I recently read a new article by David Allen, “A Model Reconstruction of What Josephus Would Have Realistically Written about Jesus,” in the Journal of Greco-Roman Christianity and Judaism 18 (2022), which tries to argue that Josephus really did write ‘something’ where we now have the Testimonium Flavianum but that it has to have been a different thing than is there now (and he elaborates on his theory in subsequent articles, but this is the only article really defending it). This of course has been speculated many times before, and never set on any foundation of evidence. It’s usually just a ‘made up’ notion, of some lost text for which there is no evidence, an apologetic to get back what is lost when admitting Josephus never wrote about Jesus. Allen attempts to fix this by assembling all the arguments he can think of for it. That makes it a useful reference point. Unfortunately, it’s riddled with fallacies. Christians also won’t like it because in result he paints Jesus as an armed militant.

The evidence overwhelmingly shows that Josephus never wrote anything at all about our Jesus. I already catalog all the evidence for this in Josephus on Jesus? Why You Can’t Cite Opinions Before 2014, so I won’t reiterate that here, except to point out that Allen ignores almost all of it. I had there made the point that we have to stop ‘ignoring’ all the results of peer-reviewed scholarship on this question after 2014 and acting as if opinions prior to that year can be at all informed. They can’t have been—because scholars of the past won’t have had access to those new findings, and therefore their opinions are literally uninformed. This was not their fault, because they cannot have had time machines or psychic powers. But for a scholar now to be ignoring it is their fault. It is a fundamental methodological failure that negates their entire argument. This is what Allen does.

Let’s go through it…

Why Are We Ignoring the Entire Science of Literary Forensics?

Allen’s first mistake is to ignore the entire science of literary forensics, and complain that forgeries should never be hybrid texts mixing the style and vocabulary of the forger and their target (114–15). Exactly the opposite is the case. Virtually all forgeries are hybrid texts mixing the style and vocabulary of the forger and their target. This is because forgers always try to emulate the author they are faking a text from; but because it is impossible to succeed at that (because the stylistic signatures of authors are extraordinarily complex semantically and discursively), they always fail to some extent. We thus detect forgeries by catching those failures amidst the successes. All forensic forgery detection is based on this principle.

I discussed this mistake already when it was made by Steve Mason in Mason on Josephus on James. Like Allen, Mason inexplicably assumes forgers do not emulate the vocabulary of their target text in any respect. In fact ‘mixed texts’ (a verbal and stylistic hybrid of the forger and the target text) are the norm for forgeries, not the exception. This is discussed throughout the work of Don Foster, Author Unknown, Joe Nickell, Detecting Forgery (pp. 104–07), and Kenneth Rendell, Forging History: The Detection of Fake Letters and Documents. They all point out the difficulty of just ‘telling’ that someone wrote or forged a text by its vocabulary or style, because forgers seek to emulate it. What these forensic experts then document is not that forgers fail entirely at it, but that they will inevitably, statistically, fail at some of it—thus resulting in a hybrid text. Hybrid texts are therefore evidence of forgery, not “editing.” Yes, editing is a possible explanation. But you need particular evidence to establish that. And the evidence all points the other way in this case. Indeed there is much more evidence in this respect than I will mention here, yet Allen ignores all of it. So one does need to review my survey in Josephus on Jesus? to see all the evidence Allen ignores. You will only get a sample here.

But to illustrate the point, consider the forgeries of Pauline letters. See, for example, the study of Bart Ehrman in Forgery and Counterforgery (who demonstrates this many times throughout; but on the Pauline forgeries specifically, see pp. 155–222 and 425–32). Ehrman notes it in discussing his first example, 2 Thessalonians, remarking that “it is easy to take over the words of another writer” (p. 158). Indeed it is. As is proved conclusively when he closes with the example of 3 Corinthians, which no one doubts is a forgery, yet its stylistic and verbal similarities with the authentic Corinthian correspondence “abound,” indeed “a large number” of “Pauline words and phrases are scattered throughout” it (pp. 428–29). Allen’s logic would fail on 3 Corinthians; he would end up arguing Paul must have written “some” of it. But that’s bollocks. He obviously wrote none of it. Allen’s method is therefore invalid.

It should also be added that one of the main occupations of ancient education at the composition level was to train authors to emulate another’s style. This is discussed throughout studies on ancient education, from Whitmarsh’s Greek Literature and the Roman Empire: The Politics of Imitation and Cribiore’s Gymnastics of the Mind, to Brodie’s review of ancient literary mimesis practices in The Birthing of the New Testament (pp. 2-79) and the classic anthology Creative Imitation and Latin Literature. Their targets of emulation were shallow (and thus detectable), like mere replication of vocabulary or idioms (and not precise valencies or statistical patterns), but this is exactly what Allen is illogically allowing himself to be duped by. It is hard to fathom how this error, and its neglect of all applicable science and ancient literary practice, would pass peer review. But there it is.

I will pause to note that exceptions here even prove the rule. For example, the style deviations in the Long Ending of Mark from the rest of Mark are unusually thorough and extreme. I discuss the evidence in Hitler Homer Bible Christ, but the result is that this is (a) not normal (rarely are forgeries so close to the target text so divergent from a target’s style) and (b) indicates the passage was not a forgery. I summarize the scholarly theories that the Long Ending was actually a piece of a Four-Gospels commentary that came to be scribally inserted after Mark because its harmonizing summary of all four Gospels solved the disappointing lack of an appearance narrative there. Still a deception (the culprits were passing it off as Mark) but not by forgery (these culprits did not compose the text; and since it was not composed for this purpose, it made no attempt to emulate Markan style). Another example is the Correspondence of Paul and Seneca, which was forged in Latin, with considerable distance from any original text by either author (e.g. this forgery is centuries after the fact and none of it was inserted into genuine collections of either author), abrogating the need to emulate Paul’s style in Greek and diminishing the ability to emulate Seneca’s style in Latin, and yet even this attempts to match the style of Seneca and Paul, and only did so poorly; it does not fail entirely (for examples see the analysis of Ilaria Ramelli).

