CRACKING"THE BIBLE CODE"

RobertC. Newman

BiblicalTheological Seminary

Forsome years now, several Israeli mathematicians and rabbis have beeninvestigating the idea that the Bible contains "code words" hidden inits text.[1] These words are claimed to validate thedivine authorship of the Bible and the rabbinic understanding of it, and togive us information about the future, especially about Israel and the end ofthe age. Recently, some of thismaterial has been published in a popular presentation for lay readers byMichael Drosnin in his book The Bible Code.[2]

Usingcomputers to search the text, these investigators look for hidden words thatare spelled by letters spaced at equal distances in the text. In some cases the letters are immediatelyadjacent; in others, they are thousands of letters apart.[3] In one example, Drosnin finds the nameof former Israeli Prime Minister Yitzhak Rabin in close proximity to the phrase"assassin will assassinate." The context also contains the year Rabin was killed, and the assassin'sname![4] In spite of this rather impressivesuccess, some of the predictions Drosnin has found have not come true.[5]

Theproposed "Bible Code," like the claim that the Gospel message isdisplayed in the constellations,[6]is attractive to many who are Bible believers. Obviously, God is capable of doing something of this sortshould He choose to. The questionis, do we have any good evidence that He has chosen to do so?

BiblicalProblems with "The Bible Code"

Thereare good biblical reasons for thinking that this is not God's work. First of all, the Bible does not make much provision for true prophecythat doesn't come true. Deuteronomy 18 clearly indicates that unfulfilled prophecy is a mark offalse prophets, and this is a characteristic feature of the biblical teachingelsewhere (1 Kings 22, Isaiah 42-46, Jeremiah 28). The book of Jonah does allow for the possibility that aprophecy might be postponed if those for whom disaster is predicted shouldrepent.[7] Presumably, something similar mighthappen with a prophecy of blessing if the recipients became arrogant (seePaul's warnings in Rom 11:17-22). I would not, however, want to make any large use of this kind ofqualification, given the Bible's emphasis on fulfillment and on God'sforeknowledge and control of history. In any case, for the Bible Code's failed prediction of the assassination of Benjamin Netanyahu, thereis no evidence of repentance.[8]

Anotherfeature also makes me suspicious. The hidden messages of the Bible Code are typically single, disjoint words or very short phrases. These are naturally rather ambiguous,and there is no context available by which to resolve the ambiguity. This is not typical of biblicalprophecies, even though these sometimes have ambiguities too.

Avery serious problem with the Bible Codeis related to this. How do we knowthat in a particular passage we have found all the coded words that arenecessary to understand the prophecy? For instance, after some of the "Bible Code" prophecies didnot come true, Drosnin found the word "delayed" in several of them![9] But what good is a prophecy for warningor guidance if somewhere in its vicinity we later find the isolated word"not" qualifying it? This resembles the work of astrologers and modern-day occult prophetsrather than that of the biblical prophets![10]

Chanceand Human Manipulation?

Butif this is not God's work, whose is it? Neither humans nor angels can tell the future with consistent accuracy,and in any case, there is no reason to believe the Bible texts have beenmodified since they were first written down by the ancient Jewishprophets. I suggest that thephenomena are just a combination of chance and human manipulation. Let's see.

Inthinking that the Bible Code is the workof human manipulation, I am not here claiming that Rips, Drosnin, and friendshave changed the text to make things work out. That is too easily checked, and they have been prettystraightforward in showing us where they found their words and in offering toprovide copies of their text and search program.

WhatI am thinking is this: Both theprime code word (usually displayed vertically on the diagrams in the BibleCode), and the secondary propheticcode-words which cluster around it, are words that are found by searching thetext using various fixed spacings between letters. But with a long enough text and not too long a search word,one is bound to find nearly any code word selected (or some suitableabbreviation or synonym thereof), as we will show below. So there is nothing to finding codewords. Around the prime code worda "context" is displayed, typically some 700-1500 characters. This is not always a real context,however, as each line shown may be hundreds or thousands of letters away fromthe line displayed above or below it. So the question is, can we find some word or words hidden in this arrayof 700-1500 characters which seem to relate to the prime word in some predictiveway, even without supernatural intervention? I believe we can, by a combination of chance andmanipulation. Let's look at thechances first.

Chance?

