A lecture by Professor Eliyahu Rips
On this page we present a lecture given about 1985.
The Russian transcript and the English translation were
made directly from an audio tape by independent professionals.
We have tried to make both of them as accurate as possible,
but please notify us if you find any problems.
The Word file needs Cyrillic font support (1251 encoding).
The Postscript and PDF files are self-contained.
English translation: HTML.
The significance of this evidence
These notes were written by Brendan McKay.
Central place amongst the evidence claimed for hidden codes in the
Torah is held by the famous rabbis experiments. According to
the official history, the first
experiment was conducted in 1986, and involved a set of 34 rabbis.
The rules, and the method of analysis, were all determined in advance,
and a list of names and appellations for each of the rabbis was
prepared by an independent expert. These claims are extremely
important, as the validity of the experimental results depends
entirely on their correctness.
The importance of this new evidence is that it appears to contradict
the official story in a number of key aspects. In fact, if the lecture
on the tape contains the literal truth, we can positively state that
the official story is a fiction and that the experiment was not properly
conducted. Since the second famous rabbis experiment
(the one with 32 rabbis published in Statistical Science) also stands
or falls on the correctness of the official story, we would be
entitled to cast that aside also.
A reply to these charges has been written by
Doron Witztum, dated January 9, 1998. It is reproduced below so that
we can refer to it in our comments. Eliyahu Rips has declined to
How many rabbis?
The official story is that the first ever experiment on the famous
rabbis was conducted using the 34 rabbis whose entries in the Margaliot
encyclopedia occupied 3 or more columns, and a date of birth or
death is given. However, the lecture says otherwise:
[Rips:] We took a list of gdolei Israel,
great Talmudic sages. The selection criteria? We took the
four-volume encyclopedia published by Mordechai Margaliot.
But the list exceeded out technical capabilities. So we
narrowed the list down in the following manner: we looked
through the entire encyclopedia, and if an article took up
three or more pages we chose it. Essentially, we ended
up with great figures. We can't guarantee that we picked every
great figure; there may have been a great man whose description
was slightly shorter, or a less great man whose description was
slightly longer; but all in all we did come up with truly great
personalities. We had an objective list which turned out to
contain some 19-20 names. Next, for the second word...
[Audience:] Since there are only 20 names, I wouldn't
mind hearing them.
[Rips:] I will tell you in a moment. Rashi, Rambam, Rabat...
[Audience:] Over three pages of text?
[Rips:] More than three...the entry in the encyclopedia took
more than three pages of text. It is called Encyclopedia of
the Sages of Israel. It's a four-volume encyclopedia by Mordechai
Margaliot. Shach, Eybeschuetz, Rambam, Maharam Me'Rothenburg,
Ha'Besht, Ha'Taz, Ha'Ramchal, Ohr Ha'Chaim Ha'Kadosh, Ha'Maharshal,
Rashi, Yosef Caro, Ha'Gaon Me'Vilna, Rabbi Eliyahu Bahur, and
Rabbi Avraham Ibn-Ezra. For the second word, we want to take
some clearly established biographical details.
Each page in the Margaliot encyclopedia contains two columns, so
3 pages is twice as much as 3 columns. The initial thought that Rips
merely mis-spoke "pages" instead of "columns" is easily dismissed:
All the rabbis mentioned have more than 3 pages. Moreover, the
number of rabbis (from WRR's list of 34) that have more than 3 pages
is truly "19-20". There are 19 clearly over the boundary, and
one which has three pages only if the bold heading is counted.
These would be impressive coincidences indeed if Rips had really
meant to say "3 columns".
In any case, Witztum does not deny that Rips spoke of an
experiment on 19-20 rabbis. Rather, he asks us to believe that
the experiment never actually happened.
Might Rips have really been discussing 34 rabbis?
The scenario that Witztum puts forward as an explanation is that
Rips suggested an experiment with 19-20 rabbis, then Witztum
decided all by himself to expand it to 34 rabbis. He didn't tell
Rips of his decision for the months that it took for the checking
of the dates, for Havlin to compile all the appellations, and for all
the computer work to be finished. Witztum didn't even tell Rips when
he gave him the results of the experiment (which in the lecture
includes such details as appellations, dates, distances and letter
arrays). Rips then went around lecturing on it, completely unaware
that Witztum had changed the experiment behind his back.
