Did Josephus write his paragraph about Jesus by slavishly copying Luke? No. In my Ongoing List of Updates to the Arguments and Evidence in On the Historicity of Jesus I maintain a section on Josephus, and as of now it simply summarizes and references my article Josephus on Jesus? Why You Can’t Cite Opinions Before 2014, originally published in 2017 but which I continue to update as the central go-to. There have been attempts since then to rehabilitate the James passage in Josephus (which all fail on basic literary theory), which is found in the twentieth volume of Josephus’s Antiquities of the Jews, and my discussions of those are listed there; but until now there really hasn’t been anything “new” on the main passage of contention, the infamous Testimonium Flavianum (or TF), the weirdly brief and fawning summary of the Christian gospel found in the eighteenth volume of the same Antiquities.

By nothing “new” I don’t mean no one has since tried, but that no one has advanced any new arguments not already refuted by the points made and studies cited in my summary article about this—which in fact was the whole point of that article: scholars are now just ignoring the actual state of the question, ignoring almost everything published in and since 2014, and thus just churning out repetitive apologetics rather than actually moving the ball down the field. And this includes the new Gary Goldberg article I will be discussing today, which although containing a “new” argument, simply ignores the research that (as if psychic) already refuted his conclusion years before (and which I already had cited in my article warning researchers to stop ignoring these new studies; Goldberg heeded it not).

Quite simply, no study can succeed that does not legitimately answer the arguments now summarized in Chapter 8.9 of On the Historicity of Jesus (Sheffield-Phoenix 2014; rev. 2023) and, expanded and updated, in Lataster’s Questioning the Historicity of Jesus (Brill 2019), pp. 36–39 and 192–202.

The “New” Argument

Of course my overall conclusion has long been that neither passage now mentioning Jesus in the Antiquities ever did when Josephus wrote—that both involve Christian interpolations of the 3rd century (indeed, we can trace their insertion to a single manuscript in a specific library: see my study in the Journal of Early Christian Studies, reproduced for more affordable access in Hitler Homer Bible Christ); but that this has no real bearing on the historicity of Jesus anyway, because Josephus wrote well after the Gospels and Christian lore built on them, and thus cannot be described as an independent source. There is no evidence he fact-checked any of this—and in fact, as we’ll see, Gary Goldberg’s new attempt to argue for the authenticity of the TF entails he didn’t, but simply mindlessly, uncritically, and slavishly copied material from a single chapter of the Gospel of Luke, thus confirming my point. This is ironic because Goldberg wants to rehabilitate the historicity of Jesus by rescuing the TF as authentic, but his effort actually destroys any use the TF could have had to prove Jesus really existed, by proving no one really checked, they just gullibly believed the Gospels. The TF is therefore useless.

On this much I agree with Goldberg: his previous demonstration that the TF is just mindlessly, uncritically, and slavishly copied from the Emmaus narrative in Luke is in my opinion conclusive (see Gary Goldberg, “The Coincidences of the Testimonium of Josephus and the Emmaus Narrative of Luke,” Journal for the Study of the Pseudepigrapha 13 (1995): 59–77). The probability of that conclusion, on the evidence he presented, simply beats all alternatives now (so no amount of trying to argue it’s “possible” these are all just coincidences can succeed, because ‘possibly’ does not beat ‘probably’). Goldberg is a Christian apologist, however, and thus does not like what this conclusion portends (that a Gospel-loving Christian, not Josephus, wrote the TF). So he wants to prove that Josephus himself, and not (as is more obvious) a Christian interpolator did this. And he does this by trying to prove merely that the paraphrase uses tricks of word-substitution that are found elsewhere in Josephus (in Gary Goldberg, “Josephus’s Paraphrase Style and the Testimonium Flavianum,” Journal for the Study of the Historical Jesus 20 (2022): 1–32).

It is of course wildly improbable that Josephus would even use a source like Luke (or even know of it); it’s even less probable that he would simply paraphrase one chapter of it uncritically. The author of the TF gullibly believes all sorts of absurd things, like that Jesus “appeared” to his followers risen from the dead or that “the prophets of God had foretold these things and a thousand other marvels about him.” They also seem to assume everyone already knows the whole story—because at no point is any strange remark in it explained, like why was Jesus executed (and by Romans no less; and Pilate in particular), or what he even taught that “won over” both “Gentiles and Jews.” It seems self-evident only a Christian would have composed the TF, and indeed not just any Christian, but a Christian writing in a context where Christianity is so widespread and well known that they would not think any random reader wouldn’t already know its story (and that was not the time period when Josephus wrote).