Thechance of finding these code words is enormously increased by severaltechniques employed in The Bible Code. All spacing between words is removed, thoughthis was never a characteristic of ancient Hebrew manuscripts, so far as I know.[11]Finding code words is also more likely in Hebrew than in most Westernlanguages, as the investigators read it (as is usual in modern Hebrew newspapers)without vowel points. The vowelpoints used in the standard printed Hebrew Bibles are removed, and Drosnin andcompany supply their own vowels when they read their code words. In this, at least, the investigatorsare following the earliest Hebrew Bible manuscripts, which were not writtenwith vowel points either. Ofcourse it is going considerably further to say that you can supply just anyvowels of your own. The choice ofvowel points in a given passage is strongly constrained by the context, so herethe investigators are assuming that one of the rules of this hidden code is tosupply whatever vowels are necessary to make it work!

Likewiseit is easy to find encoded numbers in the Hebrew language, since Hebrew, unlikewestern languages, uses (all) the letters of its own alphabet to representnumbers. Thus it will be easy tofind specific years given as part of the alleged prophecies.

Intrying to evaluate the claims of the Bible Code, we are at a serious disadvantage if we are not fluent in modern orancient Hebrew. We can look at oneof Drosnin's diagrams and have no idea what else the letters not marked might spell. However, a good test for usHebrew-challenged types will be to try to do the same thing with some famoustext in English (say, the Declaration of Independence, US Constitution,Gettysburg Address, or something from Shakespeare). After some experience with these materials, we should beable to tell whether or not there is anything really peculiar in the biblicaltext of the sort alleged by Drosnin and the others.

Forour test, we chose the text of Lincoln's Gettysburg Address.[12] We converted it to something like thecompressed and unpointed Hebrew text Drosnin and friends use by changing allthe letters to capitals and by deleting all spacing, punctuation, and vowels(a,e,i,o,u, but not y or w). As aresult, we ended up with an alphabet of 21 letters, and the condensedGettysburg Address turned out to have a text length of almost exactly 700letters!

Afterspending several hours searching the text visually (a tedious process!), wewrote a program to do the hard work automatically.[13] The program reads in whatever text weare asking it to search, then it asks us to supply a search word. Starting at the beginning of the text,the program searches for a match to the first letter in the search word. When this match has been found, itcontinues looking forward from that point till it finds a match for the secondletter. When this second match isfound, the distance between the two matches sets the spacing distance for therest of the letters of the coded word (as they must be equidistant), and thesucceeding search letters are checked against the text at that particularspacing. If successful, thestarting position and spacing of the coded word are recorded. Successful or not, the program nextgoes back and looks for another match for the second letter further along inthe text. When all second-lettermatches have been tested, the first letter of the search word is moved along toits next match in the text, and the process is repeated until the whole texthas been searched. Finally theprogram reports its results for that search word. Searches for the code word spelled backwards (a commonphenomenon in the Bible Code) are notdone automatically, but a new search word can be entered which is the first onereversed.

Someresults for the Gettysburg Address are displayed below:

FRSCRNDSVNYRSGRFTHRSBRGHTFRTHNTHSCNTNNTNWNTNCNCVDNLBRTYNDDDCTDTTHP

RPSTNTHTLLMNRCRTDQLNWWRNGGDNGRTCVLWRTSTNGWHTHRTHTNTNRNYNTNSCNCVDND

SDDCTDCNLNGNDRWRMTNGRTBTTLFLDFTHTWRWHVCMTDDCTPRTNFTHTFLDSFNLRSTNGP

LCFRTHSWHHRGVTHRLVSTHTTHTNTNMGHTLVTSLTGTHRFTTNGNDPRPRTHTWSHLDDTHSB

TNLRGRSNSWCNNTDDCTWCNNTCNSCRTWCNNTHLLWTHSGRNDTHBRVMNLVNGNDDDWHSTRG

GLDHRHVCNSCRTDTFRBVRPRPWRTDDRDTRCTTHWRLDWLLLTTLNTNRLNGRMMBRWHTWSYH

RBTTCNNVRFRGTWHTTHYDDHRTSFRSTHLVNGRTHRTBDDCTDHRTTHNFNSHDWRKWHCHTHY

WHFGHTHRHVTHSFRSNBLYDVNCDTSRTHRFRSTBHRDDCTDTTHGRTTSKRMNNGBFRSTHTFR

MTHSHNRDDDWTKNCRSDDVTNTTHTCSFRWHCHTHYGVTHLSTFLLMSRFDVTNTHTWHRHGHLY

RSLVTHTTHSDDSHLLNTHVDDNVNTHTTHSNTNNDRGDSHLLHVNWBRTHFFRDMNDTHTGVRNM

NTFTHPPLBYTHPPLFRTHPPLSHLLNTPRSHFRMTHRTH

Figure1. Gettysburg Address (No Vowels)

andthe Civil War Situation

Witha line-length of 66 letters, we get "Lincoln" (spelled phonetically)right in the middle! Above him are"Civil War" and "Battlefield." Both to Lincoln's right and left we find the name of theConfederate commander at Gettysburg, "Gen. R. Lee." The Union commander at Gettysburg,General "G. Meade," appears just below the righthand Lee. Finally, "Abraham" is seen inthe lower right corner of the text. All of these, naturally, are spelled without vowels! (LNCN, CVLWR,BTTLFLD, GNRL, GMD, BRHM)

