- At 10:39 PM 2/28/2007, Doug Weller wrote:

>Hi Ariel,

I saw this :( It was written by someone with very little

>

>Wednesday, February 28, 2007, 7:55:27 PM, you wrote:

>

> > At 06:02 PM 2/28/2007, Eliot Braun wrote:

> >>[...] Even the 'statistical' indications of the chances of one tomb

> >>producing names of so many people from the NT story of Jesus are suspect.

>

> > Yes they are. I'm waiting to see the details. If anyone has a URL,

> > please let me know.

>

>The Discover Channel website has this:

understanding of statistics. And if this is taken from the text of

the movie, then the movie won't be much help either.

But I'll use the quote to point the things I'll be looking if and

when there'll be a more serious publication of the statistical aspect

>"But Andrey Feuerverger, professor of statistics and mathematics at

As it is, this paragraph doesn't make much sense. It seems to say

>the University of Toronto, recently conducted a study addressing the

>probabilities that will soon be published in a leading statistical journal.

>

>Feuerverger multiplied the instances that each name appeared during

>the tomb's time period with the instances of every other name. He

>initially found "Jesus Son of Joseph" appeared once out of 190

>times, Mariamne appeared once out of 160 times and so on.

that Feuerverger claims a 1/190 frequency for "Jesus son of Joseph"

and a 1/160 frequency for "Mariamne", which he then multiplies

getting a 1/30400 expected frequency, or probability, for the

combination of the two. I.e. some 33 such cases in a population of a

million (which I think is a reasonable assumption for that period?).

I'm not sure what the numbers 190 and 160 represent and they seem

small compared to what I've seen on the number of names evidenced for

that period. And multiplying the frequency of a single name with the

frequency of a double one has to be shown to be valid.

You multiply frequencies, or probabilities, only when these

frequencies are independent of each other within a common set. E.g.

if you multiply the frequency of green apples by the frequency of

white grapes you don't predict correctly the frequency of green

apples and white grapes on one plant. E.g. if you multiply the

frequency of tall people (above a certain height) with the frequency

of short people (below a certain height) you don't predict correctly

the frequency of tall-short couples.

So I need a detailed exposition of the procedure and the reasoning to

understand what Feuerverger is doing.

But I'll take some elements out of this paragraph to illustrate an

important point.

There's a 1/6 probability of getting a specific number when rolling a

dice, and a 1/36 (1/6*1/6) probability of getting a specific pair of

numbers when rolling a pair of dice. If the frequency of number pairs

that is evidenced is not 1/36 for each of number pairs there can be

two reasons. 1st, the evidence is not a representative sample, i.e.

the pair of dice were not rolled enough times. 2nd, the results of

rolling the dice are not randomly distributed, e.g. the dice are loaded.

Or, if the single dice rolls result in a random distribution (1/6

frequency for each number) but the pair rolls don't, the reason could

be magnets in the dice that attract certain sides of the dice to each

other, skewing the random distribution of pair rolls without

affecting the single dice rolls.

We have as evidence a set of names from that period. The frequency in

which each specific name appears in that set is its probability. If

you multiply the frequency of "Jesus" with the frequency of "Joseph"

you get the expected frequency of these two names appearing together.

If that predicted frequency is "more or less" (see below) the

evidenced frequency in which "Jesus son of Joseph" and "Joseph son of

Jesus" appear in the set of names, all is well (statistically). If

not, either the sample is not representative or double names are not

random combinations of single names and simple frequency

multiplication is not the correct method to predict the frequencies

of double names.

The general point is that the existing evidence has to be evaluated,

assessed, in order to determine if it is usable, statistically, and

if so, what are the appropriate statistical tools.

This is something basic, and it will have to be detailed in a

scientific publication. I don't know enough about population and name

statistics to be able to criticize Feuerverger. What I'll be looking

for is reactions from experts in such statistics to see if they

accept Feuerverger's assessment of the evidence and how to use it

statistically.

