a critical apparatus and uses different typefaces to mark degrees of certainty, a typographical convention influenced by Klinghardt. His reconstructive method, however, stands closer to Roth’s source-based approach. Nicolotti reconstructs from the surviving sources without making the edition serve a larger theory about the origins of Marcion’s Gospel or the canonical Gospels.
Bilby’s normalized count places Nicolotti at 10,870 words, compared with Klinghardt’s 12,850. That difference describes the relative extent of the two editions; it should not be interpreted as though Nicolotti’s text were simply Klinghardt’s reconstruction with some restorations removed.
The edition also belongs to a sustained scholarly conversation fostered by the annual meetings of the Centro Italiano di Studi Superiori sulle Religioni (CISSR) in Bertinoro. Gianotto, Klinghardt, Vinzent, and Nicolotti participated in these meetings over several years and followed developments in the field together. This setting helps explain the concentration of important Italian work on Marcion without reducing that work to one scholar’s influence on another. Gramaglia’s 2017 engagement with Klinghardt and the 2019 volume of Gianotto and Nicolotti came out of that wider conversation.
Vinzent, Lieu, and Work That Is Not One of the Eight
Markus Vinzent’s Marcion and the Dating of the Synoptic Gospels (2014) and the studies around it belong to the revival of the literary question. Vinzent argues that Marcion’s gospel stands at the origin of the gospel tradition, a claim that goes further than “Luke used Marcion.” He is a major voice in the same conversation as BeDuhn and Klinghardt. He is not, in Bilby’s census of gospel editions, the author of one of the eight published Evangelion reconstructions. Treating him as if he had already issued a standard Greek Evangelion of the Roth or Klinghardt type would invent an edition. Treating him as a side name because he has no gospel edition on the list would invent a smaller field than the one that exists. His 2025 Apostolos reconstruction belongs to the Pauline line, not to that gospel census. The history of reconstruction includes books that change what editors think they are reconstructing.
Judith Lieu’s Marcion and the Making of a Heretic (2015) is the major recent monograph on Marcion as a second-century figure. Lieu uses reconstructions and reviews them; she does not replace them. Her Marcion is recovered from controversy, institution, and the making of heresy as well as from a printed Evangelion. Other writers who have received Knox and Tyson sympathetically belong in the same wider discussion. So does the computational project of Mark G. Bilby, whose Journal of Open Humanities Data datasets and iterative “First Gospel” work try to turn the eight reconstructions into comparable data. Bilby’s own reconstruction, announced as in progress, is not yet one of the eight. The datasets already matter, because they make the old formal choice measurable: continuous or discontinuous, maximal or minimal, Western or reserved.
On the Apostolos line, Schmid remains the technical Greek edition that first made citation habits the control. BeDuhn remains the English collection of 2013. Vinzent’s three-volume Die älteste Sammlung paulinischer Briefe und die Entstehung der kanonischen Paulusbriefsammlung (2025), prepared with Mark G. Bilby, Jack Bull, and K. Lance Lotharp, now prints a continuous Greek reconstruction of the ten-letter collection, with German translation and apparatus. English datasets and a Greek–English reader’s edition have circulated with that work. It belongs on the Apostolos line as a published reconstruction, not as a rumor of one, and it does not erase Schmid, van Manen, or Harnack. Hilgenfeld’s 1852 Galatians appendix, van Manen’s 1887 letter, Zahn’s pages 495–529, Harnack’s Beilage III, Schäfer’s Old Latin lists, Raschke’s Epiphanius catena, and Clabeaux’s A Lost Edition of the Letters of Paul are still the earlier stations of the same road. The apostle was never a footnote that had to wait until the gospel was finished.
Why the New Editions Still Disagree
The eight major reconstructions differ first in format and then in method. Hahn, Klinghardt, and Nicolotti all print continuous texts, but that formal resemblance does not imply a common reconstructive method. Nicolotti marks degrees of certainty typographically and works from the surviving sources in a manner closer to Roth. Zahn, Harnack, Tsutsui, and Roth print texts that look more like dossiers, in different degrees. BeDuhn stands between these forms and writes in English. Behind the formal choices sit the same three questions that have organized this history since Schulz named Hahn’s two faults in 1829.
The first question is silence. May a verse over which Tertullian passes, and which Epiphanius does not mark as missing, be printed as present, printed as absent, or left blank? Hahn answered from Marcion’s theology. Zahn had sometimes filled a case from theology in his first volume; his second volume then invented a notation that distinguished the three cases. Harnack warned against the habit and kept the habit. Knox put silence in an uncertain column and then built statistics on a table that still depended on Harnack. David S. Williams would use only verses quoted by both Tertullian and Epiphanius. Schmid and Roth treat silence as silence. Klinghardt treats silence as a space that a Western Luke and a synoptic theory may fill. The argument of 1829 is the argument of 2015.