Allen ignores all this. Yet it destroys the first three pages of his article.

Why Are We Ignoring Structural and Discourse Analysis?

Allen is aware of the crucial work on this question by Hopper and Goldberg (citing them each once). Which is strange. Because he completely ignores their arguments, even though they refute his entire thesis. You would think an author wishing to prove a point would notice this and thus compose some sort of defense against this fact. Instead, Allen doesn’t even mention it. So readers who don’t actually check these studies won’t know what they actually prove and why it is devastating to Allen’s thesis. This, too, should not pass peer review. But again, there it is.

I have covered all of this before. G.J. Goldberg conclusively demonstrated that the entire Testimonium was composed in emulation of the Emmaus creed in the Gospel of Luke (see my previous discussion and my followup in Goldberg’s Attempt to Rehabilitate the Testimonium Flavianum). This includes the ordering of every element across the entire paragraph and the presence of every sentence. Goldberg’s analysis thus entails no element of the Testimonium can have preceded the construction of it entire. So the thesis that Josephus wrote part of it is decisively excluded by Goldberg’s results. Allen seems not to notice this. He certainly provides no response to it. I have myself made a similar demonstration, showing how removing every semantic component of the text that violates Josephan beliefs and discourse style leaves no sentence left to have been written by Josephus, in any form (Historicity of Jesus, pp. 332–37). Allen never addresses this either.

Even more conclusively, Paul Hopper demonstrated that no component of the Testimonium matches Josephan discourse style. As I summarized before: it is not the way he uses verbs; it is not the way he tells stories. It matches, rather, the way Christians do. Its structure is fundamentally credal (as also demonstrated by Goldberg), and therefore not Josephan. My own analysis of sentence placement (which I cited above when expanding on Goldberg) anticipated and reinforces Hopper’s analysis. I will expand on that even more in my next book with the following corollary observation: we actually have several examples of how Josephus writes about Jewish sects and messianic figures, and all of them conform to a common structure—none of which can be found in the Testimonium.

• Common themes across Josephus’s accounts of messianic figures (The Samaritan; The Egyptian; John the Baptist; Theudas; and a more general accounting of ‘imposters’) include: Josephus conspicuously avoids ever using the word “Christ” (i.e. Anointed, i.e. Messiah) or “Jesus” (i.e. Joshua, i.e. the mythical founder of Israel), yet depicts them all as promising or alluding to an apocalyptic redemption of Israel through an act deliberately reminiscent of the original Joshua. This has long been proved in the literature: D. Mendels, “Pseudo-Philo’s Biblical Antiquities, the ‘Fourth Philosophy’, and the Political Messianism of the First Century CE’,” in The Messiah; Craig Evans, “Josephus on John the Baptist and Other Jewish Prophets of Deliverance” in The Historical Jesus in Context; and Rebecca Gray, Prophetic Figures in Late Second Temple Jewish Palestine: The Evidence from Josephus. Yet the Testimonium follows none of these patterns. Nor does Allen’s proposed replacement for it (which invented alternative is already fallaciously circular).
• Common themes across Josephus’s accounts of Jewish sects include: a description of some of their beliefs and customs and how they differ from the other sects’ beliefs and customs; an accounting of how they behave towards insiders and outsiders; and an accounting of how they differ in doctrines regarding the law, fate, the soul, and the afterlife. Nowhere in the Testimonium does Josephus do any of this in describing the Christian sect. Josephus also outright says there are ‘three sects’ in the Wars, and in the Antiquities a ‘fourth’ (militants not referenced here even in Allen’s proposed version) in addition to the three; Josephus apparently had never heard of anyone attributing to him an account of a fifth.
• Contrast that with Josephus’s account of John the Baptist, which does not situate John within any sect, whereas in the Testimonium the entire narrative is about the origin of a sect (the “tribe named Christians”). Even Allen’s fabricated alternative text (for which there is no evidence) establishes it as a narrative of a sect (“A teacher of men who worship him with pleasure” such that “those that followed him at first did not cease to worship him, their leader in sedition, and this tribe has until now not disappeared,” p. 42). By contrast, Josephus only describes John the Baptist as an inspirer of Jews generally, not as a novel teacher assembling a new group of distinct followers (much less as continuing after his death). And yet even then Josephus describes what John taught. The very thing we have no sign of Josephus doing for Jesus. Allen suggests maybe some material was lost here, but “maybe” does not get you to “probably.” You need evidence that that was here. You can’t just fabricate what you need and then claim that what you made up is evidence of itself.
• The text of Josephus happens to follow the Testimonium with another narrative of a religious controversy (a conflict with Isis cult in Rome), which illustrates every way Josephus would have written the Testimonium but didn’t (which I already pointed out in Historicity, p. 336): it meets all the standards of Hopper’s discourse analysis, including its employment of verbs and its clear exposition and causal narrative: we are told what things are, not merely that they are, and why they lead sequentially to the next thing, not merely that they do. The same is true of his other accounts of sects, and his other accounts of messianic pretenders. (Note that Josephus vagues up the chronological sequence by saying this Jewish/Isis cult scandal happened “around the same time” as Pilate’s disastrous actions when really it was roughly ten years before Pilate was even in Judea; but as Allen agrees, p. 123, Josephus was here mashing together events that were thematically related and not following a strict chronological order.)

Hopper’s analysis (especially reinforced by my own observations above) entails that no available reconstruction of the Testimonium—anything that could remain after removing what are supposed to be the Christian parts—can have been written by Josephus. In fact all proposed reconstructions violate every element of his discourse style more than the extant Testimonium already does, producing an account unlike how Josephus ever tells stories. This impels Allen to try and fabricate an entire lost narrative for which there is no evidence, just to try and rescue the hypothesis that Josephus must have written something, when it is far easier and simpler to just admit he didn’t write anything. But even what Allen fabricates (p. 142) violates every element of Josephan discourse style as demonstrated by Hopper (and by the ensuing story about events in Rome). It simply does not resemble the way Josephus writes or tells stories anywhere else.

