Doing the Numbers on Talpiot
Given all the hype and heat currently circulating regarding the Talpiot tomb, and particularly, how and whether the statistics done by Feuerverger are valid, I wanted to clarify one point that seems to be at the heart of most all the misunderstanding. I am not a statistician and contrary to anything you might read I did not “calculate” anything. I have consulted with statisticians, including Feuerverger, so here I want to say something about method rather than mathematical results. This is not a case of “garbage in, garbage out,” and yes, you can “prove anything by statistics,” depending on assumptions. But that said, it does not mean it is impossible to statistically test the assertion that this cluster of names is highly probable/common, and thus meaningless.
Just about every scholar who has commented publicly on the thesis that the names in the Talpiot have high probability of being identified with the Jesus of Nazareth family has made the dismissive point that the names are common. Indeed, this seems to be the most “common” assertion made this week by a host of non-statisticians who are stepping out of their expertise to offer views to the press. I don’t think any of us disagree with that, i.e., that the names are common. I know that Jacobovici and Cameron don’t disagree with that. However, the question is, what about the cluster of names per se, as well as their known relationships, in this particular tomb? There one’s intuition can be quite misleading. We have pretty good records of name frequencies from ossuaries (including but not limited to the Rahmani catalogue), from the broad range of sourses that Tal Ilan has been surveying, and from Rachel Hachlili. The remarkable thing is that the basic sorts of percentage and ratio figures for name frequencies in Palestine in the 2nd Temple period, such as those Camil Fuchs calculated, seem to hold fairly steady with the various bodies of data (i.e. literary, papyri, ossuaries, ostracon, inscriptions).
The first step, as I see things, if we are going to have a clean and consistent method, is to run the statistics on the names themselves, in this case the Talpiot tomb, without any reference at all to Jesus of Nazareth, Mary Magdalene or anyone else–just the names. What are the probabilities that these names, in the relationships specified (Yeshua bar Yehosef, Judah bar Yeshua)? There one has to decide whether to handle the names generically (count a special form Yosa as just Joseph, etc.), or include the aspect of “rarity.” I think it is best, as a preliminary step, to take the names generically, i.e., a Jesus son of Joseph, two Marys, a Jude son of Jesus, a Matthew, and a Joseph. The question then becomes what is the probability of this cluster of names and the specified relationships based on frequency ratios. This addresses the question of whether or not the cluster is common, i.e., probable. It has nothing at all to do with Jesus of Nazareth.
A second step is entirely separate as I see it. That is the matter of whether or not the specific names as we have them in the tomb, and whatever we can know of the relationships (two instances of father/son and a mDNA result that shows no maternal relationship between the Yeshua and one of the Marys), exhibit any probable relationship to anything we know or can say about the family of Jesus. This is not a statistical question, but one for the historian and biblical scholar. My own view is yes, but demonstrating that here is not the point of this post. I will write something up soon in that regard that I find highly persuasive.
Now it is possible, given some potential “fit” between what we know of Jesus of Nazareth and his family and this cluster of names, to factor in a greater rarity for some of the names, thus Joseph is not just any old 14% Joseph, it is a rare form Jose, so you give it a different number, and it is also associated with a name we know for Jesus’ second brother, etc. That is not necessary but that is what Feuerverger means when he says I took what I was given by the historians. It is not his field or his specialty to decide such thing–i.e., Yose is rare, Mariamene is rare, and they seem to fit the Jesus family, and can potentially provide a “fit” with Jesus’ brother and possibly Mary Magdalene, but at that point he is not responsible for the identification, only the numbers.
I have read initial draft versions of Feuerverger’s paper, which he will publish in good time, and I am perfectly comfortable with his 600/1 conclusion, which is actually quite low in that he added in all sorts of compensations for possible errors in the assumptions. What is important to clarify is that Andrey should not be made foolish as if he is saying “in all probability this is the holy family” as if the conclusion is based on his numbers alone. The “fit” with the Jesus family has to be based on other grounds, namely the historical records we have. He has done his work well.
P.S. Several of my colleagues, including Mark Goodacre on his Blog, have asked whether one would need to “discount” the statistical numbers for names in the Talpiot tomb that do not seem in any way to fit the family of Jesus, namely Matthew and Jude son of Jesus. The two are different cases in that Jude is said to be the son of Jesus, while Matthew we know nothing about, but the answer is no, in that this is not a statistical matter but one for the historian. In other words, the statistician could do that, if the historian said there is something about that name we could give a negative factor to–i.e, Jesus could not possibly have had a child so let’s substract some probability value. But that closes the possibility that if by other means we can say with probability that Jesus had a son, we could learn then from this archaeological find that he did. Feuerverger did not include Matthew or Jude because he was working on what we know to get the numbers, namely, names we could identify in the historical record. BTW, he also factored in for names not there, that might be expected. He was really quite careful and thorough. As for Matthew, clearly, any family tomb of this type, whether that of Jesus or anyone else, can have individuals not related by blood to the main clan, and in that sense any name could be there, and not affect the core calculations.