Just in case someone is willing to believe Witztum's story, an
important conclusion is worth mentioning. The whole credibility
of the famous rabbis experiment relies very heavily on the personal
and mathematical reputation of Eliyahu Rips. Except for that,
it would never have received even the cautious semi-support it did
from senior mathematicians and it would certainly never have been
published in a scientific journal. Now Doron Witztum, who has few
of his own, tells us that Rips had so little to do with the experiment
that he didn't even know how many rabbis were used.
Doesn't Rips' data imply that he had the results for 34 rabbis?
Witztum claims that two items in Rips' lecture match the
results for the full first list of 34 rabbis (and, presumably,
not the smaller list of 19-20 rabbis). These are:
- There were "some 150 queries".
- The score was "7 sigma... or let's say 6 sigma".
Given that Rips says very clearly that he had 19-20 rabbis, we
need a pretty good reason to not take him at his word.
The primary difficulty with the above two items is that
there is no reason
to assume that Rips had the same appellations and used the same
calculation method as were later published.
There are in fact
grounds for doubting both assumptions. Rips names three appellations
which were not in the later official list, and he also gives some
counts of word pairs that do not agree with the later list.
Furthermore, he refers to both 125 and 343 perturbations,
whereas the later preprints only used 125. He also refers to
"many" distances coming "first in 125", but there are no 1/125
distances at all in the 1986 preprint.
Looking in more detail, we find that there is not even a case
to answer. Rips defines queries in terms of word pairs, even
giving some examples, but to get a count of 150 for the list of 34
rabbis one must only include successful word pairs.
(Some word pairs fail because the distance between them is not defined.)
Even if Rips was really
counting queries in that manner (despite his words) it is very
easy to achieve that number for the 19-20 rabbis. All that would
be needed is to add appellations similar to the extra appellations that
he mentions, plus 2 or 3 more. From the 1986 preprint we know that
WRR could even contemplate using words that were not
appellations at all (see The pseudo-appellation
ha'Chaim), so this would have given no difficulty.
The case of the "7 sigma... or let's say 6 sigma" is even
less conclusive. If we take the appellations from the 1986 preprint,
and the distances listed there, the 20 rabbis achieve a score of
6.42 sigma and the 34 rabbis achieve a score of 7.17 sigma.
If anything, the 20 rabbis match Rips' statement better.
[Note added May 2001:] Also note that the measurement system given
in the 1986 differs in at least
two places from the later system.
Together these improve the performance of the 20 rabbis by about
0.5 sigma. This makes Rip's statement match even better.
How were the names and appellations collected?
The official story is that Professor Shlomo Havlin of Bar-Ilan
University was presented with the list of rabbis (the headings of
their entries in the Margaliot encyclopedia) and that he alone
produced the list of names and appellations subsequently used in
How well does the lecture match that?
[Rips:] There may be various ways of writing a name.
We took every
possible variation we could think of. For instance, Ha'gaon...
or Eliyahu... or, say, Rabbi Eliyah. If any additional
variation comes to mind, we must include it. We simply took
every possible variant that we considered reasonable.
For every name we took a number of possible spellings.
Havlin is not mentioned in the lecture. More importantly, Rips'
description is incompatible with independent data collection.
It is especially impossible to reconcile this description with the
later claims that a considerable number of complex and arbitrary rules
were applied to select appellations, such as the rule that they had to
be "pronouncable". (Also note that the appellation "Eliyahu" is not
present in WRR's later list of appellations.)
Unless Witztum or Havlin choose to produce some, it appears there
is no evidence, dating to before 1990 or so, that Havlin prepared
the appellations by himself. (Recall that the two lists of rabbis
appeared in 1986 and 1987, respectively.) In fact, none of the four
earliest known documents even claim that Havlin prepared the
appellations. These were:
- The lecture of Rips, circa 1985, that we are discussing on this
- The 1986 preprint of WRR which contained the first list of rabbis.
- The 1987 preprint of WRR which contained the second list of rabbis.
- Witztum's 1989 book, which devoted two chapters to the rabbis
The two preprints are particularly instructive, as Havlin is
mentioned in each. Recall that Havlin is supposed to have prepared
all of the data, which is surely as large a contribution as any
of the authors made. If so, why is he given the briefest of thanks
for "valuable advices", hidden amongst 14 other persons acknowledged?
