Ken Olson on the Testimonium as a middle term.

Josephus, Antiquities 18.3.3 §63-64; Luke 24.19b-21, 25-27; Tacitus, Annals 15.44.

XTalk message 20461.

Ken Olson has kindly taken the time to respond to my analysis of Josephus, Antiquities 18.3.3 §63-64 with Luke 24.19b-21, 25-27, and Tacitus, Annals 15.44, on XTalk, message 20461. On this page I reproduce the entirety of his message in italicized blocks and respond to each block in turn:


Thanks for the link to Ben Smith's page:

Let me point out two elements of subjectivity/selectivity here.

Ken is responding to Chris Weimer, who posted a few comments on XTalk about the connections between the passages in Josephus, Luke, and Tacitus.

First, Smith is comparing five and a half verses excerpted from Luke to two lines excerpted from the Testimonium to one line excerpted from Tacitus. He's not comparing whole documents.

True enough. There is value both in comparing entire texts and in comparing individual passages. For instance, the term middle term, as far as I can tell, first arose in synoptic studies as a description of the comparitive orders of Matthew, Mark, and Luke. Scholars noticed that Matthew and Luke never agreed against Mark in ordering pericopes, making Mark the middle term across all three texts in their entirety. However, this same term is also used now for individual pericopes in the synoptic tradition and is applied to the content of each text as well as to its order. In the Beelzebub controversy, for example, it has been observed that Matthew serves as the middle term between Mark and Luke.

Just as any synoptic theory must account both for the texts in their entirety and for each set of parallel passages within those texts, so our theory of relationships between Josephus, Luke, and Tacitus must account for both levels of inquiry as well. It so happens that my page comparing them focused on the passage level.

If we took Tacitus as our basis for comparison, we would find that there are agreements between his 15.44 and other parts of both the Antiquities and Luke-Acts that are not considered in his sample.

Agreed. But it needs to be made perfectly clear that any agreement that we find between Tacitus and both Josephus and Luke will contribute nothing either for or against one of these three texts being a middle term between the others. A middle term text is one against which the other two texts never agree together. Any triple agreement between all three relevant texts will by definition have nothing to offer on the topic of whether one of the three is a middle term for the others.

As Carlson points out, you could derive Pilate's office and the fact that these events occurred during the reign of Tiberius from material in the Antiquities outside the TF. Similarly you could derive all that plus the fact that Christianity came to Rome from Luke-Acts.

I would add that we can also derive the cognomen Pontius from both Josephus and Luke. However, as mentioned above, the fact that we can mine all three of our texts for this cognomen, the office of Pilate, and the emperorship of Tiberius automatically disqualifies any of these data as indices either for or against a middle term.

Only if we can derive a datum in Tacitus from Luke, but not from Josephus, is the observation that Josephus is the middle term in any way mitigated. Olson supposes that the presence of Christianity in Rome is one such datum, and indeed Josephus appears to have nothing on that while the second Lucan volume, the Acts of the Apostles, has plenty. There is, however, no way that this particular datum can count against Josephus as a middle term because the presence of Christianity in Rome is a corollary of the Neronian persecution that Tacitus discusses in the context. The persecution of Christians under Nero appears neither in Josephus nor in Luke. In other words, Tacitus had a different source for the treatment of Christians in Rome. I am not sure what that source was, but what is certain is that Nero could not have persecuted Christians in Rome unless Christianity had first managed to spread to Rome. Given our sure knowledge that Tacitus knew of the spread of Christianity to Rome apart from either Josephus or Luke, there is nothing compelling us to presume that he got this datum from Luke. This datum, therefore, cannot have any impact on our potential identification of a middle term amongst our three texts.

Update 08-07-2006: Olson notes by email that the middle term phenomenon is usually conceived of as something that can be observed in texts independently of source hypotheses; in this case, then, the presence of Christianity in Rome has to count against the phenomenon because the only way to negate it is to hypothesize as to sources, as I have done. Olson is, I believe, clinically correct. Nevertheless, the term middle term is also used in a more practical sense in the very process of forming source hypotheses. In my example of the Beelzebub controversy, for instance, Matthew is sometimes called the middle term, yet Mark and Luke do agree against Matthew in several trivial instances, and such agreements would have to be counted, say, in a statistical study of the pericope. Yet there is still value in noting that the instances of Mark and Luke agreeing against Matthew are trivial when set alongside the instances of Matthew and Mark agreeing against Luke or Matthew and Luke agreeing against Mark. Likewise, in my judgment, there is value in noting that this agreement of Luke and Tacitus against Josephus is trivially explained compared to the other kinds of agreements in the three texts.

I'm not sure what the significance would be of the Testimonium being the "middle term" between Annales 15.44 and a specific selection from Luke-Acts.

