The Jesus Dynasty / James Tabor

Someone wrote me yesterday: “A well known and oft quoted saying is part of a phrase attributed to Benjamin Disraeli and popularized in the U.S. by Mark Twain: ‘There are three kinds of lies: lies, damned lies, and statistics.’ The semi-ironic statement refers to the persuasive power of numbers, and succinctly describes how even accurate statistics can be used to bolster inaccurate arguments.”

As has been widely reported Andrey Feuerverger, Prof. of Statistics at the University of Toronto was asked to run the statistical probabilities on the names in the Talpiot tomb by filmmaker Simcha Jacobovici. It has also been incorrectly reported, and just as widely, with a untoward amount of enthusiasm, that he has essentially “recanted” on his 600 to 1 figure and has conceded he was the victim of a “garbage in, garbage out,” statistical game that means nothing, misled by Jacobovici the filmmaker. This is totally and absolutely false. I am in close touch with Andrey via e-mail and phone, so here are the facts.

One can figure name cluster probabilities in a number of ways and one decision one makes is whether or not to assign any special value of “rarity” to a name. Thus we might have 25% of Jewish women in the 1st century with a form of the name Mary, either in Greek or Aramaic (see Tal Ilan’s listings), but then individual forms of that name, such as Maria, Mariamme, or Mariamne, occur less frequently. Feuerverger did decide to assign certain rarity factors in some cases. These are mathematical decisions, based on the name frequencies, not on identification with any historical figure. Thus he has a rarity factor for the name on the ossuary Mariamenou [he] Mara, but in doing so he is not assuming, in assigning that number, that this is in fact Mary Magdalene. That is what was implied on the Koppel show and it is incorrect. Once one gets the numbers one can then go to the next step, which he chooses to do in his paper, and ask, are these names appropriate or highly appropriate for this or that person, thus Mariamenon for Mary Magdalene, or Jose for Joses the brother of Jesus? This is a second step and that step of course depends on what he is told by the historians who work on the texts. Feuerverger does not assume the identifications or the relationships that might be finally proposed in his numbers, i.e., Jose is the brother, Mary Magdalene is the mother of Jude son of Jesus, wife of Jesus, etc. This has been one of the most widely misrepresented points in this whole discussion.

In the meantime, on the numbers and statistics, one of my own consultants, David, who works in mathematical models and design, tells me this about clusters of names today. Here he is talking about his own family:

Based on very accurate date from the Social Security Administration, we have the probability of F, my dad is 0.4756%, A, my mom is 0.8419%, David, is 3.8751%, P, my wife is 1.544%, D(not David), my son is 1.7303%, and J my son is 0.7071%. The probability you’d find a family plot of six people with our distribution names is 1 in 341,116,556,446. This is the cluster of names alone, without any relationships specified. I objected to him that it sounded too high when he sent this to me I even thought it might be a typo, but the math actually works.

If one added a last name then forget it; the numbers go through the stratosphere. People don’t realize how unique sets of even common names are when it comes to simple probabilities. The lottery is another good example, look at what the odds are for a peculiar five or six number sequence, with millions playing for weeks and not getting the right combination of five slots. The Social Security Web site is available for anyone to type in a given name, date of birth, and figure out percentages of those born with that name in that year. Given the frequency one can then compute the probability of clusters by simple math. Size of graveyard/family plot, city, or area does not matter in terms of the averages. And it would not matter if a aunt named Lucy of whom we had no record was included in the plot or not. Or a Sally Sue for that matter. I am quite sure my name James, my wife L, my son S, and my daughter E, do not occur in that cluster anywhere in Charlotte metro area, with 3 million people. But the names are common. And if you put in my “real” name, which is rare for a birth name, Jimmy, then there is not even a chance. The name James was # 1 in 1946, the year I was born, but Jimmy was #42, an enormous difference. For example, just taking the data from the site above notice the following:

Mary is 5.4891%, but Maria is 0.2187% in a given year.

Taking the calculations for Robert and Mary (the two most popular names that year) we have:

P(Robert) x p(Mary) = (0.054018) x (0.054891) = 0.002965102038

Probability of finding a grave with these two names associated is roughly 1 in 337.

But if you take the more rare Maria: P(Robert) x p(Maria) = (0.054018) x (0.002187) = 0.000118137366

The probability becomes 1 in 8, 464.

Jerusalem ossuary burials are a great sample because they are limited to place and time but we have, both from the ossuaries, and the general onomastics of Palestine, good percentages on the name frequencies. It does not matter how many burials are in the Talpiot tomb, whether 35 or more. The problem is that we only had 6 names, not 20 or 35. To exaggerate the flawed logic, let’s assume that the tomb was even larger – let’s say 50, or 100, or 1,000, or even 50,000. Of course, in a tomb of 50,000 ossuaries (i.e. all of Jerusalem’s dead are in that tomb), we would almost certainly expect to find the names of all of Jesus’ family. Thus, taking an arbitrary tomb size and saying that those names are expected to be found there according to a certain probability is flawed thinking. The problem is that we only had 6 names, not 20 or 35. The Talpiot tomb was not typical, in that 6 of the ten ossuaries (or nine?) were inscribed (60%). Thus, if there were 35 ossuaries in the tomb, on average only 7 (0.2 x 35) would have names inscribed on them. If there were 20 ossuaries in the tomb, only 4 would be inscribed. Thus, even if you were to accept the argument that there were more burials (and that is not entirely clear given where the bones were found according to Gath’s notes) you would still only be able to work with 4 or 7 names in each tomb, which more or less brings it back to the number that we have been using all along, rendering the argument meaningless.