To illustrate this point, we have provided
scans of the Acknowledgement pages of
the two preprints - the only places he is mentioned in those
This is the full text of a reply made by Doron Witztum to our
initial publication of the Rips lecture. Most of the main points
have been answered above, but we will also make a few comments below.
A. The analysis of proximities between the names of Famous Rabbis
and their birth and death dates was first carried out on an original
list of 34 names (now known as "the first list").
B. Prof. Eliyahu Rips is aware of this response, but his position
is that "this kind of investigative 'truth-seeking' should not be
C. I read over sections of the English transcript that were
publicized by Gil Kalai on Dec. 31, '97. A number of things are clear
from this tape:
1. It is obvious that Eliyahu Rips was referring to the outcome
of a test performed on the list of 34 Rabbis (the first list), and
not to some other list. This is clear from the results that Prof.
Rips quoted on the transcript:
(i) "Thus we have obtained a list containing some 150 queries."
(ii) "Therefore, on the basis of all this, we have a
significant... 7 sigma... or let's say 6 sigma... never mind..."
These details correspond exactly to the outcome we obtained on
the first list of 34 Rabbis, as published in our pre-print.
In fact they correspond as well or better to the list of 19-20
rabbis, as we showed above.
[Witztum:] 2. There is however, a contradiction in the transcript
on the one hand, Eliyahu Rips refers (according to Kalai's English
translation of the transcript of the talk) to a group of 20 Rabbis,
each of whose entry in the encyclopedia is three or more pages,
and on the other hand he quotes results (above) that do not at all
match to these 20 Rabbis. The source of the contradiction is very
straightforward: Eliyahu Rips, who is the one to originally suggest
the experiment, and also who suggested using a list of Rabbis with
three or more pages of text, is not the one who actually
carried out the experiment, and was not familiar with all the details.
As was described in Document 2 ("Bar Hillel and Bar Natan Inquire:
Witztum and Rips Respond"), I was the one responsible for carrying
out the experiment, and it was my decision to compile a list of Rabbis,
whose entry in the encyclopedia was three or more columns of text
(the first list). It seems that Eliyahu gave this recorded lecture
a short time after the experiment had already been run, and was
appraised only of the final results, -not with the various details
of the experiment on hand.
3. In any case, I now have decided to run the experiment on this
very list of 20 Rabbis. The results are as follows: using our
standard probability measures of P1 and P2, the odds are less that
1 in a million- absolutely, a very strong result! This should make
it clear that there was no need to widen the list.
By these same measures, the full set of 34 rabbis does over 200 times
better! We will let readers decide for themselves if that constituted
a plausible motive.
[Witztum:] If anybody is still not satisfied, lest he claim that
attempted to decrease our odds by expanding the list
(from 20 Rabbis to 34), there is a much simpler way we could have
obtained a better result: had we subtly quantified P1 as a
measurement of all proximities that fell within the range of 0.5
(instead of 0.2 which we used; see the beginning of chapter two of
Document 4)- we would have achieved a result for these twenty Rabbis
that is even better than what we did find for the 34 we used!
Witztum's claim is simply false. By the only measure of
success mentioned by Rips, a count of "sigmas" as he calls them,
the cutoff at 0.5 for 20 rabbis does worse than the cutoff
at 0.2 for 34 rabbis. (This is true both with and without
ha'Chaim.) In fact, with ha'Chaim included, a cutoff
of 0.2 does better for the 20 rabbis than a cutoff of 0.5 does!
[Witztum:] 4. Regarding the claim that Prof. Havlin is not mentioned
on the tape:
I asked Moshe Zeldman to do a word search of the whole transcript,
and indeed, Prof. Havlin's name is not mentioned. It also turns out
that Yoav Rosenberg's name (one of the co-authors of the published
paper, along with me and Eliyahu Rips) is also never mentioned.
And even the few times where my name is mentioned are not at all in
connection with the Rabbis experiment. What does this tell us?
That maybe Yoav Rosenberg didn't really write the computer
program? Maybe I also really had nothing to do with it?
Anyone familiar with the two pre-prints we released, will see that
Prof. Havlin (and y"l Yaakov Orbach o.b.m.) are thanked in the
Acknowledgments section for their advice. For all the valuable
professional assistance they rendered, it is still simply a
technical contribution that is minor in comparison to the phenomenon
Only when the final text was being drawn up for publication in
Statistical Science, the one who edited it for us decided to also
include Prof. Havlin's name in the body of the text. (In fact,
I heard a tape of one of Eliyahu's lectures dated even after the
preprint was released, and he still makes no mention of Prof.