For example, the expertise in the field of population statistics

should indicate what is the statistical test that determines if the

evidenced frequency is indeed "more or less" the predicted one.

Statistical results are usually accompanied by such tests to

determine how significant, trustable, these results are.

>To be conservative, he next divided the resulting numbers by 25

No idea what that means.

>percent, a statistical standard, and further divided the results by

>1,000 to attempt to account for all tombs ?

>even those that have not been uncovered ? that could have existed in

That actually makes sense. A calculation of the probability of name

>first century Jerusalem.

groups appearing together in one tomb is based on an assumption that

the entire population can be divided into such groups, i.e. could

have had such tombs. Whether they did or not, or whether we know if

they did or not, is not important. What is important is if we can

assume that they could have had such tombs if they wanted to, and I

think that's a reasonable assumption.

>The study concludes that the odds are at least 600 to 1 in favor of

This is the really problematic paragraph.

>the Talpiot Tomb being the Jesus Family Tomb. In other words, the

>conclusion works 599 times out of 600."

First, I'm not really sure what is claimed because the first sentence

seems to imply a probability while the second seems to apply a degree

of confidence in a calculated probability. If it is a degree of

confidence (chi squared etc), the type of test and the result to

which it is applied will have to be specified in a scientific publication.

If it is a probability, it means that Feuerverger claims that the

probability of there being only one such set of names is 599/600. In

other words, that the probability of there being two, or more, such

sets of names is 1/600. If the population is a million with 10 people

families (i.e. "10 people per tomb") that means 100,000 families,

100,000 potential tombs. Given this number of potential tombs the

claim is that there will be 167 cases of duplicate sets of such names.

This is not a result I'd use to support a claim that this tomb is

"very likely" the one and only one ... That's a "duh?" result.

For comparison, in forensic DNA results the test is usually if the

probability of a combination of DNA markers is lower than one per the

entire possible population of suspects. E.g. in Israel (population of

~8 million including tourists etc), the result for a male suspect

should be lower than 1/4,000,000. With the assumptions above, a more

serious claim would be of a probability of two such sets of names

being lower than 1/100,000 or "odds" of at least 100,000 to 1. Even

if you cram 100 people in each tomb you'd still need odds of 10,000

to 1, not 600 to 1.

And it's very important to see if the calculation included the

frequency of the name Judah, or not. Every added name, every increase

in the number of names in the set, will lower the probability of

there being several such sets.

We all know that the evidence is not clear about what names should be

included in the "Jesus Family" set, and some of that evidence

actually claims there shouldn't be physical remains of Jesus in a

tomb at all. But let's take the largest group, i.e. including John &

Mary Magdalene. That leaves us with the son, Judah. As far as I know

there is no evidence that claims explicitly that Jesus had a son, let

alone what his name was.

Putting aside obvious problems with defining what names should be in

the "Jesus Family" set, a possible argument could be:

- The probability of there being only one family with the "Jesus

Family" set of names (without Judah) is very high.

- Therefore there's a very high probability that Jesus had a son, Judah.

But if the "very high" probability is reached by including the Judah

frequency in the calculation, the argument becomes invalid, circular.

So the bottom line is that in order to understand the claim and why

it made I will have to wait for Feuerverger's paper, or some

equivalent publication.

Ariel.

[100% bona fide dilettante ... delecto ergo sum!]

---

Ariel L. Szczupak

AMIS-JLM (Ricercar Ltd.)

POB 4707, Jerusalem, Israel 91401

Phone: +972-2-5619660 Fax: +972-2-5634203

ane.als@... - Hi Folks,

I want to thank Doug for this analysis he gave the forum a few days ago.

If there is any follow-up, or if it has been posted on a blog, please let us know.

The only other interesting discussion of the statistics that I have seen

was by Joe D'Mello and was on the NTGateway blog of Mark Goodacre.