Nowthe two longest code words, "Civil War" (5 letters) and"Battlefield" (7 letters) are actually a part of the regular text ofLincoln's speech C not code C and so their appearance is not particularlysurprising. But the same can besaid for Drosnin's "assassin that will assassinate,"[14]which is the regular text of Deut 4:42, a regulation concerning the cities ofrefuge to which the "killer that has killed" might escape.

Iadmit that I was disappointed not to be able to find the exact spelling of"Lincoln" (LNCLN) in our text, but this is mainly due to the textbeing so short. The Bible Code faces no such problem, having all the resources ofthe Old Testament text available for search.

Itis important that we think through the probabilities or statisticalexpectations involved in what we have found in our search of the GettysburgAddress. We have a 21-letteralphabet and a 700-letter text. Ignoring the fact that the various letters of the alphabet occur withrather different frequencies, we can still get a "seat of the pants"estimate by assuming all letters occur with equal frequencies.[15] Then the chance that a particularletter will occur in our text is 700 divided by 21, or 33. We should expect about 33 occurrencesof a given letter in our text, on average.

Fora two-letter search word, either letter can occur in either order at any spacing,so the probability is essentially 33 times 33. The pair will occur some 1089 times in the text!

Butfor search words of three letters and longer, the rule that the code wordsappear with equidistant letter-spacing now comes into play to make matches muchrarer. The third letter must beexactly the same distance from the second as the second is from the first, andit must be on the opposite side of the second letter from the first. Thus a particular location in the textmust contain a particular letter for a match C only one chance in 21. The resulting probability for a three letter search-word is1089 divided by 21, or 52; so we can expect about 52 matches of a given3-letter search word in a 700-letter text.[16]

Fora four-letter search word, this is divided by 21 again, giving a probability ofabout 2.5. For a 5-letter word,divide again by 21, giving .12 C about one chance in eight that a given five-lettersearch word will find a match in a 700-letter text. Not surprising, then, that we couldn't find the particular5-letter combination LNCLN in our text.

Asa formula for this rough calculation, we can say that the probability (orexpected number) N of occurrences of a given search word of length n in a textof T letters using an alphabet of A letters is:

N_n= T² / Aⁿ, n = 2 or more (1)

Nowlet's apply the same reasoning to the Hebrew alphabet and the text of theHebrew Old Testament. The Hebrewalphabet has 22 letters. We listthe calculations for a 700-letter text and a 1,000-letter text:

Table 1: Expected Matches for Hebrew Texts
Length Search Word	700-Letter Text	1,000-Letter Text
One Letter	32	45
Two	1010	2065
Three	46	94
Four	2	4.3
Five	.095 (9.5 x 10^-2)	.19
Six	4.3 x 10^-3	8.8 x 10^-3
Seven	2.0 x 10^-4	4.0 x 10^-4
Eight	8.9 x 10^-6	1.8 x 10^-5

OurEnglish text, the Gettysburg Address, was 700 letters long with the vowelsremoved. How big is the OldTestament text? We don't need tohave an exact count of letters, such as was made by the ancient Hebrew scribes;a rough estimate will be good enough for our purposes. I have here a text of the Hebrew Biblewith virtually no footnotes.[17] Taking a typical line, there are some33 letters per line, and 20-25 lines per page, so 650-750 letters perpage. Let's use the number 700 perpage, making our Gettysburg Address just the length of one page in this printededition of the Hebrew Bible. Whenwe count the pages, we find that Genesis takes up 84 pages, the five books ofMoses 335 pages, and the whole Old Testament 1360 pages. Since the number of expected matches of2-letter and longer combinations increases with the square of the text length(as we saw in multiplying 33 by 33, above), a calculation of the number ofpages squared (last column, below) will also be helpful.