D. It is interesting to note the timing of this "tape libel".
It followed on the heels of a major refutation of these critic's
(pseudo-) scientific attempts to discredit our research that appeared
in Statistical Science.
A detailed disproof of this "refutation" will be published.
We have seen that Witztum's best attempt at explaining away the
damaging evidence of Rips' lecture was a complete failure. This merely
serves to reinforce our conclusion that the lecture provides a very
strong case that the official history of the famous rabbis experiment
The Official History
In Dec 1996, Witztum and Rips described the early history of the
famous rabbis experiment in the following terms.
The complete document can be found at
In a meeting which took place in May '85, Eliyahu Rips suggested
that we move from using intuitive impressions to quantitative
measurement. To this end we defined the function which is designated
in our paper as c(w,w'). A computer program for calculating this
quantitative value was prepared over the course of that summer by
Yoav Rosenberg, and it was tested by measuring convergences from
the already existing pool of examples.
In the autumn of '85 Eliyahu suggested that we check ourselves by
examining an a priori list of word pairs. He proposed using pairs
of the type -- "personality-date (of birth or death)" -- because the
conceptual relationship between them is particularly well defined.
He proposed using outstanding Rabbinic personalities because if the
phenomenon under investigation was genuine, it seemed reasonable to
assume that outstanding Rabbinic personalities would receive fuller
treatment in the "hidden text" than others might. Until that point,
neither phenomena (i) nor (ii) had been investigated in relationship
to this topic, therefore we had no prior knowledge whether any
tendency towards "personality-date" proximities existed. On the
other hand, one example of this type of relationship had emerged
from Doron's previous work, which was an elaboration of a concept
of Rabbi Weissmandl, o.b.m., concerning ELS's with skip lengths of
special significance. (In this example the name "Herzl" was paired
with his birth date, the 10th of Iyar).
It was Doron's task to see to the preparation of the list and to
carry out the measurements. Since he had no prior experience in
the requisite areas of bibliography and linguistics, he turned for
assistance to the linguist Yaakov Orbach, o.b.m. (who passed away
last year), who directed him to the Encyclopedia of Great Men in
Israel, by Dr. Mordechai Margalioth. From this source Doron
selected the "greatest" personalities using the following criteria:
His list was comprised of those whose entry consisted of at least
three columns of text, and for whom a date of birth and/or death
was cited. These were the 34 personalities whose names and
appelations were to comprise the list (now known as the first list).
Doron approached Professor Havlin and gave him the list of the 34
personalities, and requested that he prepare a list of names and
appellations which, in his opinion, characterize these Sages.
Professor Havlin prepared the list of names in the winter of '85-86.
Then an experiment was run on this first list. When this experiment
proved successful, we decided to write a paper. This paper was
published as a preprint on Oct. 6, '86 (the "White Preprint").
Witztum, Rips and Rosenberg have repeated essentially the same
chronology of their work on many occasions. The most recent are:
- Prof Yisrael Aumann responded in writing to a lecture of
Prof Maya Bar-Hillel, after consulting Rips.
Aumann was the editor who handled WRR's paper for PNAS.
gives a detailed chronology in agreement
with that described by Witztum and Rips in the letter quoted above.
- After a talk by Maya Bar Hillel at the Center of Rationality at
the Hebrew University of Jerusalem, spring 1996, Prof. Michael Rabin,
a mathematician and computer scientist from the Hebrew University
and Harvard University asked Doron Witztum if any statistical tests
(even a single one) were made prior to the tests on the final lists
of appellations. Doron replied that no such tests were made.
- On October 21, 1997, Prof Robert Haralick reported to the TCODE
mailing list the following assurance he had received in person from
Prof Rips: "... with regard to the
Statistical Science work, there have been no experiments other than
those discussed in the 1986, 1987 and 1994 articles."
- On December 18, 1997, Prof Gil Kalai publically asked the following
question of Doron Witztum: "Did Doron perform any investigation
of the closeness of rabbis to their dates prior to the experiment
on the list of 34 rabbis that is reported in his first preprint?"
Three days later, Witztum gave the following reply via Rabbi Moshe
Zeldman: "No. The first investigation of the closeness of
Rabbis to their dates was done on the original list of 34 Rabbis."