The Center of the Torah

Parshat Tazria-Metzora

This week's post is superficially about various claims about where the center of the Torah lay, but more deeply it's about the importance of having a skeptical view on unsubstantiated claims.  This is something that everyone struggles with, as we all find it very easy to accept information that agrees with what we already think is true.  However, when we come across something we disagree with, or that conflicts with what we already know, we tend to look at it more skeptically.  Anyway, we'll see that later.  For now let's get to the topic at hand.


Where is the Torah's Center?

There is a somewhat famous Talmudic passage in Kiddushin 30a, which states (my translation):

"Therefore the Rishonim were called sofrim because they would count (sofrim) all the verses that were in the Torah, and they would say: The vav of gahon (Lev. 11:42) is the midpoint of the letters in the Torah,  darosh darash (Lev 10:16) is the midpoint of words, vehitgalach (Lev. 13:33) is the midpoint of verses....Rav Yosef asked, which half does the vav of gahon belong to?  They said, bring us a Torah and we'll count.  Didn't Rav Bar Bar Hannah say, they did not move until they counted them all?  They said to him, they were knowledgable about full and missing spellings*, we are not (implying they can't count it) Rav Yosef asked, how about the verses, which half does vehitgalach belong to?  At least we can bring a Torah (and count).  With sentences we are also not knowledgable.  In the west (Israel) they split the verse "behold I come before you in a cloud" into three.  There are 5888 verses in the Torah, Psalms has 8 more and Chronicles has 8 less."

 *The full and missing spellings are when you have an extra vav or yud in the vowel.  The grammar of these weren't set fully until the time of the Masoretes (see below).

It's true that there might be some errors in defective spellings and this could alter the movement of a letter by some amount, and make it impossible to determine which half the middle letter belongs to.  Nevertheless, the Rabbis of the Talmud made it clear that they were unwilling or unable to check the previous generation's work.  I don't really blame them.  The Torah is large and it would take a long time to count it by hand.  And even then you're pretty likely to make errors, because these kinds of tasks are ones that humans are bad at.

Nevertheless, there was a later attempt in the 10th century by the Masorete Aharon ben Moshe ben Asher, famous for setting the version of the text with all the cantillation notes we use today.  In this work, called, Sefer Dikdukei ha-Te'amim (Grammar or Analysis of the Accents), he also compiles lists of various quantities in the Torah, including verse amounts, word amounts and letter amounts along with letter counts for each letter.  In his count there are 5845 verses in the Torah, a significant difference from the Talmud's count of 5888.  I should mention that one of the main purposes for Ben Asher's work was to elucidate the grammatical rules of biblical Hebrew and thus standardize the spelling and pronunciation of the text.

Aharon ben Moshe ben Asher also calculated the midpoints and came up with answers that were very different from the Talmud's.  He found the midpoint of letters to be in Lev 8:28, 4829 letters off from the Talmudic count.  The midpoint of words is in 8:26, 743 words off, and the midpoint of verses is Lev. 8:8, 164 verses off.  Furthermore, the skew is very strange, where the Talmud's calculation have the midpoints of words and verses three chapters off, this calculation has them differ by only 18 verses (in the other direction).  For the Talmud's calculation to be correct, the first half of the Torah would have to be full of long sentences with a lot of words.  Even if the Talmudic Rabbis made a bunch of errors in defective spellings and verse distribution, they would at least have realized that something was way off had they actually sat down to to the work.  Aharon ben Moshe ben Asher's results can be found here, along with some apologetics which we'll get to in a bit.

We'll get to these problems in a minute, but since the theme is checking other people's work, why not do it myself?  So I did just that, using unicode text from mechon mamre and python to parse the text.  My calculation for verses is 5844, or one less than Aharon ben Moshe ben Asher.  My letter count is also very close, with 4 more letters.  My word count is considerably different, as I have almost 150 more words.  I also have considerable differences with the alphabetical makeup of the Torah, and some differences in what the middle letter, word and verse is.  Nevertheless, my middle values are much more consistent with Ben Asher's than they are with the Talmud's.  At the end I'll post the output of my code so you can see where the differences are.

Why is the Talmud so Far off?

There are three possibilities for the discrepancies of the midpoints mentioned in the Talmud.  They are.

1) The text that the Tana'im and Amoraim were using were significantly different from the text we have today.  This difference would have to be larger than can be accounted for by defective/full spellings which would not affect word or verse count anyway.

2) The Rabbis in the Talmud (Tana'im) erred in their counts.

3) The Rabbis in the Talmud (Tana'im) actually meant something else entirely and were misinterpreted by the Rabbis of the Gemara (Amoraim).

Answer 1 does show up sometimes in various sites.  For example, this criticism of Judaism, seems to imply that this is the most likely answer.  However, I actually find it a bit unlikely that the Torah in the time of the Talmud would differ by that much compared to current day.  If this were true, we would expect to see many more differences in the verses that the Talmud quotes, and we see hardly any.  I think it's pretty likely that we are working with fundamentally the same text as the Talmudic Rabbis were.