The names that were found are the sample size. They are not part of an arbitrary larger list of names in which they just happen to be found. Another way of looking at this is the following: First, assume that there were 35 inscriptions found in a tomb. Then what is the chance that out of these 35 names, just the names of Jesus’ family happened to be found together in the tomb and all the rest are subsequently unaccounted for? There would have to be an explanation pertaining to why only these were specifically selected to remain in the tomb, while the rest were removed, implying some sort of deliberate manipulation. This lies outside the realm of statistics, and points to the flawed nature of the argument from a statistical point of view.

We learned a few years ago from Camil Fuchs, Prof. of Statistics at Tel Aviv University, and others, regarding the James ossuary, that the most simple relationship cluster such as: “James, son of Joseph, brother of Jesus,” goes against all our intuition, thus, notice, as recently calculated by John Koopmans, a demographer in Ontario, based on Tal Ilan’s data for ratios:

Joseph 231 or 1 in 8.0
Joshua 103 or 1 in 17.9
Jacob 45 or 1 in 40.9

No special rarity value is added to any form of these names, such as Yeshua rather than Yehoshua, or the unusual spelling of Joseph on the ossuary. This could be done but to keep things simple and make a point:

Thus assuming a family of six as a model (and you could change the size, it doesn’t matter):

1. The probability of the name “Joseph” (Yehosef or Jose) occurring is 1 in 8.0;
2. The probability of the name “Jesus” (Joshua) occurring is 1 in 17.9;
3. The probability of the name “James” (Jacob) occurring is 1 in 40.9;
4. The probability of the father-son family relationship between Joseph and James is 1 in 3.0 (either could be father, both could be brothers); and
5. The probability of the brother relationship between James and Jesus (assuming Joseph is the father of at least one) is 1 in 2.0 (Jesus could have been brother or son of James).

By assuming that Joseph will not have two sons with the same name, but could have a son with his own name, the following are the revised probabilities:

1. The probability of the name “Joseph” (Yehosef or Jose) occurring is in 8.0;
2. The probability of the name “James” (Jacob) occurring is 1 in 40.9;
3. The probability of the name “Jesus” (Joshua) occurring is 1 in 17.4;
4. The probability of the father-son family relationship between Joseph and James is 1 in 3.0; and
5. The probability of the brother relationship between James and Jesus (assuming Joseph is the father of at least one) is 1 in 2.

Thus, the probability of a man with the name Joseph having a son with the name James, who is the brother of a man named Jesus is:

1/8.0 x 1/40.9 x 1/17.4 x 1/3 x 1/2 = 1/34,160 families

Therefore, out of 34,160 families (a population of 204,960), this particular combination of names and relationships would occur only once. However, in Jerusalem at the time, this is approximately 4 times the assumed number of families in Jerusalem at the time (50,000). Thus, statistically, the likelihood that this combination of names within a family was unique, is more than certain.

Stephen Pfann and others have made the point that in looking at ossuaries in the Israel State collection we find that 16 of the 72 personal names account for 75% of the inscribed names. The top male names are Simon, Joseph, and Judah, and the top female are Mary, Salome, and Martha, but the frequency ratios in this subset, i.e., the ossuaries found, are amazingly consistent with the much broader base that Tal Ilan has compiled. The rareness of the “cluster” remains, mathematically speaking, as with James/Jesus/Joseph here, or at Talpiot: Jesus son of Joseph, Mary, Mary, Jose/Joseph, Matthew, and Jude son of Joseph. In all of the figures I have asked to be run I have used the common generic names, not the special rare forms (Yose, Mariamenon) in order to be as conservative as possible.

I mention the James ossuary here, not because it has been demonstrated to have come from the Talpiot tomb, that remains unsettled, but I want to make the point that something as simple as that single individual, with those “common” names, is counter-intuitively rare. So, if the inscription is genuine, then we undoubtedly do have the brother of Jesus of Nazareth, but even without the (genine) James ossuary added, the cluster is still very rare–and it goes through the roof if one assigns higher probability values to the special names.

The question to be addressed then, once the statistical probabilities of the cluster are shown to be sufficiently unique, is to ask, as a hypothesis, do these names, in the form they come to us, and the relationships known, offer a “fit” with anything we know of Jesus of Nazareth and his family. More on that shortly.

The past ten days we have heard dozens of biblical scholars, archaeologists, and historians weigh in with opinions on statistics to the press, stating repeatedly, “Interesting, yes, but those names are extremely common, so this tomb means nothing. There would be lots of graves in Jerusalem from this time with these precise names.” I am not sure if such spokespersons intend such statements to be taken as just personal opinion, or whether they are offered as an academic contribution outside one’s field of expertise. Maybe we need “peer review” for statements made in interviews outside of ones field. What I have done is consult with the statisticians as I surely do not think that I have the expertise to speak on this from my training as an historian. I like Shimon Gibson’s answer when he was asked about the names, even with his years of experience and his survey of hundreds of tombs in the Jerusalem area–”I don’t know, they seem somewhat common, but as to the significance of the cluster you would need to consult with a statistician.”

The statisticians with whom I am working are in touch with several others who have proposed alternative models, and I hope that from those exchanges more clarity will emerge over the next week or so on this subject.