The End of the Arabic Testimonium

One of the things I talked about at the Society of Biblical Literature conference in Notre Dame last weekend was the demise of a popular argument for the authenticity of the Testiminium Flavianum (that fawning paragraph about Jesus in the Jewish Antiquities of Josephus). This is a summary for a lay audience of that part of the story. I’ll write about my whole presentation in a following post. But this one issue is complicated enough to get its own treatment first.

Often when expert opinions are cited in favor of some form of the Testimonium having been written by Josephus (even if Christians later fixed it up), they are opinions based on the Arabic fragment and the ensuing argument of Shlomo Pines. But that theory was decisively refuted by Alice Whealey in 2008 (see my following article for bibliography). Therefore, all opinions that were based on the Pines thesis, are now null and void. They are not only uninformed, they were misinformed.

Here’s what happened. In an Arabic Chronicle by Agapius, written in the 10th century, the Testimonium gets quoted, yet in such a way that looks much less credulous, maybe more like what a Jew might have written. Pines argued that this must be an Arabic translation of an earlier manuscript tradition of the Antiquities than the one known to Eusebius (the first author ever to notice the passage existed in Josephus), and thus must reflect what Josephus really wrote, as it bypassed the meddling hand of Eusebius or whoever preceded him at the Library of Caesarea (where the manuscript of Josephus was produced that Eusebius used).Chart that shows the causal pathway Whealey proved, from the Church History of Eusebius in the 4th century, to the Syriac translation of that in the 5th century, to the Syriac Chronicle of Theophilus in the 8th century that used that 5th century Syriac, and Theophilus's Chroncle was used in turn to produce the Arabaic of Agapius in the 10th century, and copied into the Syriac Chronicle of Michael the Syrian in the 12th century.

However, Alice Whealey proved, quite conclusively, that in fact Agapius was translating the Syriac edition not of Josephus, but of Eusebius. And it therefore certainly did not come from any earlier manuscript tradition untouched by Eusebius, but the very same one, in fact from Eusebius himself! Moreover, Agapius was translating this passage from the Syriac Chronicle of Theophilus, written in the 8th century, the exact same text copied by Michael the Syrian in the 12th century in his own Syriac Chronicle. And though we don’t have Theophilus, we can tell from Michael that Theophilus’s text was essentially identical to the known Greek text of Eusebius’s quotation of the Testimonium in his copy of Josephus (but for one key difference I’ll get to in a moment). Agapius was therefore taking liberties, and altering the text in some way that suited him. His translation was thus not reliable, but more like a crude and speculative interpretation. His Arabic can no longer be used to argue for anything authentic appearing in the Antiquities.

Any opinion not aware of this finding, is uninformed and unciteable. It has no value. Unfortunately, that means most of the opinions anyone ever cites.

Now, Whealey may have destroyed the favorite prop experts once grasped to try and vindicate Josephus having mentioned Jesus. But she herself is fanatically dedicated to the Testimonium’s authenticity. So she tacked onto her findings regarding the Arabic, which were based on well-evidenced and sound argument, a new argument based in no evidence nor sound reasoning.

Whealey’s new defense of the Testimonium is an argument from improbability. But it’s not logically valid.

Every Greek text of the Testimonium declares Jesus “was the Christ,” a sentence no faithful Jew, much less Josephus, would ever utter of Jesus (long the clearest evidence the Testimonium is a Christian forgery). But the Syriac quoted by Michael declares Jesus “was believed to be the Christ,” a softer assertion that perhaps maybe a Jew like Josephus might write. This same emendation appears in the Latin translation of Jerome, written just half a century after Eusebius. Note, however, that in both cases, Jerome’s Latin and Michael’s Syriac are quotations of the early 4th century Church History by Eusebius, and not of any manuscript of Josephus. Yet they are eight hundred years apart. And Syriac writers were rarely influenced by Latin readings.

Accordingly, Whealey argues that it’s improbable that Jerome and Michael would both, independently, make this change (in each case deciding that surely Josephus, being a Jew, meant “was believed to be” rather than “was,” and maybe the text had become corrupted or misinterpreted—a thought process we know moved many a scribe to creatively “emend” the manuscripts they copied), and therefore, Whealey argues, that can’t have happened. The only other alternative, she claims, is that the original Greek of Eusebius must have read, in fact, “he was believed to be the Christ,” and that this original reading is preserved in two translations, the Latin of Jerome and the Syriac. And this explains why Michael’s Syriac text in the 12th century read that way—because he was copying from Theophilus, and Theophilus was copying from the 5th century Syriac translation of the Church History.

We don’t have Theophilus, so we can’t confirm how his text read. But we do have a few other manuscripts preserving readings from that 5th century Syriac translation of the Church History that he used…and they both say “he was the Christ.” Not “he was believed to be” as Whealey’s thesis requires. Hmmm. So Whealey’s thesis runs aground on two logical problems here that she doesn’t seem trained to deal with. And this is my own observation; no one has as yet published this, though I may do:

  • First, if all possible explanations of the evidence are improbable, it is no longer logically valid to rule out a theory because it is improbable. Because then the one thing you know for sure is that the correct explanation is improbable. One has to, instead, compare the relative probability of the competing hypotheses. And in this case, all possible explanations are improbable.
  • Whealey’s thesis is actually the least probable of all the possible explanations of the evidence. And that includes the explanation she rejects (that two unconnected authors coincidentally made the same emendation) but also explanations she overlooks (that both authors relied on a single corrupted text in common).