The simplest (the least ad hoc) explanation is that the previous “sad calamity that put the Jews into disorder” Josephus references at the top of Antiquities 18.65 is the one he concludes with in 18.62. There was no material in between. At all. Exactly as we can tell from the works of Origen, who scoured the text of the Antiquities for anything to address about Jesus—yet never had any clue any such passage was here (not even of some other kind). Allen tries to “fix” this by fabricating the line that “many of his followers, the Galileans and Judaeans, were slain and thus repressed for the moment,” an egregious re-edit from already-suspect material in Tacitus that was never attributed to Josephus (at all, much less as appearing here). But there is no evidence of this. For example, if 18.65 had begun with a reference to Galileans, which is missing from the prior event described in 18.62, then we’d have evidence there was a reference to Galileans in between, which would indeed be evidence for at least something like what Allen is proposing. But no such evidence exists. And you can’t just declare things probable for which there is no evidence.

Put all this together and the idea that Josephus wrote anything here is literally the least likely hypothesis. Trying to get out of this by gerrymandering a Cartesian Demon that threads the needle of evidence to somehow get all these damning facts still to be true and there still be something here about Jesus that Josephus wrote, is a fundamental violation of logic, a rhetorical tactic essential to apologetics, but destructive of real scholarship. You cannot make improbable evidence probable again by inventing an even more improbable hypothesis. You still end up with an improbable theory. Which means your theory is probably false.

Why Are We Ignoring the Entire Science of Textual Criticism?

Allen’s next illogical move is to complain that when Ken Olson documents the passage to be far more Eusebian than Josephan in its vocabulary, he uses Eusebius’s invention of a speech for Licinius as an example of his dishonesty, when (as Allen puts it) everyone knew ancient speeches had to be invented (p. 117). But that isn’t what Olson argues (in the 2013 anthology Eusebius of Caesarea: Tradition and Innovations). Olson’s point is not that Eusebius faked a speech. His point is that Eusebius lied about it. And he did so to authenticate a document (functionally, an affidavit) that he forged. Even Allen admits this: Eusebius invented “fake witnesses” to authenticate his Licinian speech text (p. 115). That is exactly what we are saying the Testimonium itself is. It first appears in the historical record when (coincidentally) it serves to authenticate a narrative Eusebius was promoting (three different times, across three different books!)—it’s just this time, his fake witness is Josephus. Eusebius does this a lot (see How To Fabricate History: The Example of Eusebius on Alexandrian Christianity). So we should not act surprised.

What are the odds that a passage that only first appears in the record as ever being in Josephus just happens (!) to contain vocabulary and stylistic elements distinctive of the very same author who “discovered” it? Allen believes in coincidences. We do not. Unlike Olson I will allow that Eusebius may be the dupe of his tutor Pamphilus—we cannot rule that out because Eusebius could have acquired his style from his tutor, who had all the same motives and proximities, and we have no writings from Pamphilus to compare so as to rule him out as the forger—but it has to be one or the other. There is no other plausible explanation for how the content came to be the way it is—as even Allen admits, since he has to allow much of our version is still a Eusebianesque fabrication. And if Eusebius or Pamphilus could fake up half of it, why not all?

That is why we need evidence to conclude any of it was there before they got to it. And that evidence has to explain why Origen never heard of it; why the passage does not fit structurally where it has been put (with the following story referring to the previous one as if this one wasn’t there); and why none of it fits any aspect of Josephan discourse or storytelling style; and it has to do all that without being so ad hoc and convoluted as to be a priori improbable. It is not enough to invent an excuse as to why it deviates considerably from Josephan vocabulary yet still makes a go at emulating it (“Maybe only bits of it were forged!?”). That doesn’t explain any of this other data, which is all very improbable on that thesis. Allen is ignoring all this (just as Goldberg did).

Allen’s consistent violation of logic continues when he advances the notion (as if a fact) that a single manuscript variant rendering “Jesus” with “a certain Jesus” (Codex A, from the 15th century) is therefore ‘the original text’ and so we can hang his entire convoluted hypothesis on that premise (p. 115, 126–29). This is illogical for two reasons.

First:

The evolution of the Arabic and Slavonic versions of this text demonstrate that later scribes were attempting to make the passage sound less obviously Christian. This therefore rules out any assumption that such variants must be original. It is a standard principle in textual criticism to follow the lectio difficilior, the “most difficult reading.” The idea is that it is unlikely that a scribe would make a text more bizarre or confusing; they are more likely to “help” the text by making it say something easier or more suitable. This is actually not always true (we have many counterexamples; the Latin poem the Aetna is rife with them; for the New Testament alone, the Swanson Guides are full of them; and I’ve come across more examples across ancient literature than I can remember). So it’s more like a rule of thumb or a “criterion” to tick the box of to move the meter a little. But it doesn’t even apply here: fixing an embarrassing text by making it sound more plausible from a Jewish author is not a lectio difficilior. It’s the lectio facilior. We actually expect this to happen.

Allen therefore has the logic here exactly backwards. And this is proved by the Slavonic text of the Testimonium (as difficilior a lectio as can ever be!) but also by the infamous “Arabic” text, which for a long time had been used as proof that an original (pre-Eusebian) text read “he was believed to be the Christ” instead of “he was the Christ.” That has since been conclusively proved to actually derive from Eusebius and therefore post-dates him (see The End of the Arabic Testimonium). Someone changed the text to read that way after Eusebius. The scholar who proved this, Alice Whealey, didn’t like that conclusion and so she still insisted that this reading nevertheless was the earlier one, and therefore even Eusebius’s texts must have once read that way. But that commits an even more egregious error in textual criticism, the very same one Allen is with the Codex A variant: it is far less probable that some cabal conspired to remove this word from every manuscript of this and every other book Eusebius repeated the Testimonium in, and every manuscript of the Antiquities as well—and across the whole of Europe and the Middle East no less—than that a single scribe tried to improve just one manuscript of one book in one place.