Table 2: Text Length - Hebrew Bible
Section	Pages	Letters	Pages Squared
Genesis	84	58,800	7056
Pentateuch	335	234,500	1.12 x 10⁵
Old Testament	1360	952,000	1.85 x 10⁶

If we now compare thelast column of Table 2 with the 700-letter text (= single page) column of Table1, we see that the expected matches for 7-letter search words (per 700-letterpage) is 2 x 10^-4 and the squared length of Genesis is about 7 x 10³,so the two multiplied together C the chance of finding a given 7-letter match inGenesis C is 1.4. We should typically expect to find 7-letter matches in Genesis, if ourformula is not too rough. For8-letter matches, we would probably need the Pentateuch to search, or even thewhole Old Testament. And, indeed,Drosnin points out that the 8-letter combination spelling "YitzhaqRabin" in Hebrew occurs only once in the Hebrew Bible, just about what wewould expect.

Oncewe have located our long "prime code word" (of say six to eightletters) to be displayed, what size code words can we expect to find in itsvicinity? If we have only 700letters of text displayed as the "context" of our prime code word,then our typical cluster words will have four letters or less, as we saw above.If our context is, say, 1400 letters (twice as long) then the chances of anyparticular combination appearing with be four times larger. In general, we can find out the minimumtext length in which to expect to commonly find words of length n by solvingequation (1), above, for the text length T, with N_n set = 1 and A to22, for various values of n:

T= A^n/2 = 22^n/2 (2)

The results are given inTable 3:

Table 3: Text Needed for Search Word
Letters in Search Word	Length of Text
3	103
4	484
5	2270
6	10,678
7	49,943
8	234,256
9	1,098,758

Notice,in our example of the Gettysburg Address (text length 700), that most of thecode words which we displayed were four letters in length (LNCN ="Lincoln"; GNRL = "Gen R Lee"; BRHM = "Abraham"),in accordance with the largest sorts of words one will easily find in a text ofthat length. We could have chosento display many words of 3 letters or less, but we only chose to display one(GMD = "G Meade"). Wewill come back to this point when we discuss the matter of human manipulation,of which more later.

Wedid, however, find two even longer words CVLWR = "Civil War" (5letters) and BTTLFLD = "Battlefield" (7 letters), which are longerthan we would have antecedently expected. In general, one can expect to find some longer code words even in a textof 700 letters. If ourcalculations for these probabilities are not too far off, perhaps one in eightof the 5-letter search words we try will be successful. But both of our longer words came fromthe regular text of the Gettysburg Address, just as Drosnin got "assassinwho will assassinate" (11 letters) from the regular text of Deut4:42. Naturally, we can expect toget long meaningful phrases from the regular text of our "context"since this by definition is a meaningful, connected text. So, in our example, we could haveexpanded CVLWR = "Civil War" to GRTCVLWR = "Great CivilWar" (8 letters) and BTTLFLD = "Battlefield" to GRTBTTLFLD ="Great Battlefield" (10 letters) or even GRTBTTLFLDFTHTWR = GreatBattlefield of that War" (16 letters)!

Letus now look at the decoding diagrams given in the Bible Code to see what sorts of words Drosnin has found. We will tabulate the length of thewords which Drosnin marks and comments on. Our table 4 will distinguish between coded words (labelled"Spaced"), which have some space between letters in the original textthough they are often aligned vertically in the diagrams with no spacing, andregular text words (labelled "Unspaced"), which have no spacesbetween letters in the original text, even though some of these will have thewords divided differently than in the Bible, or will be read backwards. To give some idea what is going on ineach diagram, we will also spell out (sometimes abbreviated) the main code wordin each.