While they are in some senses complementary I believe that Doug has done

the best overall job of really trying to wrestle directly with the statistical issues,

writing with a real savvy and flair and comprehension. D'Mello emphasized more

the *interpretation* of the existing numbers given by the film makers rather than trying

to find a proper and accurate *formulation*. And the latter is ultimately far more important.

The fly in the ointment of much of the discussion (and was never even remotely

discussed in all the brouhaha) is the difficulty of what is sometimes called "post facto

probability" calculations. This pops up in all sorts of venues and defining the proper

input parameters is very dicey since there already is a bullesye target (what actually

happened) staring you in the face. So it is very difficult to go back to *before* what

happened.. happened.. and develop the proper formulations and questions that give

a sound resultant calculation number.

To give an example of a post facto probability (one that was never really

done, afaik) in the public arena that actually made a lot of sense and could

have well-defined input .. when Hillary made her $100g on futures trading

one might be able to go back and say .. given a-b-c .. what is the likelihood

that she could gain that amount by *luck* as opposed to a little "help" from

the person entering the profits and loses (ie. perhaps doing a little winnowing

out of losses to help her accrue gains.)

However the variables in the ossuary situation are a lot more difficult. I will

give an example of one. Jose is an extraordinary name in one sense, in that

it did not show up on ossuaries much. On the other hand it is all over the

Talmud and other ancient Hebraic realms and was apparently simply much like

"Joe" is today to Joseph, a shorthand, a nickname. John Gill discusses this as well.

And most graves today do not have a Joe, they have Joseph. So if you hit a grave site

today with "Joe" how easy would it be for you to figure out what Joseph or Joe was

involved ?.. (answer, very difficult .. especially as most people with one name are

sometimes called the other).

Difficulties like this are all over the ossuary calculations.

Enough for now. Thanks, Doug.

Shalom,

Steven Avery

Queens, NY

http://groups.yahoo.com/group/Messianic_Apologetic

Doug Weller ->I just found this statistical analysis - like the other one, it assumes the names are correct.

>I have computed that the probability of the tomb in the Talpiot

>district of Jerusalem being the family tomb of the Jesus of Nazareth

>is at least 12 to 1 *against*. Apparently the makers of the movie

>calculated the probability that more than one family living in ancient

>Jerusalem would produce a cluster of names like the ones discovered in

>the tomb in the Talpiot district of Jerusalem, and found that this

>probability is very small and that therefore this must be the tomb of

>Jesus of Nazareth's family. But I think they asked the wrong question.

>The right question is: How many families living in ancient Jerusalem

>would produce a cluster of names in a tomb that would appear to be as

>similar to the names in Jesus' family as the cluster of names actually

>found? And the answer is that more than 12 families would have

>produced such remarkable cluster of names in a tomb.

>

>Here is how I computed this number. According to the gospels Jesus'

>family consisted of Joseph and Mary, Jesus, and four male brothers of

>Jesus named James, Joseph, Simon, and Judas (besides unnamed female

>siblings). We also know the approximate frequency of names in ancient

>Palestine. According to

>http://benwitherington.blogspot.com/2007/02/jesus-tomb-titanic-talpiot-tomb-theory.html

>

>these are: 9.2%, 8.3%, 6.2% and 3.8% for Simon, Joseph, Judas and

>Jesus respectively for male names, and Mary's name frequency is a

>whooping 21.3% for female names. The tomb discovered in Talpiot

>contained 10 ossuaries, of which 6 carried inscriptions. The relevant

>inscriptions here are "Jesus son of Joseph", "Mary", "Mary", and

>"Joseph". These inscriptions were in different languages and used

>different forms for these names, but that's about it. (see:

>http://www.sciencedaily.com/releases/2007/02/070228135009.htm ) I read

>somewhere that first century Jerusalem had about 50,000 inhabitants.

>As the period in question spanned various generations I used a

>population of 10,000 families.