Remember, Eusebius wrote this passage down in three completely different books, each of which will have had countless manuscripts circulating. And it was circulating on its own in a fourth book: the Antiquities itself. How, then, can Allen explain this word being missing from literally all of them, other than by a massive conspiracy theory? Once again, Allen’s is literally the least likely hypothesis. As is Whealey’s, because it depends on the same violation of the logic of probability. This isn’t like one Gospel, for example, where a single original reading can survive in one late manuscript simply by its one unique textual tradition avoiding alteration (while all the others derive from the reformed text). Rather, this would be more like three identical verses in three separate books (like Mark, Matthew, and Luke) that were written by the same author (unlike Mark, Matthew, and Luke) quoting another author (as if we also actually had Q, for example, and this same verse was in it, too, and thus replicated across four distinct books), all agreeing on how the verse reads—except one, single, late Medieval manuscript of only one of them. The probability that that was the original reading (and not a later corruption) would be extraordinarily low. You would need quite a considerable basis of evidence to the contrary to render such a notion probable again. (Honestly, this is textual criticism 101 here.)

As I wrote last time for the Arabic reading:

If one single manuscript of the Church History of Eusebius was emended in the Greek to “was believed to be,” perhaps by a scribe who assumed “he was” was surely a mistake, and that manuscript (or a descendant of it) was used by Jerome to make his translation, and a descendant of that manuscript was used to emend the Syriac translation [of Eusebius] in the 5th century (whose author we already know changed several other things from the earlier Syriac), then all the evidence is explained by positing only a single textual corruption in a single manuscript of a single book (the Church History of Eusebius), and one emendation based on it (in a translation we already know was emended in several ways).

That’s vastly more probable than what Whealey’s thesis requires: that Eusebius wrote “he was believed to be” because Josephus wrote that; and then someone altered a later manuscript of the Church History of Eusebius to say “he was,” and someone altered a later manuscript of the Theophany of Eusebius (which also quotes the Testimonium, and all manuscripts of which read “he was”), and someone altered a later manuscript of the Demonstration of the Gospel by Eusebius (which also quotes the Testimonium, and all manuscripts of which read “he was”), and someone altered a manuscript of the Jewish Antiquities of Josephus to say the same, and someone altered a manuscript of the 5th century Syriac translation of the Church History to say so as well, and by chance coincidence all these altered manuscripts are the only ones to have descendants today for us to consult, and yet somehow Theophilus or Michael had access to an unaltered textual tradition of that same 5th century Syriac translation, and yet somehow no surviving manuscripts attest to there being such a tradition. The odds against that are almost as astronomical as odds can get in mundane earthly affairs. Obviously, two guys centuries apart thinking up the same emendation is, even by itself, millions of times more likely. That they both relied on the same singular emended textual tradition, even more so.

So there simply is no way to get from “one scribe of one manuscript a thousand years later added one word to ‘fix’ the text” to “that was originally there, even in Eusebius, and even before Eusebius.” Like Whealey, Allen is going against all the logic of textual criticism as a science here. The data we have demonstrate (as even Whealey herself proved) that there are no versions of the Testimonium today that do not derive from the manuscripts of Eusebius. I demonstrated this myself under peer review (see my article on it, cited by Allen and reproduced now in Hitler Homer Bible Christ).

Even Allen’s own evidence demonstrates it: guess what reading is not in Codex A? Oh right. Whealey’s “he was believed to be the Christ.” If this manuscript derives from the original text of Eusebius as Allen avers, then why does it lack that reading, which also makes the text sound more plausible from a Jewish author? It can’t be both, right? So clearly scribes were happy to make these kinds of changes to the texts of Eusebius—yet, curiously, never to the text of Josephus: no surviving manuscripts of the Antiquities contain either variant. How can that be—unless all manuscripts of the Antiquities derive from the very same manuscript Eusebius produced? They therefore do not derive from any version that escaped his hand. They therefore do not go back to Josephus.

It doesn’t work the other way around either. “Maybe a later scribe removed ‘he was believed’ because he didn’t like it” leaves improbable that that same scribe would keep the supposedly unlikable “a certain Jesus.” A scribe can’t have simultaneously held contradictory intentions. Either they were getting rid of words they didn’t like, or they weren’t, and were preserving the text as received. Indeed, at this point Allen’s theory is even self-contradictory: if these words were so problematic, why did Eusebius keep them? Allen’s theory isn’t like Whealey’s. Whealey can answer this question because she holds that Eusebius did not doctor the passage at all. Which is improbable on its own terms. But Allen is saying Eusebius went to town on it. And that hypothesis makes no sense of these supposedly problematic words at all. Maybe if Codex A were a Codex of Josephus Allen could begin to have a point. But this is a Codex of Eusebius. Which his theory makes no sense of.

Allen’s entire thesis is destroyed here.

Why Are We Acting Like Christian Apologists Now?

Unperturbed, Allen then proceeds to make a bunch of weird arguments that are, admittedly, common failure modes in biblical studies, usually coming from apologists with an agenda. For example, Allen relies on the I Know Better How to Write Than They Did fallacy to argue (following J.C. Paget) that surely had a Christian interpolated the Testimonium they would have put it in with Josephus’s John the Baptist material instead of the Pontius Pilate material (p. 116; on this as a fallacy see my discussion in The Backwards and Unempirical Logic of Q Apologetics). But, news to the world, Eusebius was not Allen or Paget. So it is not likely he will think like them. “I would have liked it better there” does not get you to “Eusebius would have preferred it there too.”

If we want to estimate what Eusebius was likely to do, we have to consider who Eusebius was, what he was doing with this, and his entire context (internally and externally). First, whoever composed the Testimonium drew every single sentence of it from the Emmaus sequence in Luke, in order (indeed, meticulous order: Goldberg shows twenty parallels, nineteen of which are in exactly the same order between both texts). But there is no reference to John or the baptism in the Emmaus narrative. There is, however, a reference to the crucifixion, which is clearly linked to Pontius Pilate in the previous chapter (to which it is obviously referring).