Such are the general arguments already refuting Goldberg’s new thesis right out of the gate. But there is more specific evidence against it, already outlined in prior studies.

What’s Wrong with Goldberg’s Argument

I’ll demonstrate that last point in my closing section. First we should survey what Goldberg’s new thesis is and on what evidence and methods it’s based on. His general approach is to ask, in effect, “if” Josephus had paraphrased the TF from Luke 24, “then” what would we expect him to do, in terms of word choice, when transforming that passage’s material into his own voice. Goldberg cherry-picks several examples and finds they match practices one can likewise cherry-pick from the collected writings of Josephus (where we can see how he paraphrases other texts that we have). But Goldberg never performs a control: he never asks what we could expect any Greek author to have done, particularly one well versed in Josephus’s works (like, say, Eusebius—or his predecessor and tutor, Pamphilus, which I’ll circle back to).

Goldberg’s approach is thus methodologically illogical: it fails to compare his thesis with competing explanations. He thus tries to argue for a high likelihood, but never for a low likelihood of its best competitors; which renders his argument formally invalid. “This has a high likelihood, therefore it has a high probability” is a fallacy of incomplete Bayesian argument (see If You Learn Nothing Else about Bayes’ Theorem, Let It Be This and The Scary Truth about Critical Thinking). Goldberg also commits a standard cherry-picking fallacy by only including in e (the evidence to be explained) the examples he found matches for. He does not include all the discordant ones (as documented by Ken Olson, a point I’ll get to later). But the likelihood of e must be that of all of e, not just the parts that agree with your thesis.

Worse, Goldberg myopically only focuses on a model of slavish paraphrase of a poorly articulated narrative; he does not consider that in fact Josephus never does that. This was the point of Paul Hopper’s study of Josephan discourse style (not just verbal or grammatical style). Which I’ll also get to later; but the gist is simple: Josephus writes causal narratives; the TF lacks any causal component (it never explains why anything it lists happened, it just lists things happening, unexplained); and it lacks appropriate explanatory style. For example, Josephus never says things like “the prophets of God had foretold these things and a thousand other marvels about him”; he would explain what he means, knowing his readers won’t otherwise know what he is talking about. Llikewise Josephus would not cite an appearance from the dead without cross-referencing his previous discussions of Jewish resurrection beliefs, for example, or even indicating what he himself thinks happened. What the TF instead looks exactly like are Christian credal statements, a style of composition Josephus never employs—even when, indeed, specifically describing sectarian and religious movements (which he does often).

On top of all that, most of Goldberg’s argumentation also commits the possibiliter fallacy (treating something that is merely possible, as if it is therefore probable).

Examples Thereof

I won’t fisk Goldberg’s entire paper. I don’t think that’s necessary, as the same errors are just repeated over and over, such that once you see the error in a few examples, you can see them in all.

Goldberg opens his paper with a general case for Josephus being known for revamping what he paraphrases in his own words, rather than merely copying verbatim. But in fact all Greek writers were taught to do this; it isn’t an affectation of Josephus, it’s an affectation of the entirety of the ancient Greek educational system and literature. So it does not help to show that whoever wrote the TF did this. Anyone who would have written the TF would have done this. The more bizarrely slavish adaptation of sources in the Synoptic Gospels is actually so highly peculiar in Greek literature that it is more correctly described as redaction: as with the many redactions of the Life of Aesop (which are a rare instance of this practice found outside the Gospels), the authors of Matthew and Luke are trying to pass their stories off as the originals, by simply plagiarizing their sources with minor tweaks (and adding or subtracting stories as needed). The Gospel of John, by contrast, rewrites the Synoptics in his own words—as was in fact a more typical way someone then would use sources (see OHJ, Ch. 10.7). Paraphrasing, by contrast, was the second most typical way someone would use sources, as we find in, for example, Christian summaries of the Gospels (like the Long Ending of Mark, on which see Hitler Homer Bible Christ).

