Chapter 4 of 6

    The Classical Reconstructions

    12 min read

    0:0019:14

    🎧 Listen while you read

    The second half of the nineteenth century did not produce another Tübingen. It added editions, and an English and a Dutch argument that the German monographs had not absorbed. William Sanday tried to end the question with style and diction. Walter Richard Cassels, writing Supernatural Religion anonymously, changed his mind after Sanday attacked him. Hajo Uden Meyboom wrote the first large Dutch Marcion book and then declined to redo the text. Van Manen reconstructed Galatians and reversed the charge of falsification. Zahn rebuilt gospel and apostle together. Harnack rebuilt both again and told the twentieth century that the literary question was closed. The classical reconstructions became the books later people used; the work that made them is larger than those books.

    English and Dutch Work Before Zahn

    Sanday’s first piece was a rejoinder, in the Fortnightly Review for 1875, pages 855–875, to the Marcion chapter of Supernatural Religion. He revised it as a chapter of The Gospels in the Second Century (1876), pages 204–237, and added an appendix, pages 362–372, that set out Luke 5 as the sources attest it. The remainder of the evidence, he wrote at the close of that appendix, could easily be produced if necessary; he did not think it would long remain in doubt that “our present St. Luke was really the foundation of the Gospel that Marcion used.” The German controversy, Sanday thought, had revolved too much around whether Marcion’s text could be explained from his dogmatic system. He believed the dogmatic argument pointed toward the traditional view, and he also thought it should not be pressed too far. He was tempted, he wrote, to call the almost exclusive and certainly excessive use of arguments from the history of dogma “the prime fallacy” at the root of Tübingen criticism (page 221). His substitute, pages 223–230, was linguistic. The style and vocabulary of the passages absent from Marcion’s gospel, Sanday argued, matched those of the passages present in it, and he inferred that the gospel Marcion used presupposed “our present St. Luke, in its complete form” (page 230).

    Cassels gave way in the preface to the 1879 complete edition of Supernatural Religion (volume 1, pages iii–iv). Since the sixth edition, he wrote, he had been convinced that Marcion’s gospel was based on Luke. He did not name Sanday there. English-speaking scholarship largely followed Sanday until John Knox. Knox’s 1939 article, “On the Vocabulary of Marcion’s Gospel,” argued that the proof had a hole. Sanday had examined Luke without reference to Marcion’s actual recoverable wording and had assumed that the pericopes appeared in the same form in both books. What he had proved, Knox said, was the linguistic homogeneity of canonical Luke, which had never been in doubt. On that argument the evidence had no necessary relevance for Marcion’s gospel. The first famous method in English was a method that did not touch Marcion’s text. It still mattered, because it taught later scholarship that a purely theological evaluation of Marcion’s gospel was not enough. Knox later admitted that his own attempt to invert Sanday’s statistics had also asked too much of the recoverable wording.

    Other English attempts stood beside Sanday’s. Sabine Baring-Gould’s 1874 The Lost and Hostile Gospels had already offered the first extensive reconstruction of Marcion’s gospel in English (pages 248–255). A. H. Charteris printed a collection of early testimonies, including a Marcion reconstruction, in Canonicity (1880, pages 393–408). Charles B. Waite gave another in History of the Christian Religion (1881, pages 243–251). Hill’s 1891 English Hahn has already been mentioned. None of them took the argument further than the Germans had in the 1850s. They matter because they show an English-language reconstruction history already running in the 1870s and 1880s, beside Zahn and before Harnack. Readers who encountered Marcion’s gospel in Baring-Gould or Hill were already reading a reconstructed book.

    Meyboom’s Marcion en de Marcionieten (Leiden, 1888) devoted a section to Marcion’s canon, summarized Epiphanius, and listed the Lukan sections over which Tertullian is silent (pages 125–164, especially 128–142 and 153–156). Zahn’s verdict was exact. On the Apostolikon Meyboom relied, by and large, on Hilgenfeld’s 1855 essay; on the gospel he relied on Volckmar; he did not correct the method or use the reports more completely. Meyboom said why. He was writing the character and history of Marcionism rather than doing textual criticism. If reconstruction is defined as a new Greek text, the Dutch book is easy to omit. Yet it transmitted Hilgenfeld and Volckmar to a new language, and it already overlooked the difference between Volckmar in 1850 and Volckmar in 1852.

    Van Manen and the Apostolos Line

    W. C. van Manen’s “Marcion’s brief van Paulus aan de Galatiërs,” published in two installments of the Theologisch Tijdschrift in 1887 (pages 382–404, 451–533), is the first largely continuous Greek reconstruction of Marcion’s Galatians printed as a text in its own right. Van Manen did not reconstruct the Evangelion. He reconstructed the letter with which Marcion’s Apostolos began, and he used that letter to reverse the old charge. It was not Marcion, he argued, who had falsified a Paul transmitted intact by the Catholic Church. A Catholic after Marcion had reworked, in a Catholic spirit, the letter Marcion had preserved in its original form.