An example is concluding Joe won the lottery. That is always improbable. The probability that Joe, of all people, won the lottery is, let’s say, a million to one against. However, abundant evidence attests that he won. Newspaper announcements, state government records, his winning ticket receipt, his newfound wealth. That all that evidence was forged or in error is also improbable. In fact, it’s far more improbable—let’s say, a trillion to one against. That means that even though Joe’s winning the lottery is a million to one against, that he won is still a million times more likely than the only available alternative explanation of the evidence. And that means, all else being equal, the odds that Joe won the lottery are in fact a million to one in favor. Even though the odds of his winning the lottery were a million to one against. Whealey is acting like someone who argues against this by saying, “but, that Joe would win the lottery is improbable, so he can’t have won!” That’s logically invalid.

A visual comparison showing how Whealey's theory, Theory 1, is far less probable than an obvious alternative, Theory 2, as explained in the text.That two authors hundreds of years apart would come up with the same conjecture (that surely a Jewish author like Josephus meant “he was believed to be” and therefore that is a more accurate translation of the author’s intended meaning) may be improbable. But it’s not that improbable. Far more improbable is every other possible explanation of the evidence. For example, Whealey does not even consider this scenario (diagrammed at right): 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 an ancestor of it) was used by Jerome to make his translation, and a descendant of that manuscript was used to make the Syriac translation in the 5th century, 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).

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 millions of times more likely.

And that’s not the only alternative. It’s entirely possible the reason all extant manuscripts of the Syriac translation of the Church History of Eusebius read “he was the Christ” and not “he was believed to be the Christ,” is that in fact that’s what it always said, because that’s what the Greek text the Syriac author was translating from said, because that’s what Eusebius wrote (because that’s what Josephus wrote). And what happened instead is that the single corruption in the Greek that Jerome used to translate his Latin from survived in descendants of that manuscript that were known to Michael (or Theophilus), and he (or Theophilus) used that corrupted Greek text to correct his text by (a common occurrence in ancient and medieval translation and transcription work). Then all the evidence is explained, including why all extant copies of the 5th century Syriac don’t agree with Michael’s, by positing only a single textual corruption in a single manuscript tradition of a single book (the Church History of Eusebius). Only one other ad hoc assumption is required: that Michael or Theophilus had access to a Greek manuscript of the Church History derived from that same corrupted manuscript translated by Jerome, and used it to emend his lift from Theophilus—or Theophilus used it to emend his lift from the Syriac translation of Eusebius (and Michael simply copied that).

That’s actually more likely than the alternative that the Syriac and Latin translated from a Greek text derived from a single corruption that changed “he was” to “he was believed to be.” Because that thesis requires two things to be assumed, each increasing the improbability: that someone later “fixed” the textual tradition of the Syriac to say “he was,” and that history gave us two other coincidences besides, that the “changed” text is the only version that survived in any manuscripts known today and that the other version (the one that produced no surviving ancestors) was nevertheless the one used by Michael or Theophilus. That the Syriac would be changed to match all extant Greek textual traditions (in two of Eusebius’s books and in the Antiquities of Josephus) and the surviving Syriac translation of a third book by Eusebius (in which he also quoted the Testimonium) is the same presumption that Michael or Theophilus did, so the improbabilities there are equal. And yet still I’d say that Michael and Jerome acting independently is more likely—because it requires none of all these added assumptions. And yet even those added assumptions are vastly more probable than the impossible managerie of coincidences, bordering on widespread conspiracy, that Whealey’s thesis requires.


In 2008 Alice Whealey refuted the thesis that the Arabic fragment of the Testimonium supports its authenticity. It doesn’t. Because unlike what experts thought, the Arabic fragment does not come from Josephus at all. It comes from Eusebius, through the intermediary of a well-known Syriac translation of Eusebius.

Meanwhile, Whealey’s alternative proposal, that a massive conspiracy of altering five separate books, by two different authors in two different languages, exactly the same way, is so astronomically improbable it has no plausible chance of even being possible.

It’s vastly more likely the reason Jerome’s Latin and Michael’s (or Theophilus’s) copy of the Syriac translation of the Church History of Eusebius both say “he was believed to be the Christ” is that two smart scholars out of the hundreds dealing with this text over the course of nearly a thousand years both independently had the same notion that surely Josephus meant that, and so they “interpreted” the text that way when they translated it (just as modern Bible translations in English “interpret” all kinds of things in ways not exactly matching the literal text).

It’s even vastly more likely (if not as much) that the reason is that Jerome and Michael (or Theophilus) were referencing a corrupted Greek text of the Church History, caused by a single scribal emendation or mistake in a single manuscript of Eusebius in the mid-4th century.

It’s even vastly more likely (if not nearly as much) that the reason is that Jerome and the 5th century Syriac translator whose edition was used by Theophilus (and thus Michael) were translating from a corrupted Greek text of the Church History, caused by a single scribal emendation or mistake in a single manuscript of Eusebius in the mid-4th century (and someone then “corrected” later copies of that Syriac translation to match the more widely extant Greek that read “he was”).

All three hypotheses are vastly more likely than Whealey’s. And all of them establish that Eusebius did not write “he was believed to be the Christ,” nor ever read such words in any manuscript of Josephus known to him. Whealey would notice this if she had learned the logic of probability, and understood that “x is improbable, therefore x is false” is invalid—the same fallacy as asserting “Joe winning the lottery is improbable, therefore Joe didn’t win the lottery.” What matters is not whether the explanation is improbable. But what it’s probability is relative to all possible alternative explanations of the evidence.

And people still wonder why historians need to learn Bayes’ Theorem already?



Add a Comment

I only publish comments by my patrons and anyone who or whose work I discuss in the article commented on. Comments must also follow good etiquette. Those who engage in dishonest, abusive, or harassing behavior may even be banned as commenters and patrons.