So there is no reason to conclude that Eusebius (or any forger composing this text) would prefer to put it with John the Baptist (where Josephus doesn’t present anything chronologically or credally relevant to a Christian, although Christians probably did meddle with that text too) over instead locating this material exactly as a historian of their day would: at the conclusion of Josephus’s account of Pilate’s actions in Jerusalem—the very context in which a passion narrative like the Testimonium is actually suited. Since the Testimonium centers on the death of Jesus (as all Christian creeds did), not his baptism, its most natural location for a forger would be to sneak it in to the narratives of the only personage pertaining: Pontius Pilate; and, moreover, roughly chronologically where it would belong. This is certainly no less likely than what Allen “wishes” Eusebius did; and is arguably more so. It is fallacious to ignore all of this evidence, and just “assume” that what you would do is what he would do.

Similarly, Allen echoes a common apologetical line when he says (p. 117):

Contra Feldman’s argumentum ex silentio [regarding Justin’s second-century Dialogue with Trypho], Trypho never claimed ‘Jesus had not lived’; Trypho was denying Jesus was the ‘Christ’ by saying a messiah was not born yet, not that Jesus was not born yet. Justin could not have used the original [Testimonium] in his argument against Trypho (Dial. 8); [because] as I show later, the statement that ὁ Χριστὸς οὗτος ἦν (‘he was the Christ’) was not in the original [Testimonium]…

There are a couple of things going wrong here.

First, Allen has confused two different things the Jewish character Trypho is made to say by Justin, conflating them into one thing—and all in line with popular apologetics but not the text of Justin (see my discussion in On the Historicity of Jesus and the Rhetoric of Justin’s Dialogue with Trypho). In the actual text, Trypho both argues that Jesus, if he existed, was not the real messiah (“the Christ—if he has indeed come and is anywhere is unknown and does not know” yet that he is the messiah), and that Jesus could be a made up person anyway (“But you, having accepted a groundless report, invent a Christ”). Which is why Justin has to answer both charges with “we have not believed empty fables, or words without any foundation.” Justin thus knows he has just composed an attack on the very historicity of his man, and not just his status as the messiah (I cover more on how that follows from the entire treatise’s structure in the above link).

Second, Allen has not made a cogent argument against the argument from silence here anyway. Because it is not simply that our text of the Testimonium would have been a coup for Justin to quote, but that any form of the text would have been necessary to quote. This holds even for a lost hostile text as Allen is proposing. Because then that would be a part of the public debate between Jews and Christians that the entire Dialogue of Trypho was written to address. In other words, if such a passage existed, Justin would then have to have his fictional Trypho quote or cite it, and then compose a response to it (thus answering the attack). That the passage never comes up on either side of the debate means it didn’t exist—at all, in any form. Justin did not know he had to rebut a negative text, because it didn’t exist yet. The same reason Justin would not know to quote a positive text that didn’t exist yet.

A more cogent argument for Allen would be to say that we cannot be sure Justin even knew the text of Josephus, that it may have been obscure even in Jewish circles and thus for that reason was not being cited in public debate. This is why the silence of Justin is at best only a weak argument to Feldman’s point—and correspondingly why the silence of Origen is a decisively strong argument to Feldman’s point: because there, we have abundant evidence Origen struggled to scour everything he had to use or address from the Antiquities of Josephus, and milked every piece of weak tea in it he could find. So that Origen never heard of it (not even as a hostile text he needed to rebut) is indeed too improbable to credit; unlike Justin.

Allen seems to miss this point entirely (p. 118). The very reason Origen had to compose a massive multivolume work Against Celsus (and later Christians then endeavored to preserve it) was that there was a hostile text casting Jesus into ill repute that therefore had to be answered. This exact same motive would impel Origen to compose something against Josephus, if Josephus had done essentially the same thing as Celsus—even in general, but especially in his work against Celsus because Origen decided to repeatedly cite Josephus as an authority against Celsus, which means he absolutely could not leave unanswered a hostile text there as well. If Origen is directing readers of Celsus to Josephus, he would be undermining his entire project if he was directing them to an author who supported Celsus, which fact readers of Celsus could then use gleefully against Origen. That would not do. Origen therefore had to address the passage. But he doesn’t. To the contrary, Origen seems blissfully unaware of any hostile passage in Josephus to address. His only complaint about Josephus as an authority is that he was Jewish (and thus not a Christian, ergo “did not accept Jesus was the Christ,” which is the literal definition of any non-Christian Jew, not a quotation of anything Josephus said).

Allen isn’t a Christian apologist (at least I don’t think; he’s in the “Jesus may have been an armed militant” camp, like Bermejo-Rubio), but he is an apologist for an overly-particular theory of Christian origins (involving a lot of convoluted reinterpreting of texts and inventing of others) and thus sounds a bit like a crank (albeit one more erudite than usual). His adoption of illogical reasoning typical of apologetics is thus explicable. But it’s still not commendable.

Why Are We Letting Ourselves Be Duped by Christian Grifters?

Allen is at least correct that the silence of the medieval author Photius is uninformative, and thus cannot mean “his” manuscripts of the Antiquities lacked the Testimonium (pp. 118–19). Obviously, Photius would not complain only of Justus of Tiberias not mentioning Jesus if he thought Josephus also had not. Since all extant manuscripts of the Antiquities derive from the one Eusebius used or produced at the Christian Library of Caesarea, odds are strong any Photius would have known would have had the same ancestry and thus would say pretty much the same as ours do now. It appears none of the original unedited manuscripts of Josephus survived Late Antiquity. Everything in circulation in the Middle Ages was probably the Eusebian edition. Allen is also correct that the extant “tables of contents” for the Antiquities are too brief and generalized to have mentioned any of this material about Jesus, so an argument from their silence is similarly ineffective (p. 120).