>

>I wrote a computer program simulation that actually randomly produced

>10,000 families of 10 members each (keeping the right name

>frequencies) and then proceeded to compute the following average

>numbers:

>

>111 families would have a Jesus son of Joseph. I understand that in

>fact other ossuaries have been found with the inscription "Jesus son

>of Joseph".

>

>75 families would moreover have at least one Mary.

>

>71 families would moreover have one more name that belongs to Jesus'

>family, be it Joseph, Simon or Judas, or maybe a second Mary

>(supposedly Mary Magdalene).

>

>43 families would have two more such names. One of such clusters might

>be: [Jesus son of Joseph, Mary, Mary, Joseph], as is the one

>discovered in the Talpiot tomb.

>

>And 16 families would have three more such names. Here is in detail

>the very first family case my simulation produced: An unnamed (i.e.

>with no relevant names) couple have three children: a daughter Mary, a

>son Joseph, and an unnamed second daughter. Their daughter Mary

>marries Simon and produces an unnamed daughter. Their son Joseph

>marries an unnamed wife and produces two children, Jesus and Mary.

>Voila: A family cluster of 10 whose tomb might have had inscriptions

>"Jesus son of Joseph", "Mary", "Joseph", "Mary", and "Simon" - all

>names related to the circle of Jesus of Nazareth, but this is not

>Jesus Christ's family. Nevertheless this cluster would appear to be

>even more statistically conspicuous than the one discovered in the

>Talpiot tomb.

>

>The film producers have tested the DNA of one of the Mary's and

>discovered it is not maternally related to Jesus' DNA. I compute that

>adding this condition we still get 12 families. Here is the very first

>such case my simulation produced: A Jesus (whose father was named

>Joseph) marries a Mary and has 4 children: Simon, Jesus, and two more unnamed ones. Their son Simon marries a second Mary but have no children that would be buried in the family tomb. One unnamed daughter marries Joseph and has a daughter Mary. That's the second Mary who is also not maternally related to "Jesus son of Joseph". - So, any of these 12 families might have produced tomb even more conspicuous than the one found, but at most one of these families could be Jesus'. Hence the chance of the Talpiot tomb being Jesus is less than 1/12.

>

>Further: Taking into account that Jesus' family was not from

>Jerusalem, that his family was too poor to afford a family tomb, that

>if Jesus' bones were put in an ossuary one would expect the ossuary

>itself or the inscription on it to be more special in some way, and

>that if Jesus' body was buried in a tomb to decompose and then put in

>an ossuary then probably somebody would have found out back then when

>so much was made of Jesus' bodily ascension to heaven - taking all

>that into account the probability of the Talpiot tomb being of Jesus

>is much less than 12 to 1 against. Finally, if the movie producers

>really believed that this was Jesus' tomb one would expect that they

>would have asked neutral professional archeologists to evaluate their

>evidence or argumentation - which they haven't done.

>

>There is some more arguments, such as the "James son of Joseph brother of Jesus" ossuary (see:

>http://en.wikipedia.org/wiki/James_Ossuary )

>having come from this same tomb, but the archaeologist who first

>studied the tomb flatly denies it (see:

>http://www.jpost.com/servlet/Satellite?c=JPArticle&cid=1171894527185&pagename=JPost%2FJPArticle%2FShowFull )

>and the archeological report on the Talpiot tomb counts six ossuaries

>with inscriptions, all accounted for (go to

>http://dsc.discovery.com/convergence/tomb/explore/explore.html

>and

>click on "Enter the Tomb", then on "Download Documents" and then on

>"Download PDF").

>

>I did the above computations in a hurry and it's possible that I have

>committed some mistake. If you send me an email to

>dianelos@... I will gladly send you a copy of the program I

>wrote, so that you can check it yourself. It's written in Pascal, and

>it's a simple 150 lines program that any programmer can read.

>

>Doug Weller