The second question is the witness. Is Tertullian a transcript, a translator, a rhetor, or some mixture that can be studied? Hahn took the embedded citation as a quotation. Zahn trusted the sources more than his predecessors had, and still decided some unattested verses from Marcion’s theology. Harnack asserted that Tertullian’s reproduction is reliable, as nearly every page shows, and that one can almost always tell a verbatim citation from a report of the wording. Clabeaux, Schmid, Braun, and Roth made citation habits the first task of the edition. Tsutsui took a verb of saying as a mark of reliability and was told that this was a step backward. An edition is only as good as its theory of how a hostile writer quotes.
The third question is Luke. Every reconstruction is also a relation. Semler, Eichhorn, Schwegler, the Ritschl of 1846, the Baur of 1846/1847, Couchoud, Tyson, BeDuhn, Vinzent, and Klinghardt have, in different ways, made Marcion’s book, or a book close to it, earlier than the Catholic Luke. Hahn, Olshausen, Harting, the Ritschl of 1851, the Volckmar of 1852, Sanday, Zahn, Harnack, Loisy, and Blackman have made Marcion’s book a form of Luke. Roth leaves the relation open, though BeDuhn charges that his method leans toward Luke. Hilgenfeld, the Baur of 1851, and Knox after him stood between those poles, and not in the same place. The new editions have reprinted the old poles with better tools rather than creating a fundamentally new position.
Bilby’s word counts make the reprint visible. Hahn at 14,442 words and Klinghardt at 12,850 are maximal continuous texts. Nicolotti’s 10,870-word edition is a continuous, source-based text whose smaller size should not be described merely as a reduction of Klinghardt’s reconstruction. Zahn at 10,572 is a large discontinuous text. BeDuhn at 7,419 is a text of middle size, converted from English by Bilby’s sampling ratio. Harnack at 4,338 and Roth at 4,169 are sparse dossiers. Tsutsui at 3,296, converted from Latin by a Vulgate ratio, is the sparsest of all. The numbers rank daring rather than quality. Two editors can share a literary theory and still print books of different sizes. Two editors can print books of similar size and still deny each other’s theory. The field’s appearance of chaos is the old map drawn at new scales.
One Unattested Verse
Luke 5:39, the saying that no one after drinking old wine wants new, is a place where the editions can be compared without a later survey. Tertullian and Epiphanius do not quote it for Marcion. Harnack treated the silence as a dogmatic excision: Marcion, on this view, could not bear a sentence that favored the old. Chapter 4 has already named 5:39, 9:56, 22:43, and 24:40 as verses where Harnack turned a lack of citation into positive evidence. Roth’s 2015 edition leaves 5:39 unattested. A continuous edition such as Klinghardt’s cannot leave the place blank. It has to print the verse or leave it out, and give a reason drawn from somewhere other than the fathers. The same silence is therefore an excision, a blank, or a decision made on other evidence, depending on what an editor permits silence to mean. A reader who wants the actual texts can open Harnack’s Beilage IV, Roth 2015, and Klinghardt’s second volume at that verse.
The Fight Recurs Because the Object Requires It
Marcion published a gospel and ten letters. The men who quoted him were not his copyists. Every later editor has had to decide what a quotation is, what a silence is, and what Luke is. Those three decisions generate a text. They also generate an opponent. Hahn’s forensic Luke produced Schwegler’s pairs and Ritschl’s Grundstamm. Ritschl’s Zusammenhang produced Baur’s doubts and Franck’s objection, and Volckmar and Hilgenfeld produced Ritschl’s retraction. Baur’s 1847 priority produced Baur’s 1851 Ur-Lukas. Volckmar’s 1850 mediation produced Volckmar’s 1852 confirmation of the fathers. Zahn’s notation produced Harnack’s slimmer dossier. Harnack’s closed sentence produced Knox’s reopened question. Knox’s German summary produced an English account that Volckmar 1852 and Ritschl 1851 do not support. The editions of 2013 to 2021 then reprinted the old fight at new lengths: Roth’s sparse dossier, BeDuhn’s charge that such caution leans toward Luke, Klinghardt’s overflowing *Ev, and Nicolotti’s source-based continuous edition, typographically indebted to Klinghardt but methodologically closer to Roth. The sequence looks like progress when it is described as a list of better editions. It looks like recurrence when it is described as a list of the same three decisions.
The book has followed the people who already made those decisions, so that they do not disappear behind the next edition. The famous editions do not exhaust the company, and they do not close the argument.
A reconstruction will be needed for as long as the manuscript is missing, and the next will face the same test.