So what we really want to know is not whether Josephus did this, but whether it was done in any way more like Josephus did it than any other prospective author—like Eusebius or Pamphilus. And this is where it becomes important that far too frequently the answer is no (as Ken Olson demonstrates): the TF contains too many instances where the vocabulary choices match the tendencies of Eusebius and not Josephus. Goldberg ignores all these “misses” and tries to collect what “hits” he can instead, and then like a Vegas psychic claim the hits prove the prescience of his thesis. But this is illogical. The likelihood of all the transformations in the TF from Luke must equal the probability of the entire matrix of changes, not a cherry-picked sample. It’s not just “how often did other authors use the same transformation steps Goldberg finds” (though he never calculates that and thus never completed any logical argument from what he did find) but “how often would we expect that and all the incongruous transformations as well.” The answer is not going to favor Goldberg’s thesis. It is, in fact, going to refute it.

For example, Goldberg tries repeatedly to make hay out of “shared word roots” whereby “Josephus’s revisions typically retain a few words from their sources, often with a change in inflection or conversion into a cognate.” But this is true of nearly any paraphrase technique. It is in no way peculiar to Josephus. It’s just as expected a Christian paraphraser will do the same thing—especially considering their peculiar reverence for the source-text. None of this evidence thus advances Goldberg’s thesis at all. It must all be tossed out as useless data.

Likewise, Goldberg burns word count trying to prove the TF doesn’t match the style of Luke-Acts; but since we all agree there is no way the author of Luke-Acts wrote the TF, that it would not match their style is entirely expected regardless of who wrote it. This therefore has no value for arguing Josephus wrote it. And while Goldberg claims his thesis would explain the TF’s deviations from Josephan style (as illustrated by Olson, for example), this is simply false: most of those deviations (particularly the ones most closely matching Eusebian style) do not exist in Luke and therefore cannot be explained as coming from there. There is only one probable explanation of where those oddities could come from: the forger of the TF; who has to be most likely either Eusebius or someone close to him in stylistic preferences (which is why his predecessor, Pamphilus, who taught him his style—and who would have handled or produced the manuscripts of Josephus Eusebius then relied on—is also a leading suspect; we just lack texts from him to compare, yet have abundant writing from his student).

I must agree with Goldberg, though, that attempts to rehabilitate the TF by completely rewriting it and then passing off that hypothetical version as what Josephus wrote are methodologically invalid. Or as he puts it: “A major criticism” of such approaches “is that the selection methodology” of what to drop or retain (or even suppose was deleted) “is based not on an objective foundation, but on suppositions.” And it’s suppositions in; suppositions out. This is GIGO. It has no validity as a method. Which is a major point of Lataster’s Questioning the Historicity of Jesus: that historians of Jesus rely absurdly much on “hypothetical sources.” Which can only get you to hypothetical conclusions. Trying to edit the TF to sound more like Josephus illustrates that same unsound practice—even apart from the fact that, per Olson and Hopper (as we shall see), they also fail to sound more like Josephus, but end up sounding even less so; another consequence of ignoring these crucial studies of the TF; and also apart from the fact that Goldberg’s prior thesis conclusively refutes all such “versions” anyway: once you scratch off every line derived from Luke 24, there is nothing left to creatively “delete.” So the TF had to have been composed entirely as we have it, all in one go, by a single author paraphrasing Luke 24. No other thesis even bears plausibility now. That argument is dead.

But for his new thesis, Goldberg has only one other kind of evidence to cite: he argues “there are eight expression substitutions…that have identifiable precedents” in Josephus (and another eight with “indirect support” and another three “with little or no support”), without showing that such expression choices (out of all 89 words the TF consists of) is unexpected even on chance, much less given specific suspects, like Eusebius or Pamphilus, who were also familiar with Josephuan style. Even though they failed, and could never likely have succeeded, in replicating Josephan style completely, we cannot suppose they made no choices uninformed by what they thought to be typical moves for him. So they will certainly have performed better than chance. While indeed it is all the ways they failed (per Olson and Hopper, below) that makes it unlikely any remaining congruences came from Josephus.

For example, Goldberg notes that Josephus will often transfer what is a conversation in a source into third-person narrative in his text. But since anyone composing the TF would have done this, this is not evidence Josephus is the one who did it. For example, Goldberg notes the TFs author changed Luke’s “the things that have happened” into “it happened that.” But the latter is an extremely common phrase (it is not peculiar to Josephus), and it is the most logical way for anyone to compress a statement referring to previous discussions now absent. After all, no one paraphrasing this could repeat the reference to previously related events (“the things having happened”), because, unlike in Luke, those things haven’t been related in the Antiquities to refer to. Whereas starting a sentence in the way we find in the TF is common across Greek literature, and obvious to any reader of Greek as a mode Josephus employed (even though he rarely used it: as Goldberg says, just “twenty-nine times at the beginnings of sentences in his works,” works that collectively contain literally thousands of sentences, making chance hits like this highly likely). So this transformation evinces nothing specific to Josephus.

