Wednesday, March 09, 2011

How likely is it that Jesus' sisters were called Mary and Salome?

My recent NT Pod on the sisters of Jesus mentions Richard Bauckham's suggestion that, on the basis of statistical studies of female names, there is a 50% likelihood that Jesus' sisters were called Mary and Salome. He makes the claim in three works* in the following passage:
So is there any degree of probability that the tradition known to the Protevangelium of James, the Gospel of Philip and the source used by Epiphanius correctly preserved the actual names of two sisters of Jesus? A recent study of the recorded names of Jewish women in Palestine in the period 330 BCE-200 CE finds that, although the 247 women whose names are known bore 68 different names in all, 61 of these 247 women were called Salome (including its longer version Salomezion) and 58 were called Mary (Mariamme or Maria). In other words, these two names account for 47.7% of the women. Every second Palestinian Jewish woman must have been called either Salome or Mary. Individual sources for the names also show high percentages of these two names, making it likely that the sample is in this respect representative . . .

In the light of this statistical finding, it seems that the tradition which gives the names Mary and Salome to Jesus' sisters has a 50% chance of being correct, even if it was not based on historical memory!
But surely the inference Bauckham makes here is incorrect.  Assuming the accuracy of the statistics, and assuming two sisters (about which more anon), there is only a (nearly) 50% chance of at least ONE of the sisters being called Mary OR Salome.  Assuming only two sisters, and assuming the accuracy of the data, the chance of their being called Mary AND Salome is more like 5.8% (i.e. 61/247 x 58/247), isn't it?**  Or am I forgetting something?

* Richard Bauckham, "Salome the Sister of Jesus, Salome the Disciple of Jesus, and the Secret Gospel of Mark", Novum Testamentum 33/3 (Jul., 1991): 245-275 (253-4)
* Richard Bauckham, Jude and the Relatives of Jesus in the Early Church (Edinburgh: T & T Clark, 1990; 3rd edition: London & New York, T & T Clark, 2004): 43
* Richard Bauckham, Gospel Women: Studies of the Named Women in the Gospels (Grand Rapids: Eerdmans, 2002): 233-4.

** It occurs to me that if the first sister is named Mary, the second is unlikely also to be called Mary, so we should perhaps reduce the pool to 186 names, in which case the calculation would be 61/247 x 58/186 = 7.7%.  is that right?

Update: After a useful discussion in the comments thread (see below), we have consensus that the correct figure is in fact 15.3%.  What I had forgotten was that the first child, in these calculations, could be called *either* Mary *or* Salome.  So it is still a much lower figure than Bauckham's 50% figure, but it's a bit higher than the 7.7% I was thinking about before.


Adam's Apple said...

Hey Mark,
Love it when you go deep for 15 minutes on a specific topic like this one. Taking 15 minutes to delve into a specific question is brilliant. No one else is doing it. It's easier to remember, especially when I'm driving or walking. When you try to cover big topics I tend to drift and get lost. For example, the podcasts that covered the differences between Mark and Matthew or Mark and Luke, find me drifting from the swelter of multiplying fractals. Also, please don't hesitate to delve into the scholarly literature. I think you can interpret some of those discussions and bring them to a wider audience. For example, I had to laugh at the podcast on whether or not Paul uses the word "testicle"--oh heavens--I've forgotten the text. But thanks. Keep it up!

Mark Goodacre said...

Thanks, Dan. Appreciated the comments. I think you are right, actually, that this kind of topic works best in the NT Pod format. That testicle one (1 Cor. 11.15) actually morphed into an article in the end that is scheduled to come out in the Journal of Biblical Literature later this year. I have quite a lot on Jesus' sisters that may make it into a similar piece, if I have the patience.

Thanks again for your helpful comments.

Adam's Apple said...

On another note, correct my impression. I think it must be easier for an orthodox Christian to accept Q than to accept that Luke is changing Matthew around in such drastic ways. It would go like this. Matthew and Luke usually amplified or elaborated on Mark, and then took their other material from Q. But on your view, we have to say that Luke just changed the hell out of Matthew, setting up a potential conflict between the two gospels as religious authorities. What do you think?

Mark Goodacre said...

Yes, I think that that is about right. The only caveat would be that there might be occasions when Luke has additional materials that he uses in his redaction of Matthew. But I don't think that happens very often. And, after all, they already appear to have that attitude with Mark.

Stefanos Mihalios said...

Thank you Mark,

This was a great pod; very informative! By the way, it seems to me that the chance of Jesus’ sisters being called Mary AND Salome is 11.6%, rather than 5.8%. The chance doubles because both sisters could be either Mary or Salome (one chance for Mary AND Salome, and one for Salome AND Mary). Nonetheless, your main argument still stands.

Richard Fellows said...

Mark, your calculation is more or less correct, although you have to double your answer, as Stefanos says.

We might also want to account for the fact that a woman is more likely to have a Hebrew/Aramaic name if her family members also have Hebrew/Aramaic names (rather than Latin or Greek names).

You mention the possibility that two sisters would not be given the same name and this has interesting implications. Each group of sisters could then have no more than I Salome and 1 Mary. Since about a quarter of women were called Salome and a quarter were called Mary, the number of daughters born to each couple could not be more than about 4 on average.