But these aren’t any leading arguments against authenticity (you’ll notice neither appears in any of my summaries, linked and cited above). When it comes to those (the actually good arguments against authenticity), Allen gives completely bizarre responses that literally ignore everything scholars making those points actually argue from them. For example, he actually falls for the obtuse apologetical argument that “Origen’s observation that Josephus did not believe Jesus as the Christ…would not have been brought up unless Origen had a version of the [Testimonium] that did not mention Jesus was the Christ.” That’s a non sequitur. Origen does not say Josephus even said this, much less in this passage—to which Origen never refers at all.

Perhaps if Origen had actually said Josephus said this, and that he said it somewhere in some passage other than the James passage (the wrong passage, in the wrong volume of the Antiquities), then Allen might begin to have some point. But the opposite is the case. If Origen even meant Josephus “said” this, he writes as though he said it in book twenty when discussing James, not somewhere else (but we now know Origen had confused a volume of Hegesippus with a volume of Josephus). And Origen did not mean that anyway. When Origen confesses that Josephus did not accept Jesus as the Christ, he is simply saying Josephus was a Jewish author and not a Christian one. And contrary to Allen’s assumption that this admission “weakened Origen’s own argument,” to the contrary it supported it: Origen’s point was that Josephus said the things Origen (mistakenly thought) he said about James even despite not being a reverent Christian, but being instead an objective third party, making his remarks therefore an admission against interest and thus more credible, not less. Yes, Origen completely goofed this up (his fatal mistake of confusing two authors’ scrolls pretty much tanks his argument). But he did not know that at the time.

(And as an aside, Origen appears never to have learned of his mistake. Eusebius was still confused by it a century later, struggling and failing to find what Origen claimed in Josephus, in what would even have been the same manuscript of the Antiquities Origen had, or a copy of it made in the same library. Eusebius had to rescue Origen by juxtaposing Hegesippus, the one who actually said all the things Origen thought Josephus said about Jesus, with a quote of Origen himself, juxtaposed in turn with the James passage in Josephus now duly interpolated—once again, by amazing coincidence, for the first time on record—with a reference to “Christ,” in order to somehow fudge it into something like what Origen claimed, but which what Origen said completely contradicts. Eusebius did the best he could. But it remains a self-evident mess of error and mendacity.)

Similarly, Allen simply declares the Slavonic and (another) Pseudo-Hegesippan version of the Testimonium as “prior to Eusebian tampering” (pp. 118–19). But there is no evidence of this at all, and it is quite improbable. As already discussed, that these authors tried to make the passage sound more suitable from a Jewish author is not evidence of authenticity but the reverse. Otherwise they are both obviously paraphrases of our Testimonium. The Slavonic (Ref. IV) is almost a sentence-by-sentence paraphrase with multiple Christian aggrandizing additions, only framed to sound like they are coming from a fawning Jewish observer (hence the omission of any reference to “the Christ”). The forger in this case did the same to the last two sentences by relocating and rewriting them in an ensuing part of the narrative (Ref. V). Their liberty was quite loose and free, demonstrating that we cannot discern any “pre-Eusebian” text here. All we have is an elaborate Christian rewrite of the Eusebian Antiquities text—and again, still not producing anything like Allen’s proposed version, but something again wildly improbable from the hand of Josephus.

Pseudo-Hegesippus, meanwhile, is an even briefer paraphrase set inside a third-party commentary adding a digression to this material, all based on Eusebius (it even paraphrases things Eusebius says around his own presentations of the Testimonium). It has no similarity to the Slavonic. And being just an essay about what Eusebius says, it can’t even be used as a “text” of the Testimonium at all. I can only conclude that Allen does not know what he is doing. This text is, ironically, exactly like what we should have from Origen, but peculiarly don’t: a description and apologetical commentary on the text, which omits anything that could cast suspicion on it or harm the argument it is being used to make. This proves the author of this text knew the Testimonium (at least through the medium of Eusebius; as all the evidence just referenced indicates they were not using a manuscript of the Antiquities here) in precisely the way Origen’s bizarre and complete silence proves he didn’t.

Allen is thus letting himself fall prey to ancient and medieval Christian propaganda. Their very effort to trick him into thinking Josephus actually said these things is actually working. Which is embarrassing.

And Why Are We Not Actually Reading the Greek?

Instead of being savvy to all that, Allen gives us these perplexing howlers (on p. 120):

First:

• “If we take it that Origen referenced the” Testimonium “and it is more likely that he did” (note that Allen has given no evidence of this even being likely, much less “more” likely) in Against Celsus 1.47, then, Allen avers:

It looks like Origen cites three passages from Josephus’s Antiquities in that chapter: first, he cites the Baptist passage; secondly, he could have seen the [Testimonium] passage as he asserts (regretfully) that Josephus did not believe Jesus as the Christ and they put Jesus to death; and thirdly, he cites what he thinks is the James passage, saying James’s death was the cause of the fall of Jerusalem.

Which is a completely inaccurate description of Origen’s text, which actually reads:

I would like to say to Celsus, who represents the Jew as accepting somehow John as a Baptist, who baptized Jesus, that the existence of John the Baptist, baptizing for the remission of sins, is related by one who lived no great length of time after John and Jesus. For in the 18th book of his Antiquities of the Jews, Josephus bears witness to John as having been a Baptist, and as promising purification to those who underwent the rite. Now this writer, although not believing in Jesus as the Christ, in seeking after the cause of the fall of Jerusalem and the destruction of the temple, whereas he ought to have said that the conspiracy against Jesus was the cause of these calamities befalling the people (since they put to death Christ, who was a prophet), says nevertheless—being, although against his will, not far from the truth—that these disasters happened to the Jews as a punishment for the death of James the Just, who was a brother of Jesus called Christ (the Jews having put [James] to death, although he was a man most distinguished for his justice).

After which Origen goes on to elaborate on that last point.

So. Where is this reference to the Testimonium that Allen claims is here? Sorry. I don’t see it.