Answer 3 is one commonly used by apologists.  They say that what the Tana'im actually meant with regard to the central letter was that it was really counting the big and small letters of the Torah, in which the vav of gahon is the 8th out of 16 (note that there's no actual "middle" one).  They bolster this claim by noting that the middle letter claimed for Tehilim (Psalms) in the Talmud is also the middle big/small letter, but not the actual middle letter. Similarly, with regard to the middle word of the Torah, it claims that darosh darash is the middle "double word" of the Torah.  We'll examine both these claims in a bit.

I actually think the apologists are on to something, but not quite what they intend.  As I'll explain now.  I actually think the second approach is the most likely, and I think it goes a bit further than the Tana'im just making an error.  I think this goes very much to the way Chazal approached questions of this type, questions with definite answers.  Instead of verifying it in the scientific way we think of today, they just assumed an answer they heard was correct, or just chose one that seemed good and then it became the answer due to what is essentially an argument from authority.  The answers quoted in the Talmudic passage all seem good.  Why?  It's exactly what the apologists explain, except they get it backwards.

The vav of gahon is a big letter.  There are a few such large and small letters in the Torah, the actual amount differs between versions.  One of them is the vav in question, and another one (the next, 9th) is the gimel in vehitgalach the first word in the supposed middle sentence.  It's very likely to me that the thinking of Chazal went something like this:

"This vav is big, why?  Well it's roughly in the middle of the text, it must be the middle letter!"  "But wait, this gimel is also big, and also near the middle, maybe it's the middle letter."  "Well, the gimel is in the first word of the sentence so maybe it's the middle verse instead."  "Cool, and what about a middle word?  There's no other weird letters nearby."  "How about the nice double word darosh darash wish essentially means to really learn about, that sounds like a great middle for the Torah."

Admittedly, this is rampant speculation, but if you've read enough Talmud, this type of argumentation should be familiar enough to you to be plausible.

Returning to the apologist explanation, it's worth looking at it a little more deeply.  It's clear that the Amora'im who are doing the conversing in the Mishnah are not talking about big and small letters.  The conversation of them not being able to count because of confusion over extra vavs and yuds makes no sense in that context.  So this must mean that the Amora'im misunderstood what the Tana'im where saying about the middle letter.  However, this is also problematic if you're one of those people who relies on the integrity of the oral tradition.  If they make such a blatant misunderstanding here, where else did they err?

Rather, this form of apologetics is a blatant way of saving face for Chazal.  But another question to ask is are the apologists even correct?  We'll look at the double word issue, since the large and small letters differ between versions.  From the apologists site, the claim is that there are 77 double words of which darosh darash is the 39th, or middle one.  They give two examples for double words, lech lecha, and avraham avraham.  The first example is important, since lech lecha is actually two different words entirely that just happened to be spelled the same way.  So we're looking for consecutive words that are spelled the same, but may not be pronounced the same.  I searched through this and counted 91, of which darosh darash was the 47th (not the middle).

All the verses with double letters will be included on a separate page here.

The count of 77 is wrong.  No doubt about it.  To make darosh darash the middle set, we would have to discount 2 instances in the first half.  Perhaps you can argue this by eliminating Exod. 7:17 which is actually one that straddles verses (verse markers post-date the Torah though, although based on the passage in the Talmud they had some form of marking).  7:16 ends in coh and 7:17 begins in coh; Perhaps you can also eliminate Exod 15:25 which is sham sam, two words that are spelled with the same written characters, but you can differ between the shin in the first word and the sin in the second.  I'm not sure how they got the count of 77. Even if you discount all the words that appear across broken phrases, these would be: Gen 6:9, 11:10, 11:27, 25:19, Lev 13:38, Num 14:34 it does not remove enough, and the middle count would then be more wrong.  So it's not clear to me that the answer offered here to validate Chazal's claims makes any sense anyway, and it's also done by people who were also very sloppy in their calculations.

What Else Have I Learned From This?

So the cool thing about these sort of exercises is you actually see new things that you didn't expect.  One thing I noticed, when looking through all the double words was the relative rarity of them in Devarim (Deuteronomy).  There are only 7 double words in Devarim, and the 7th appears in Ha'azinu, which is sort of a special case being poetry.  Furthermore, when going through the words for the middle three books, a very large number come from the phrase ish ish, which literally means "man man" and idiomatically means, "each or every person".  This phrase is completely absent in Deuteronomy, despite having a lot of similar legislative style content.  Similarly, several of the double words in Vayikra (Leviticus 11:41-43) come from the "kosher" passage which talks about unkosher animals using the phrase hasheretz hashoretz, a phrase that's completely absent when Devarim talks about kosher animals, (it actually doesn't mention insects at all in those kashrut laws!)  When people talk about differences in the way the books were written, this is what they are talking about.

Again, the full output of my code including verse counts, and the location of the double letters can be found here.  If you actually want the python code used for analysis, send me an email and I'll be happy to provide it.