Also, if parents were reluctant to give the same name to two daughters, they would likely call their first daughters "Mary" and "Salome" and, having done so, would have to call on less common names for any subsequent daughters. Thus these popular names would have been more common for the first daughter than for the fifth, for example (since the name would probably have already been taken by the time the fifth came along). These two names would then have been very popular indeed for first daughters.

The name Salome is more common in the diaspora than Bauckham implies in Gospel Women p234, I think. The statistics in part III of Ilan's Lexicon were not available when Bauckham wrote.

Matthew C. Baldwin said...

The statistics model you use to criticize Bauckham would not be accurate.

We don't multiply the percentages together, as we would with, say, withdrawing cards from a deck.

Yes, draw one card and the likelyhood of drawing a spade from a fresh deck of cards is 1/4 (25%); draw two cards and the likelihood of drawing a second spade is 1/4 x 12/52 (5.7%) (or something like that).

But human parents don't work that way, and neither do names.

First, there is no "deck" of names (there is only the inexhaustible resource of language itself). So the available names are not reduced by naming, the way a deck of cards is reduced by drawing.

And second, names are not selected from a randomly shuffled pile (names are selected by the human will in a cultural matrix).

Here's how I would think about it. Of known individual females from ancient Palestine roughly 25% are named Mary and 25% are named Salome.

Assuming they belonged to this demographic, it is reasonable to presume that Jesus' parents would have had a 50% chance of naming his first sister either Mary or Salome. We all agree on that.

But then, once one name is given, a second sister would also have to be named. She won't be given the same name as her sister, but there is a fairly strong likelihood that she will be given the next most popular name. Yes it is no longer a 50% chance. Now it is something like a 33% chance.

In the model of naming the second girl that I would propose it is more like rolling a 3 sided die. 1 side says "Salome" (or "Mary") and the other two sides each say "some other name."

So the overall chance of both sisters having the two most popular names in that context would be 50% x 33% or 16.5%


Another hopeless liar using statistics.

Mark Goodacre said...

Thanks for your kind words. On the figures, once one daughter is called Mary or Salome, though, the second one is not *also* going to be called that name, are they? But I should have allowed for the first daughter to have been called either Mary or Salome, and that increases the figure overall, as Matthew notes. Does that seem right to you?

Mark Goodacre said...

Thanks for your helpful comments, Richard. I see what you mean. And yes, Bauckham is working with the older figures.

Mark Goodacre said...

Thanks for your helpful response, Matthew. It's good to be put right on the Maths! On your first point, you note "the available names are not reduced by naming". But aren't they? Once you've named daughter 1 Mary, you don't also name daughter 2 Mary? On the second point, I agree, of course, that "names are not selected from a randomly shuffled pile" but absent the kind of detailed cultural data that we would need (e.g. how many first daughters were called Mary, how many second, how many two-daughter-families were Mary+Salome etc.), don't we have to do the stats as if it is random? That's not a rhetorical question but a genuine one -- I am keen to learn here.

Thanks for spotting that the first sister could be called *either* Mary *or* Salome and that I should have built that into my little calculation. D'Oh! So figure one should be 47.7%. You give a figure of 33% for daughter 2 being called Mary or Salome, presumably roughly based on the 61 and 58 figures. Is that right? In other words, my error was essentially to forget that daughter 1 could have been called *either* Mary *or* Salome for this calculation, right?

agiering said...

There are two scenarios here. One is that the first child is named Salome and then the second is named Mary. The other is that the first is named Mary and the second is named Salome. These two probabilities must be added together.

First scenario: there's a 61/247 chance that the first child is named Salome. Then there's a 58/186 chance that the next is Mary.

Second scenario: there's a 58/247 chance that the first child is named Mary. Then there's a 61/186 chance that the second is a Salome.

So we get (61/247)(58/186) + (68/247)(61/186). I'm getting about a 16.7 percent chance.

I think my logic is similar to Matthew's, only I used more exact figures (not 25 percent odds for each name).


Mark Goodacre said...

Slight adjustment needed on the figures, agiering -- second scenario would be 61/189, not 186, right? This gives:

((61 / 247) * (58 / 186)) + ((58 / 247) * (61 / 189)) = 0.152797903

So 15.3%, a litle lower than your figure.

agiering said...

Yes, Mark, thanks for catching that error. Sounds good to me.

Stefanos Mihalios said...

Yes, the final 15.3% seems right to me as well.

Gail Stevens Shourds said...

This conversation is quite fascinating and enjoyable. It occurs to me, however (and please correct me if my assumptions are wrong), that it is unlikely that Mary and Joseph (or Mary and anyone) would have had a daughter named Mary since to give a child the name of someone still living would be unthinkable. Depending on the date of death of Mary's mother, Anne, wouldn't it be more likely considering Jewish naming tradition, that at least one of Jesus' sisters was named Anne? Personally, statistics notwithstanding, I go with Anne and Salome.

Unknown said...

So is there any evidence of jesus having another sister?

Unknown said...

I wonder if Bauckham was/is aware of the tradition--though late, coming from Patriarch Nicephorus (early ninth cent.) in his Chronography--that Jesus had two sisters, named Esther and Tamar.