Harnack knew what Marcion’s gospel was, and his confidence is the problem. Roth quotes him saying that one would no longer need to spend even a single word on Marcion’s gospel being anything but “a falsified Luke.”
Our position is the opposite of his, and a reader should have it in hand before the assessment begins. The Evangelion is the gospel Marcion received and published, and canonical Luke is the later expansion of it rather than its source. What follows is an account of how much of Harnack’s work still stands once that base is removed, and a surprising amount of it does.
The Model Harnack Worked In
For him the Evangelion is canonical Luke with deletions, a handful of additions, and a set of tendentious corrections, resting on a text-type he calls the W-text. He states it flatly on his twenty-sixth page: “the observation that M.’s text, apart from the tendentious corrections, is simply a W-text (also in its harmless individual readings) has likewise gained acceptance.” Everything else in Part IV of the pamphlet, and much of the Addenda, is built on that base. Marcion is an editor working on a text we possess.
His skill inside that model is remarkable and should not be patronized. He is the one who saw that Tertullian had before him a Marcionite Latin New Testament, which he rightly calls an important new insight for the history of the canon and the text. He ranks the Old Latin witnesses for their closeness to the Marcionite text, placing the Palatine codex first and then Veronensis, Colbertinus and Vercellensis, a judgment of 1923 that I have not checked against current assessments. He notices the influence on Sinaiticus. And when Bauer and Lietzmann press him, he concedes on his twenty-seventh page what a defensive scholar would not: that he “might have emphasized, or established, the influence of the Marcionite text upon the Church text somewhat more strongly,” both beyond the W-text and within it. His closing demand on his twenty-eighth page (that the Western text be restored for Luke and Paul, quoad fieri potest, from the Old Latin and Marcion’s text, so as “at last [to] put an end to the preferential treatment of B ℵ Origen”) is a research programme that later scholarship has partly carried out.
He also records, on his twenty-ninth page, the observation that matters most for us: “so far as we know, Marcion was the first to call a book ‘the Gospel’.” Whatever else is disputed, the noun is his.
And there is one further thing to his credit that cuts against his own model. On his twenty-second page, in a footnote, he refuses the argument that Marcion’s rejection of John proves the fourth gospel had no standing in his circles, and he insists, “against Eichhorn and Bauer”, that John would not have suited Marcion better than Luke, since in his judgment John’s general features “blossomed up out of Judaism” and would have repelled him. That is Harnack declining an easy point.
What Has Changed Since 1923
Now the difficulty, and it is the largest single change in the field since 1923.
The direction of dependence between the Evangelion and canonical Luke is no longer settled. Four bodies of work have unsettled it in different ways, and they should not be run together.
Ulrich Schmid did the methodological work, in Marcion und sein Apostolos (1995), on the Apostolikon rather than the gospel. His contribution is a discipline rather than a thesis: patristic citations must be assessed as citations, with attention to how a polemicist quotes, before they are treated as evidence for a manuscript. He stated the rule most sharply in a later paper: “Arguments e silentio, creating positive evidence out of a lack of evidence, should not be allowed, even if the alleged omission would match supposed theological preferences of Marcion.” That transfers directly to the gospel. A reading is not attested for Marcion merely because Tertullian argues as though Marcion held it.
Jason BeDuhn, in The First New Testament (2013), begins from the premise that Marcion collected rather than composed: that he took up texts already in circulation and gave them back in collected form. That changes what counts as a variant, and it leads him to a conclusion that must not be confused with the priority thesis below. He holds that the differences between the Evangelion and canonical Luke are better accounted for on non-ideological grounds. Here he revives a suggestion of Semler’s: the two are alternative versions, shaped by the distinctive needs of the Gentile mission. On that account both stand behind Marcion, and neither is the parent of the other.
Dieter Roth is the necessary brake, and his discipline is the most useful thing here for anyone tempted to overreach. In The Text of Marcion’s Gospel (2015) he sorts the evidence into three classes rather than two: verses attested as present, verses attested as not present, and verses simply unattested. If Tertullian nowhere quotes a verse, that belongs in the third class and not the second, because Tertullian quotes what serves his argument. And the deeper methodological point is the one that bears on Harnack. Reconstruct Marcion’s gospel by starting from canonical Luke and asking what has been removed, and you will produce a mutilated Luke. That is what the method is built to produce.
Matthias Klinghardt argues the strong thesis, in Das älteste Evangelium und die Entstehung der kanonischen Evangelien (2015): that the gospel behind Marcion’s is prior to canonical Luke, and that the canonical gospels are to be explained partly as responses to it. Markus Vinzent argues a different thing in Marcion and the Dating of the Synoptic Gospels (2014), and the two must not be merged. His is a dating argument about when the canonical gospels became available and authoritative, and it would reposition Marcion in the sequence even if Klinghardt’s textual case failed. Neither is consensus. Both are serious.
And Judith Lieu stands behind all of it with the warning that applies to every reconstruction, Harnack’s and Klinghardt’s alike. We know this text through writers who were arguing against it. Their literary interests shaped what they reported, and their portrait of the man has a life of its own.
The Position Defended Here
Here is the position I think can be defended, and I want to be careful about its limits, because two claims are regularly confused and only one of them is safe.
“The Evangelion is earlier than canonical Luke” and “the Evangelion is not simply a mutilation of canonical Luke” are not the same claim. The first is a live hypothesis with able defenders and able opponents, and the True Testament does not need it. The second is much better founded, and it is enough. The Evangelion is an early second-century gospel text; it must be handled as a textual witness in its own right; it cannot responsibly be described as a mutilated copy; its relationship to canonical Luke remains open; and a great many individual readings remain uncertain. That is a defensible historical position and it does not require anyone to have won the priority debate.