The significance of a middle term, provided that there are enough parallels to make the probe worthwhile, is that such a textual relationship is not an expected result of three independent testimonies to the same event (in this case the execution of Jesus). If in any trio of texts one of the three happens to be a true middle term between the others, we are left with four likely scenarios (in what follows let us imagine that B is the middle term between A and C):

  1. A was a source for B, which in turn was a source for C.
  2. C was a source for B, which in turn was a source for A.
  3. B was the source for both A and C.
  4. A and C were the sources for B, which is a conflation of them.

And here is a sampling of the possible arrangements that are rendered unlikely:

  1. B was a source for A, which in turn was a source for C.
  2. C was a source for A, which in turn was a source for B.
  3. B was a source for C, which in turn was a source for A.
  4. A was a source for C, which in turn was a source for B.
  5. A was the source for both B and C.
  6. C was the source for both A and B.
  7. A and B were the sources for C, which is a conflation of them.
  8. B and C were the sources for A, which is a conflation of them.

Another unlikely scenario is that A, B, and C each copied independently from a fourth source (or from the historical event itself), for this arrangement would not explain how A and C managed to avoid agreeing with one another except with the consent of B.

To be able to narrow all the possibilities down to only four basic options, eliminating a host of other arrangements while simultaneously rendering mutual independence unlikely, seems quite a significant step.

I think we might get the same result if we substituted some of the summaries from the speeches in Acts for Lk. 24 (though some of them do mention Pilate).

I am not certain what Olson means here by the same result. Does he mean that once again Josephus would come out as the middle term? Or does he mean that some speeches in Acts would come out as closely related to Josephus as the Emmaus text is? (Or does he mean something else entirely?)

If he means that Josephus would once again come out as the middle term he might well be correct (though the number of parallels would be seriously reduced; see my next paragraph). But then I do not understand the objection. If further comparisons between Lucan and Josephan material retain Josephus as the middle term then we are that much closer to showing that Josephus is a middle term between Tacitus and both of the Lucan volumes on the scale of entire texts. Ironic, given that Olson kicked off his review with the complaint that I was not dealing with whole documents.

Suppose we were to discover that certain speeches in Acts did bear the same kind of relationship to Josephus and Tacitus that the Emmaus passage does. What could that possibly bring to bear against the original relationship? We would have even more confirmation of the same pattern, and it would only go to show either that Josephus combed through the two Lucan texts for material for his Testimonium or that Luke used the Testimonium more than once to construct his speeches. The only thing that would nullify Josephus as the middle term would be an unexplained element shared between Tacitus and Luke against Josephus.

If, however, he means that there are some speeches in Acts that are as closely related to the Testimonium as the Emmaus text is, then I think he is mistaken. G. J. Goldberg, the originator of the hypothesis of a link between the Testimonium and Luke, has a table laid out of summary speeches both in Acts and in other sources compared to Luke and Josephus; the Emmaus text cannot be beat for closeness of correlation. What is especially impressive is the order of correspondences. On my analysis the elements in Josephus that find parallels in Luke are A, B, C, E, G, H, I, ~K, L, M, N, and O. In Luke these parallels fall into exactly the same order except for G, which Luke places between L and M.

But what does this mean? I do not think it is enough to establish literary dependence among any of the passages examined.

Olson may be correct, and there is no remedy for a fool who finds an intricate connection where none exists. But I for one find the apparent interconnections between Josephus, Luke, and Tacitus unusual and worth further investigation.

Second, I think some of the parallels in the chart are a bit strained. It takes a bit of imagination to see the TF's "teacher of men who welcome the truth with joy" as a *parallel* to Tacitus' "deadly superstition." Smith seems to be interpreting it to be a negative response on Tacitus' part to what was in his source. That's a possibilty, but I don't know that it counts as a parallel. (I see Smith has marked it with a ~, which I guess indicates a weak or questionable parallel). But I tend to think that the belief that Jesus was the one who was going to liberate Israel (Lk. 24.21) would indeed have qualified as a destructive superstition in Tacitus' eyes.

Olson is referring to element D on my analysis and has correctly guessed that the tilde ~ next to it indicates a degree of reservation on my part. Element K is also awarded a tilde. Since his second example of subjectivity or selectivity is one that I myself have identified as weak, nothing much needs to be said here. What is of more interest is how he would treat the stronger examples.



Kenneth A. Olson MA, History, University of Maryland PhD Student, Religion, Duke University

My heartfelt thanks to Ken Olson for taking the time to respond to a page on my site.


The challenge to compare the entire texts of Josephus, Luke, and Tacitus is a worthy one. I can think of quite a few agreements between Josephus and Luke against Tacitus: Theudas, the census, Judas the Galilean, the Egyptian, not to mention those within the Emmaus passage and the Testimonium. I can also think of a number of agreements between Josephus and Tacitus against Luke: Caligula and his effigy, the temple prodigies, the ambiguous oracle, not to mention those within Annals 15.44 and the Testimonium.

Besides the spread of Christianity to Rome, already discussed, are there any agreements between Luke and Tacitus against Josephus?