With all this information, we can test a hypothesis regarding the omissions of the table of contents, which is that the omissions are also likely to be omissions from the text of the War, while that which is present in the table of contents is also more likely to be present in the War.
In the Text of the Antiquities, Mentioned in the Table, and in the War (105 sections)
In the Text of the Antiquities, Mentioned in the Table, and not in the War (111 sections)
In the Text of the Antiquities, Not Mentioned in the Table, and in the War (41 sections)
In the Text of the Antiquities, Not Mentioned in the Table, and not in the War (122 sections)
Total: 379 sections according to the divisions of the Niese text of the Greek for Antiquities 18
And we can create a table.
Sections of Antiquities Mentioned in the Table | Sections of Antiquities Not Mentioned in the Table | |
Also Mentioned in the War | 105 | 41 |
Not Also Mentioned in the War | 111 | 122 |
Chi-square value: 21.5846
degrees of freedom: 1
p-value: < 0.0001
The above, however, isn't necessarily valid, because using smaller sections will create more "points of data," which will tend to cause the p-value to be small but without necessarily providing a secure basis for further reasoning, if the unit size of what is being counted is arbitrarily small. We need to use units of data collection that are not arbitrary or overly small, as the Niese sections could be considered in this context. So we will do this one more time using the unit of "four sections" (instead of just one), which corresponds to the median length (in sections) of a Nodet sense unit. This results in an evaluation along roughly the same level of granularity as the table of contents itself, as this would tend to provide a more robust result.
"4x Sections" of Antiquities Mentioned in the Table | "4x Sections" of Antiquities Not Mentioned in the Table | |
Also Mentioned in the War | 26 | 10 |
Not Also Mentioned in the War | 28 | 31 |
Chi-square value: 5.5894
degrees of freedom: 1
p-value: 0.0181
The groups are significantly different at the p-value < 5% significance level.
What does this mean? It means that the null hypothesis, that the parts of the Antiquities not mentioned in the table are just like those which are, in respect to this other quality of being also mentioned (or not) in War 2, is more or less falsified.
Instead, there seems to be a relationship between material that is not mentioned in the table and material that is not mentioned in War 2 (as the majority, about 75%, of the material not in the table is also not in War 2 but only 25% of the material not in the table is in War 2, which is considerably different than the roughly 50-50 ratio that we see for material that is mentioned in the table).
This can be said to go some way towards establishing the likelihood of Nodet's hypothesis:
"Sometimes, the order of the titles does not match the narrative flow of Ant. 18 ... Several titles of the summary are not elaborated in Ant. 18 ... Conversely, some titles correspond to large sections ... Moreover, the summary ignores many portions of Ant. 18, almost one third of the book. ... Now if we hypothesize that this summary is not a table of contents written after the book but a preliminary sketch, composed before its final redaction, all the previous difficulties disappear. A comparison of the summary with War 2 shows no discrepancy in the order of the topics, but the previous narrative is interlarded with new pieces of documentation." (Etienne Nodet, "Josephus and Discrepant Sources," in Josephus, Interpretation and History, p. 269)
Lastly, there is the brute fact that 43% of the material, counted by sections, of Antiquities 18 is not covered by the outline (the table of contents), which means that there is very little cogency in an argument from the lack of mention in the table to the absence of a section from the text, in general terms. Combined with the impression of several scholars that the original compilation of the Greek summaries appears to have been performed by a Jewish person** and plausibly even Josephus or one of his assistants, the argument from the omission from the Greek table of contents (of Jesus or John) cannot be sustained without fresh arguments that the table of contents is primarily of Christian origin.
** I have a blog post coming shortly, partly on this.