    Hilgenfeld had already done the first meticulous analysis of Marcion’s Galatians as an appendix to Der Galaterbrief (1852, pages 218–234) and a later examination of the whole Apostolos in “Das Apostolikon Marcion’s” (1855). Those studies remain, in the words of Bilby, Bull, and Lotharp, “unwieldy hodgepodges of critical commentary, piecemeal Greek reconstructions, Latin attestations, and scholarly cross-references.” They are still the beginning of the Apostolos line as a scholarly object distinct from the gospel line. Van Manen made the object visible as a printed letter. Zahn, whose own Apostolikon text was already complete when van Manen’s work appeared, said that it cost him an effort to add critical remarks against the Dutch book. Those remarks stand in the apparatus to his Galatians (Geschichte des neutestamentlichen Kanons 2.2, pages 495–504), with a general verdict in the note at pages 449–451; together they are the first major collision between a classical edition of both books and a reconstruction of Paul that denied Catholic priority.

    The Apostolos line does not pause for the gospel editions. Hermann Raschke in 1926 arranged Epiphanius’s Greek scholia on Galatians, 1–2 Corinthians, and Romans. Karl Theodor Schäfer in 1939 compiled Old Latin readings for Galatians, including Tertullian’s attestations of Marcion. Clabeaux’s 1989 A Lost Edition of the Letters of Paul collated Marcionite wording against manuscript and patristic variants in order to recover a pre-Marcion Pauline text. Schmid’s 1995 Marcion und sein Apostolos then produced the first methodologically controlled reconstruction of the whole collection, pages 315–344, on the basis of a full study of citation habits. BeDuhn printed an English Apostolikon in 2013. Harnack’s Beilage III and Zahn’s pages 495–529 stand on this line, not above it. A history of reconstruction that mentioned Paul only when Zahn and Harnack printed him would treat the later, stricter work as a sequel. It was a parallel discipline, and for a time it was ahead of the gospel.

    Zahn’s 1892 Edition

    Theodor Zahn gave long sections of the Geschichte des neutestamentlichen Kanons to Marcion: volume 1, pages 585–718, and volume 2, pages 409–529. After summarizing the debate among Ritschl, Baur, Volckmar, and Hilgenfeld, he said that a clear and complete presentation was still lacking. He intended to supply one. The result is a source criticism followed by a method and then by two texts.

    Pages 409–449 of volume 2 examine the witnesses, Epiphanius first, then the Adamantius dialogue, Jerome, the anonymous “Adversary of the Law and the Prophets,” and uncertain reports. Pages 449–455 state the method. Zahn’s hope of advancing beyond Hahn and his successors rested on reports previously neglected or unknown, on a criticism of Epiphanius that eliminated errors and made Adamantius usable, and above all on “a fundamentally different position toward the sources.” The difference applied first to what the witnesses report. Only what is transmitted as Marcion’s text may count as his text until the transmission has been shown to be wrong, and Zahn charged everyone before him with making Marcion’s text as like the customary Luke as possible, even against the reports. It applied to silence as well. The silence of Tertullian or Epiphanius proves neither that a portion of Luke was absent in Marcion nor that it was present, whether in divergent or in identical wording. Zahn described three moves his predecessors had made where the sources say nothing, and he rejected each as incapable of proof. One says that Marcion could not have kept a passage because of his doctrine. Another says that he could have kept it by interpretation. A third says that Marcion never found the passage in his copy and that it was later smuggled into the Catholic Luke, which Zahn called divinatory criticism without historical foundation so long as the absence had not been proved (page 451). At Luke 24:39 Hilgenfeld and Volckmar had restored words against the threefold testimony of Tertullian, Epiphanius, and the dialogue, on the ground that Marcion could well have kept them. Zahn answered in one sentence (page 454): “Yes, he could have done so; but history is to answer the question whether he did so.”

    The notation of the printed reconstruction follows from the rule. A bare chapter and verse number means the passage stood in Marcion and nothing secure is transmitted about differing wording. Round parentheses mark a passage whose presence is not established by a positive report or a secure inference. Square brackets mark a passage whose absence is attested in the same two ways. Greek is printed where Marcion’s wording diverges from Tischendorf’s eighth edition, as far as Zahn thought necessary. A question mark, or a variant in round parentheses, marks uncertainty in the transmission. The reconstruction of the Evangelion occupies pages 455–495; the Apostolikon occupies pages 495–529. The installment that holds these pages was issued in March 1891, and the finished volume is dated 1892. For the first time both books of Marcion’s canon could be read as editions produced by one method.

    Zahn’s result was the fathers’ result. Marcion had found not some text merely related to the Catholic Luke but the Catholic Luke itself, and he had worked it over (volume 1, page 681; compare page 713). The ancient conclusion, Zahn said, had confirmed itself on every side. He also held that Marcion knew and rejected Matthew, Mark, and John. The edition that taught later scholarship how to mark presence, absence, and uncertainty was an edition written to close the literary question in the traditional direction.