What we have is:

• “In the 18th book of his Antiquities of the Jews, Josephus bears witness to John as having been a Baptist”

And then:

• “This [same] writer, although not believing in Jesus as the Christ…says…these disasters happened to the Jews as a punishment for the death of James the Just.”

There are only two references to the Antiquities here. Neither is the Testimonium. Allen has mistaken a mid-sentence digression (a participial deon clause, introducing a sentence in the third person infinitive) expressing Origen’s personal opinion (that Josephus “ought to have said that the conspiracy against Jesus was the cause of these calamities befalling the people, since they put to death Christ, who was a prophet”) for a statement about what Josephus says somewhere. But Origen is here saying the opposite of what Allen claims: that Josephus did not discuss the death of Jesus but should have. If Origen meant what Allen avers, he’d have written something like “since he admits they put to death Christ, and that he was a prophet,” not “since they put to death Christ, who was a prophet,” which is in the Greek syntax an indirect statement of Origen, not Josephus (δέον αὐτὸν εἰπεῖν ὅτι … ἐπεὶ ἀπέκτειναν τὸν προφητευόμενον Χριστόν, lit. “he ought to say that” these things happened “because they killed the prophet Christ”). Then Origen describes what Josephus said (ὁ δὲ καὶ … φησι, “he nevertheless … does say”). Is Allen not competent in Greek? Or is he so lost in his illogical apologetics that he just forgot how Greek works? Either way, he has no correct reading of the text here. It’s the other way around: this proves Origen did not know Josephus had narrated the death of Christ or his reputation as a prophet, and only believed he had mentioned John the Baptist and James.

(Notably, Origen also does not say here that Josephus had said this James was the brother of Christ; those are, again, Origen’s words, not those of Josephus. The extant text of Josephus says this the other way around, lit. “the brother of Jesus (called Christ), the name for whom was James.” Origen is thus paraphrasing his source, not quoting it, and intermixing that with his own commentary and beliefs, which facilitated his error in confusing what he was describing as in Josephus when it is in Hegesippus. Whereas the first author to ever quote the version of the James passage we now have is…Eusebius. In the parlance of the savvy we call that a “red flag.”)

Second:

• “Another reason to indicate Origen was aware of the [Testimonium] was his remark, ‘For the Jews do not connect John with Jesus, nor the punishment of John with that of Christ’ (Cels. 1.48). In Antiquities it does not connect the Baptist movement with the Jesus movement. Also in Antiquities, the execution of John (beheading) is different from the execution of Jesus (crucifixion). Therefore, these two passages taken together (Cels. 1.47, 48) show that Origen used Antiquities in his fights with Celsus.

Huh?

If you are as confused I was, let me try to walk you through what just happened. Allen actually thinks that because Origen knew that Josephus never linked Jesus to John, that therefore Josephus must have described Jesus. Think that through for a minute. This is bonkers. “Mary says Galaxy Quest never mentions Sigourney Weaver starring in Ghost Busters, therefore characters in Galaxy Quest must have discussed the movie Ghost Busters.” Yeah. No. That is loony tunes logic. I struggle not to laugh here. Obviously if the Testimonium never existed, it would still be the case that Josephus never links John to Jesus, which would still be a fact for Origen to observe and remark upon. In no way does this afford any evidence Origen knew of the Testimonium.

It’s only worse that Allen has again completely misdescribed what Origen said. At the end of Against Celsus 1.48, Origen is not talking about Josephus at all (much less the Antiquities). He is talking about a character Celsus invented in his dialogue, whom Celsus portrays as a Jew and thus as a representative of Jews (not of Josephus). To this Origen applies various rebuttals, concluding with the observation that Celsus has incorrectly supposed a random Jew on the street would link Jesus to John and their deaths: “such a statement is not appropriately placed in the mouth of a Jew,” Origen says, “For the Jews do not connect John with Jesus, nor the punishment of John with that of Christ.” This is a statement about public Jewish discourse, not Josephus. And it is discussing what isn’t in that discourse—not what is. So it couldn’t even implicate Josephus—who also did not “connect John with Jesus, nor the punishment of John with that of Christ” even if he never wrote anything about Jesus.

Allen has completely failed at even the most rudimentary logic here.

How Do We Not Know Evidence Trumps Priors?

Allen then spends a page arguing that Eusebius can’t have invented a whole passage because he usually uses or edits existing passages (pp. 120–22). That would support a lower prior probability—except that this is entirely washed out by all the evidence that Eusebius did fake this whole passage. You can’t say O.J. Simpson didn’t commit one murder because he usually doesn’t murder people. That’s bonkers logic. It’s true insofar as we accept it as a principle prior to the presentation of evidence (Simpson was certainly, legally, “innocent until proven guilty,” although he manipulated that process like Eusebius did and got away with it). But once evidence is stacked up proving he committed a murder (and let’s be honest here: he did), that prior is washed out; the posterior is now on the other end of the probability meter. That’s what evidence does.

Now, Allen could have made a savvier argument here, which is the one I make: that we do not actually know for sure Eusebius did it; it could have been his tutor Pamphilus, decades before Eusebius even read the manuscript of the Antiquities that Pamphilus would have left him. There is a lot of incriminating evidence that points to Eusebius—and to Eusebius inventing it all, not just editing something Josephus wrote (all the evidence is against that). But all that same evidence could implicate Pamphilus (e.g., if Eusebius’s style was learned from or emulated his tutor’s)—there is therefore reasonable doubt hovering over Eusebius. While in a real trial that could have been quashed by bringing in stylistic evidence from the writings of Pamphilus, we have lost all that evidence, so we can’t generate that outcome now. That leaves Pamphilus on the suspect list. Eusebius in that case may simply be a dupe of his tutor’s forgery (or complicit—since there is also no evidence Eusebius didn’t know what Pamphilus did and even approved of it as convenient).