In just the same way, Goldberg notes with odd surprise that Josephus occasionally transformed Septuagintalisms favored by Luke like “in those days” to a more contemporary idiom like “around that time.” But that “modern” idiom appears commonly across ancient literature; and the exact form is as commonly employed by Eusebius as by Josephus (even subtracting his repetitions of the TF). So it indicates nothing particular to Josephus. Likewise, Goldberg is surprised the author of the TF said “a certain Jesus” rather than something more credal like just “Jesus” or even “Jesus of Nazareth.” But if someone were composing a vignette in the mouth of a purportedly neutral author like Josephus, that transformation is fully expected. That was a common way across all Greek literature to introduce a new character when you don’t want to give their full patronymic or description of origin, and a forger would not want to introduce elements Josephus hadn’t already discussed (like who Joseph was or where Nazareth was). If you just wanted to say, “Hey, there was this Jesus…” this is what you’d say. There is nothing particularly Josephan about this.

Not only are none of Goldberg’s examples peculiar to Josephus, in some cases his thesis strains against Josephus. For example, in respect to identifying who this Jesus was (per above), Josephus actually more typically (not always, but more often) would elaborate, especially when it’s quite relevant to his account. Yet, if he was using Luke, indeed slavishly trusting it like Goldberg’s thesis requires, why wouldn’t he know about its Nativity and genealogy, and remark upon it? Or indeed even Jesus’s endorsement by John the Baptist, a character Josephus did write about? Likewise its account that Jesus was a Galilean and began his movement there—and thus was a fellow-countryman of Josephus (who famously governed Galilee). We can’t say Josephus would surely have brought in these details; but it is weird that he wouldn’t bring in any, that he would just summarize Luke 24 and never use any other part of that source to compose his story. Whereas this is more expected of a hasty Christian paraphraser. Again, it’s not black or white. Either is possible. But the balance of probability is not in Goldberg’s favor here.

We could list example after example of this kind of thing. Goldberg says Josephus occasionally used the phrase “whether one should say.” But so did countless other authors. So there is nothing peculiar to Josephus here. In fact, the TF uses εἴγε; yet Josephus only ever uses εἴτε in such expressions. So this even looks more like a miss than a hit. Likewise when Goldberg wants us to believe a match occurs with “doer of strange deeds”—even though he can find no instance anywhere in Josephus; and the closest matches he can come up with all use different wording and phrasing. So there is no connection with Josephus here at all. But you know who uses that exact phrase several times? Uh. Yeah. Eusebius.

Likewise, Goldberg wants us to be surprised that Josephus’s tendency to reduce “alls” to “manys” is exhibited in the TF; except in the case he means, the meaning of the sentence was changed by the paraphraser requiring that transformation, so any writer would have had to do that. Luke says Jesus was powerful before “all the people,” but the paraphraser said he persuaded (“won over”) “many” people. Obviously the word “all” is impossible there; even a Christian author would have had to change it. Hence in cases like this, Goldberg seems intent on ignoring the actual context and intent of the sentences he is talking about; and is definitely ignoring alternative hypotheses to his own.

So whenever you survey his examples, notice what is never told you. For example, a ‘prophet man’ in Luke becomes a ‘wise man’ in the TF. Okay. But is that unusual? Even in general? (Isn’t this the kind of paraphrasing we might expect from any author, Christian or otherwise?) But also in particular: this affectation, while Goldberg can find no examples of it in Josephus, is peculiar to Eusebius. It’s one of the evidences Olson collects that Eusebius wrote the TF. Goldberg and Olson both give “just-so” stories as to why this transformation happened, neither supported by any direct evidence, so that much is a wash. But given the evidence that this is a Eusebian idiom and not a Josephan one, the balance of probability still favors Olson here, not Goldberg. (Note, again, I would expand Olson’s thesis to include more suspects than Eusebius: no author’s affectations arise in a vacuum, and Pamphilus had equal means, motive, and opportunity; as surely would have others similarly situated that we don’t know about.)

You can pick any example Goldberg presents as evidence, and it will fall to one or more (even sometimes all) of the above objections, to the point of failing to evince anything.