    The closure was not as clean as the notation. In volume 1, pages 684–704, Zahn had argued that, given the weakness of Ritschl’s criterion of connection, a more certain proof came from asking whether the differences between Marcion and Luke were better explained by Marcion’s theology or by a Catholic redactor’s. He used that test mainly on attested readings and denied, at page 704, that all differences were dogmatic. He also used it to create evidence for alterations or omissions. Luke 22:39–46, discussed at volume 1, pages 686–687, is one of those cases; further examples stand at pages 706–707. The notation of volume 2 and the argument of volume 1 do not always keep the same rule. Harnack praised Zahn for setting the right principles and completing the work so carefully that later examinations could, on many points, only revise and continue his results (Marcion, page 41*). In the same place he complained that the impractical form of Zahn’s presentation made it extraordinarily difficult for a reader to get to know Marcion’s text of the letters. Gerhard May’s later judgment is sharper and fairer: Zahn overtook the older research, and his restoration, like Harnack’s, is still not critical or methodologically strict enough. The classical edition became classical because later scholars built on it, while the method remained unfinished.

    From Zahn to Harnack

    Hermann Freiherr von Soden put Zahn’s reconstruction to work in his New Testament. He took Marcion’s gospel as a mutilated Luke and concluded that by about 140 Luke was already being read in Asia Minor or Rome in the I-H-K text. August Pott contested the second conclusion. Marcion, Pott argued, had a “Western” text before him, and many readings previously called tendentious were simply Western readings. Harnack supported the Western diagnosis without accepting all of Pott’s assignments. Heinrich Vogels, reviewing Harnack, remained unconvinced that Marcion had a Western text or that Marcion was innocent of stylistic changes now attested in the Western tradition. The classical period was already a text-critical period. Reconstruction had become a way of dating Luke’s manuscript history, not only a way of deciding which book was first.

    Adolf von Harnack’s Marcion: Das Evangelium vom fremden Gott appeared in 1921 and in a revised second edition in 1924. The reconstructions sit in the Beilagen. In the 1921 printing the Apostolikon is Beilage III and the Evangelion is Beilage IV. The 1924 edition, which scholarship cites, revises and re-paginates those appendices; the Apostolikon text begins at page 67*, the Evangelion at page 183*. Harnack’s 1923 Neue Studien zu Marcion, thirty-six pages, is a supplement to the monograph, not a third reconstruction. The 1870 Dorpat prize essay, published much later from the manuscript, shows that Harnack had been living with the older scholarship, including Hahn, Ritschl, Hilgenfeld, and Volckmar, for half a century before the famous book.

    Harnack adopted Zahn’s discontinuous approach and many of his indications, then pared Zahn down. He used nearly every available source. At page 45*, introducing the Apostolikon, he asserted that Tertullian’s reproduction of Marcion’s text is reliable, as nearly every page shows, and that one can almost always tell a verbatim citation from a report of the wording. He did not demonstrate how that discrimination is to be made. He said, more usefully, that what Marcion left in the text matters more for Marcion’s thought than the omissions (Marcion, page 66), and that for a large number of passages it remains unclear whether Marcion excised them or his opponents simply passed them over. Those are the right cautions. The printed text does not always obey them. Ulrich Schmid later put the rule Harnack needed: when a source does not cite or evoke a passage, it does not cite or evoke it. Arguments from silence that turn a lack of evidence into positive evidence, even when the alleged omission would match Marcion’s supposed preferences, should not be allowed. Harnack still created such evidence at Luke 5:39, 9:56, 22:43, and 24:40. The first, third, and fourth of those verses are unattested in the sources; Harnack treated them as dogmatic excisions. Luke 9:56 is an unattested element that he treated as a Marcionite addition.

    He also reconstructed without a sufficient study of how his witnesses cite scripture outside the anti-Marcionite books, and he compared Marcion’s wording to Tischendorf’s apparatus after von Soden’s material was already available. Pott observed that many readings Harnack called unique or unattested would not have looked that way in von Soden. Vogels, reviewing the second edition in 1925, listed nearly one hundred and fifty verses whose apparatus he judged deficient. Harnack granted the defect, declined to work with von Soden, and published a short Verwahrung in the Theologische Literaturzeitung in 1926. The classical reconstruction was already being argued over as a text-critical instrument, not only as a literary verdict.

    The sentence that closed the literary question for a generation stands at page 240* of the 1924 edition. Marcion’s gospel, Harnack wrote, is nothing other than what the ancient church said it was, a falsified Luke. “No more words need be lost on the subject.” Harnack also held that Marcion’s Luke, apart from Marcion’s alterations, was a pure Western text; that it already showed the influence of Matthew and Mark, probably because Marcion had received a harmonized copy; that Marcion’s influence on the Catholic text was very small; and that Marcion knew the other three gospels and justified his rejection of them in the Antitheses. The reconstruction and the history of the canon were one project. Once the project was printed, it became, in Helmut Koester’s later sentence, the work on which “all further research is based” (History and Literature of Early Christianity, page 329).

    Harnack’s English afterlife has been uneven. John E. Steely and Lyle D. Bierma translated the main monograph in 1990, not the Beilagen. The reconstructions themselves, as running English scholarly editions, are later. Work on the same problem continued after Harnack in English, French, German, and Italian, and in Japan.