Take five cases where Harnack committed himself, and see what the witnesses actually bear.
Nazareth, and Luke 4:16. Harnack’s reconstruction on his thirty-second page is his showpiece. He argues that Marcion’s text omitted τεθραμμένος, “brought up,” and αὐτῷ, “his,” so that the opening reads “And having come to Nazareth, where he was, he entered, according to custom, on the day of the sabbath into the synagogue”, with the custom no longer Jesus’s own. He calls it “a cabinet-piece of Marcion’s textual criticism,” reasoning that Marcion could not allow Jesus to have been reared in Nazareth or to have attended synagogue by habit. The argument is coherent and the motive plausible. But he arrives at the reading partly by asking what Marcion would have needed to remove, which is the method Roth warns against. The True Testament text here reads “Jesus came to Nazara, where he went into the synagogue on the Sabbath, just as was the custom” (Evangelion 1:7 TT): impersonal custom, no rearing. That is essentially Harnack’s reconstruction, and it should be labelled as such: a reconstructed Marcionite reading standing in the True Testament text, not an independent ancient witness to it.
The old wine, and Luke 5:39. Harnack argues on his thirty-third page that the saying “no one who has drunk old wine wants new; for he says: the old is good” is absent from Marcion, absent from a group of Old Latin witnesses and Eusebius, and present in all the rest; therefore genuine, and deleted by Marcion “for very understandable reasons.” The True Testament text is consistent with that: the sayings about the wineskins and the unwashed patch stand (Evangelion 3:4–6 TT) with no praise of the old. But this is one of four verses Roth singles out (with Luke 22:43, 24:12 and 24:40) where the Marcionite text is in fact unattested and where Harnack nonetheless concluded deliberate excision, and Roth’s verdict is that “at several points Harnack is guilty of precisely such a ‘creation of evidence’.” Schmid, writing on this very verse, put it more bluntly: “This is simply creating positive evidence (in this very case positive negative evidence) out of no evidence at all.” The inference from silence to deletion assumes what is at issue; and if the shorter form is prior, the same evidence shows the saying being added by those who wanted the old defended.
“Father, forgive them,” and Luke 23:34. Here Harnack argues the other way, and this case needs the most care. He holds that the saying was added by Marcion and then penetrated the Catholic manuscripts, and in the middle of his textual note he supplies a reading: the sentence “now acquires a pregnant sense: the henchmen of the world-creator can know nothing of the good God.” That gloss is Harnack’s interpretation and nothing else. It is not in the verse, which reads simply “Jesus said, ‘Father, forgive them, for they don’t know what they’re doing.’” (Evangelion 15:32 TT). Whether the saying is Marcionite in origin is a real question, and the manuscript evidence is genuinely split. Whether it means what he says it means is a separate question, and the answer does not follow from the textual evidence at all.
Peter at the tomb, and Luke 24:12. Harnack argues that the verse is well composed in Luke’s manner, cannot be attested for Marcion, is otherwise missing only from a small group of witnesses, and was therefore deleted by Marcion, “who did not want Peter here.” The True Testament text has the women return and the disciples disbelieve, with no Peter running (Evangelion 16:13–14 TT). But this is the clearest of Roth’s four cases. Harnack’s own words give the game away: the verse “cannot be attested” in Marcion, which is a statement about the witnesses’ silence and not about their testimony, and he then treats the inference as a result.
The man working on the sabbath. On his last page Harnack raises a possibility about a story he reports as standing only in Codex Bezae, between Luke 6:4 and 6:5. It may be originally Lucan; Marcion may have deleted it; and the deletion may have succeeded almost completely, even in the ecclesiastical Bibles. His three reasons are good ones. The story is synoptic and most like Luke in style. It would have been intolerable to Marcion, since it asserts the law’s continuing validity within limits. And the very completeness of the excision is paradoxical but explicable, since the saying would soon have embarrassed Catholic readers too. This is Harnack at his most attractive and also at his most speculative, and he says as much by framing it as a possibility rather than a finding. It remains a possibility.
What comes out of those five is a consistent picture of the method rather than of the text. Harnack’s individual judgments are frequently acute. His procedure asks, at every point, what an editor working on Luke would have done; and that question cannot generate any answer except an edited Luke.
Which leaves the question a confessing reader will actually ask: if this much is uncertain, on what footing does the True Testament Assembly receive the Evangelion as scripture?
Not on the footing that the textual questions are closed. They are not, and a confession that needed them closed would be hostage to the next monograph. And not on the footing that Marcion said so: that is circular, and it would make the authority of the text depend on the authority of a man about whom we are told mostly by his enemies.
The footing is this. The Evangelion is the gospel that was transmitted together with the Apostolikon, as the first part of a two-part scripture, and it is in that pairing that we receive it. It proclaims a Father who was not previously known and who is known only through the Son. And its textual identity can be studied honestly, with the uncertainties named, without the confession collapsing, because what is confessed is the Father whom the text proclaims, not a theory about the text’s stemma. A reading may be reconstructed, secure, disputed or lost, and the proclamation does not change. That is the ordinary condition of every scripture anyone has ever read; it is only unusual here because the reconstruction is recent enough that we can watch it being done.
Three Distinctions
Three distinctions follow, and this afterword will keep them. A reading may be attested for Marcion by a patristic witness. It may be reconstructed by modern scholars, Harnack’s or anyone’s, from a mixture of attestation and inference. Or it may be a True Testament editorial decision, made in preparing a readable text. These are three different things, and the True Testament text is not a manuscript witness. Where this afterword quotes the Evangelion in a disputed place, it will say which of the three it is quoting.
The Evangelion never stood alone. It was the first half of a collection, and the second half was ten letters of Paul in a particular order. Whether that order is an argument is the question that follows.