This Was Already Refuted Years Ago

Besides being internally a failure, Goldberg’s thesis doesn’t even survive any of the traditional arguments against the authenticity of the TF: the TF doesn’t fit its context; it’s implausible from a Pharisaic Jew; it’s improbably brief; it’s improbably obscure; and it was unknown to Origen, who knew the Antiquities well and struggled to find or respond to anything in it on Jesus; and all attempts to “fix” these problems rely on baseless speculation. But it is also refuted by specific studies. Goldberg cites the old but not any of the more recent work of Ken Olson (which refuted the attempted rebuttals Goldberg does cite); and he cites Paul Hopper, but seems completely unaware of the contents of Hopper’s study and how it refutes everything Goldberg is saying. At any rate, he never addresses Hopper’s arguments; or Olson’s latest.

Those are now required reading for anyone who wants to doubt this:

• Ken Olson, “A Eusebian Reading of the Testimonium Flavianum,” Eusebius of Caesarea: Tradition and Innovations, eds. Aaron Johnson and Jeremy Schott (Harvard University Press: 2013), pp. 97–114.
• Ken Olson, “The Testimonium Flavianum, Eusebius, and Consensus,” The Jesus Blog (13 August 2013).
• Paul Hopper, “A Narrative Anomaly in Josephus: Jewish Antiquities xviii:63,” Linguistics and Literary Studies: Interfaces, Encounters, Transfers, eds. Monika Fludernik and Daniel Jacob (de Gruyter: 2014), pp. 147–169.

We should be asking how Josephus describes other sects. That’s the proper comparand—because he does this a lot (separately for the Sadducees, the Pharisees, and the Essenes, for example), yet the TF does not resemble any of those in any way—yet all of those resemble each other in ways particular to Josephus. Likewise how Josephus usually describes religious controversies, like the very next story (after the TF) about a religious scandal in Rome. Likewise his accounts of messianic pretenders: those all share a common structure and running metaphor, depicting men acting like the Biblical Joshua, a.k.a. Jesus, indicating messianic purpose—while conspicuously avoiding using either word (“Jesus” or “Christ”). That is how Josephus describes movements like this. Several times in fact (see OHJ, pp. 68–72). Look at all those and you will see that is how Josephus tells stories. In respect to storytelling, the TF resembles nothing in the entire corpus of Josephus. It is in fact almost exactly the opposite of how he relates stories.

This kills Goldberg’s thesis dead. As does the fact that Olson documents numerous parallels between the style of the TF and Eusebius, more than Goldberg can claim for Josephus, and the matches with Eusebius are unprecedented or weird in Josephus. And as these deviations from Josephan style cannot be explained as coming from Luke, they must come from another author—someone other than Luke or Josephus. Eusebius already fits the bill better than they do. Others could have. Read my summary of Hopper. Goldberg cannot win against that. Hopper’s evidence is devastating. Read my summary of Olson. Goldberg cannot win against that. Olson’s evidence is devastating. All Goldberg has is evidence that the TF uses commonplace scholastic methods of periphrasis in ancient Greek literature; nothing linking any of this to Josephus as its author. Whereas Hopper and Olson together prove conclusively there is no possible way Josephus wrote the Testimonium Flavianum.

Hopper also proves conclusively that all “truncated” and “edited” versions of the TF, every “hypothetical” TF that this or that scholar invents to try and rescue it, is even less likely to come from Josephus. Because it then looks even less like how Josephus tells stories. The only plausible “hypothetical original” would have to be more elaborate—it would have to have a causal narrative (explaining why each thing he says happened came about), it would have to have an explanatory framework (explaining what each weird thing he says means and how his readers should understand it), and it would have to have a contextual framework (explaining what this story does to further Josephus’s argument and overall narrative; what connects it to the stories before and after it). Because that’s how Josephus composes narratives. By contrast, the TF looks (as Hopper shows) a lot more like Christian creeds as a genre, a genre Josephus would never use (and never does).

To see an example of Josephan storytelling style and what that tells us about what he could or couldn’t plausibly have written, see my essays on the James passage as an example, especially Reading Josephus on James. I apply the same principles to the TF in OHJ, line-by-line (pp. 332–42). Finding that the author of the TF used typical methods of periphrasing does nothing to counter all this evidence against Goldberg’s conclusion. The traditional arguments defeat him; Olson’s arguments defeat him; Hopper’s arguments defeat him. He’s trounced. There’s nothing left—except Goldberg’s own conclusive proof that the author of the TF just slavishly abbreviated Luke, which sooner rules out Josephus as its composer. And that’s it. It’s time to give up on the Testimonium Flavianum.