But the evidence Allen has here is not strong enough to win a case against Pamphilus, either. In a trial, Pamphilus could rightly point out that this was simply the one instance in which Eusebius worked up a whole insertion and got away with it—and then prove it with stylistic evidence demonstrating its content was more like his student’s than his own, his own eyewitness testimony that the passage wasn’t there when he checked and only mysteriously turned up in a copy Eusebius was carrying around one day, and so on. Evidence trumps priors. We might not have this evidence, but that is not because it did not exist; it was destroyed (we can’t question Pamphilus or anyone else who worked in that library; all his stylistic evidence is lost; and so on). So no argument from silence can work here either. We simply don’t know whether Pamphilus did it, or Eusebius. But the evidence leaves no reasonable doubt one of them did it.

And yet anyone stumping for Eusebius as the culprit would still not be warranted in going off him because Eusebius “didn’t usually” fake up whole texts. People do do one-off crimes. And the evidence does overwhelm any prior against it. The probability of the evidence is much lower than the prior probability that Eusebius resorted to this only the once. And that’s assuming we don’t include Eusebius’s faking up an entire speech from Licinius as “a second time,” on Allen’s weak-tea argument that everyone made up speeches then: that’s still fabricating a whole passage, and most authors at least based their constructions on real sources, whereas Allen himself admits Eusebius didn’t but pretended to. So we know Eusebius is not averse to fabricating whole texts.

But even if we ignore that datum, still at best all we can say is that all the evidence we have (which Allen ignores) that the Testimonium was faked up in its entirety does still leave Pamphilus a suspect in that crime. A possibility with which someone gunning for Eusebius could still agree. But in no way does this get you to Allen’s conclusion. We already know large interpolations were the rarest form of forgery, so arguing it would be rare for Eusebius too does not get to “Eusebius never did it,” any more than it being rare for the Christian community means the Christian community never did this. As rare as they were, wholesale interpolations nevertheless abounded. There are several in the New Testament alone. “But edits of an existing text are a hundred times more common” is simply besides the point. That doesn’t make the many solid interpolations go away. “Christians rarely faked up whole passages” is entirely empirically congruous with “Eusebius rarely faked up whole passages,” which is entirely empirically congruous with “Eusebius faked up a passage this one time.”

Allen is like someone who, failing at even third grade math, does not comprehend that “rarely” is not a synonym of “never.” And then, failing at sixth grade math, he also does not comprehend that unusual events can nevertheless still be proved to have happened—with evidence. All of which he ignores.

And Then It Crashes and Burns

And then…the circular arguments. Allen argues that because the Testimonium could only fit where we found it if it described a calamity throwing the Jews into disarray (because the very next sentence says so), it must have done so; Allen then uses this as an argument (and hence as evidence) that it did (p. 123). But you cannot use a theory as evidence for itself. You need evidence independent of the theory that supports it. The evidence we have supports the opposite: the calamity throwing the Jews into disarray is the previous passage describing Pilate assaulting rebellious Jews over sociopolitical disagreements. There is no evidence of a second additional story in between. Allen is thus treating the total absence of evidence for his theory as evidence for his theory. You cannot get more illogical in historical reasoning than that. Allen does the same with Pseudo-Hegesippus and the Slavonic Testimonium and Codex A (pp. 124–26), which failure of logic I already exposed above. Every argument he makes here is circular. He never considers alternative explanations of the material, or its context, or employs upon it any scientific or professional method. The Devil just planted all the fossils because…look, fossils!

Allen also now starts contradicting himself—admitting Whealey proved “both Agapius and Michael the Syrian recensions are more primitive than the [Testimonium], yet both hark back to Eusebius” (p. 127). Um. It cannot be both. Their text cannot be more primitive than Eusebius and hark back to Eusebius. Then he says “They are also important for proving an additional redactional layer after Eusebius,” yet this contradicts what he said earlier about their version being more primitive. For example, these are the versions that show the later redaction of “was the Christ” into “was believed to be the Christ.” Thus, Allen is simultaneously claiming variants of the text that distance the author from Christianity must come from Josephus “and also” constitute later Christian editing. Allen has literally refuted himself. Allen likewise just insists Pseudo-Hegesippus “follows Antiquities directly” (p. 128) and therefore its version predates the version of Eusebius. But since the material in Pseudo-Hegesippus is demonstrably a paraphrase of Eusebius, Allen’s entire thesis is here contradicted by the evidence.

And yet Allen goes on to completely fabricate a new version of the Testimonium for which no evidence exists and which violates Josephan discourse style in every possible way, and is therefore even more impossible by his hand than the existing Testimonium (pp. 128–42). Allen’s reconstruction is based entirely on heaps of epicyclic speculation, following no discernibly legitimate methodology. His approach resembles a creative reinterpretation of the Zapruder film in which rampaging apes must have been in frame and secretly removed because Kubrick liked them and Kubrick and Zapruder are both from the 1960s and the CIA can do things like that. Therefore they did. (I’m not kidding. Just follow the train of thought, for example, spanning pp. 129–30. Tell me that’s not full-on tinfoil hat.)

Even the coherent nonsense of “Josephus can’t have written that, so he must have written this, therefore he did,” a mode of argument Allen repeats over and over, is an utter non sequitur that has no place in a serious historical study. But his lines of reasoning get even more convoluted than that, and stack upon each other to marvelous heights. It makes even less sense than Christian harmonists who try to rewrite the passion narrative so the Gospels don’t contradict each other anymore, producing an absurd keystone cops fiasco that is precisely not what any author ever wrote. This violates the logic of priors (massively speculative and epicyclic ad hoc theories grow less, not more probable) and the logic of evidence (theories cannot be evidence for themselves; if there is no evidence for your theory, it is not likely your theory is true; you can’t use unproven speculations as facts; and to be evidence at all, an agreed fact has to be less probable on any other theory than yours, not just as or more probable on comparable alternatives—much less simpler ones).

Allen’s entire paper is a travesty of logical mistakes. All to get a wildly implausible new text into existence proving Jesus was an armed rebel against Rome who got a ton of people killed.

If this is what historicists have to resort to, historicity is dead. Time to move on.