Table 4: Length of Diagrammed Words in Bible Code
Page	Spaced (Code Word)	Unspaced
15	8 Yitzhaq Rabin	11
16 = 15	8 Yitzhaq Rabin	14, 6, 4
17	7, 6, 4, 4 Rabin assas.
19	7, 4, 4, 3 Fire on 3 Shev	5, 3
27 = 15	8 Yitzhaq Rabin
28 = 15	8 Yitzhaq Rabin	11
29 = 15	8, 4 Yitzhaq Rabin	(11)
32	7 Clinton	5
33	7 Watergate	16
34	8, 5, 4 Econ. Collapse	5
35	7, 5 Man on Moon
36	8, 4, 3 Shoemaker-Levy	4
37 … 36	8, 3 Shoemaker-Levy
40	5, 4 Hitler	7, 5
47	6, 4, 4, 3 Shakespeare
48(1)	7, 6 Wright Brothers
48(2)	6, 4, 4 Edison
49(1)	6, 5 Newton
49(2)	9, 3 Einstein	10, 8, 6
54 = 15	8 Yitzhaq Rabin	11, 11
55	9 Holocaust of Israel	6
56	10 Atomic Holocaust	5
58	9 The Next War	10, 4, 3
62	8 Libyan Artillery	5
64 = 56	10, 5 Atomic Holocaust	3, 3, 3
66	8 Atomic Atilleryman	9, 5
70	7, 4, 3 Autobus	5
71	7 Autobus	11, 5, 5
73	9 Pr Min Netanyahu	4, 4
76 = 15	8 Yitzhaq Rabin	11, 11, 6, 6
80 = 58	9 The Next War	10, 3!
81 = 73	9 Pr Min Netanyahu	10, 7, 3, 3
86	6 In 1995-96	11
87	8 World War	11
88 = 56	10 Atomic Holocaust	11
89	7, 5 End of Days
92 = 89	7, 5, 4 End of Days	5
93 … 89	7 End of Days	11, 4, 4
96	5 Made by Computer	21
99(1)	6 Bible Code	11
99(2)	4 Computer	24
104	5 (year) 1997	13, 7, 7
106(1) = 87	8 World War	9
106(2)	7 Roosevelt	15, 4
107 = 56	10, 3, 3 Atomic Holo.	3
108	11 Pres Kennedy to Die	5
109	7 Oswald	14, 3
110 = 109	7, 4 Oswald	10
111	6, 6 R. Kennedy; Sirhan	10
113	12 Captivity of Toledano	9, 4, 3
114	7 Goldstein	19, 5
117	8, 7 Oklahoma	7
118	9 Murrah Building	6, 6, 3
119	9, 6, 4, 4 Name Timothy	8, 5
124(1) = 87	8, 4 World War
124(2) = 87	8, 6 World War
125 = 56	10, 6 Atomic Holocaust
128 = 87	8, 2 World War	6, 4
129	8, 5, 4, 4 Communism	3
132	7 Atomic Weapon	6, 5
133 = 87	8 World War	5
134	13 Armegeddon Asad Holo.	9
135	5 Syria	13, 11, 4, 3
138	6 In (year) 2113	18, 5
139(1)	7, 4 Great Earthquake
139(2) = (1)	7, 5 Great Earthquake
141 … 139	7, 4, 3, 3 Gt Earthquake
142	6 L.A. Calif.	5, 4?
143 = 139	7, 4, 4 Gt Earthquake	3
145	7, 4 Kobe, Japan	6, 5
146	7 Year of the Plague	9
147(1) = 139	7, 5, 3 Gt Earthquake
147(2) = 34	8 Economic Collapse	10
149	8, 4, 4 Dinosaur	5
151	6, 4 Swift	11
154	7, 5 Comet	10
155(1)=154	7 Comet	6, 4
155(2)=154	7 Comet	8, 4
157 = 81	9 Pr Min Netanyahu	9
158 = 81	9 ditto	9, 3
160 = 81	9, 4 ditto	12, 9, 3, 3, 3
161 = 58	9 The Next War	12
163 = 87	8 World War	12, 5
164	8, 4, 4 25 July 1996	18
166 = 55	9, 6, 4 Holocaust of Isr	3
167 = 146	7, 4 Year of Plague	9
168 = 87	8 World War	11, 6
169 = 55	9, 5 Holocaust of Israel
170 = 56	10 Atomic Holocaust	7, 5
171(1) = 55	9 Holocaust of Israel	3, 3
171(2) = 93	7 End of Days	11, 5
176(1)		16, 4 Future Bkwds
176(2) = (1)		16, 6 ditto
180 = 56	10 Atomic Holocaust	11, 7

Lookingat the middle column of this table, we see that a number of 8-letter code wordshave been found, somewhat fewer 9-letter, and one each of 10, 11, 12, and13-letter words (most of these are actually phrases):

8-letter:

AtomicArtilleryman

Communism

EconomicCollapse

LibyanArtillery

Oklahoma

Shoemaker-Levy(2x)

25July 1996

WorldWar

YitzhaqRabin

9-letter:

Einstein

HisName is Timothy

Holocaustof Israel

MurrahBuilding

TheNext War

PrimeMinister Netanyahu

10-letter:

AtomicHolocaust

11-letter:

PresidentKennedy to Die

12-letter:

Captivityof Toledano

13-letter:

ArmageddonAsad Holocaust

Wewould expect, given the length of the text of the Old Testament, that one couldfind virtually any 8-letter word desired, or at least some synonym. It should also contain lots of 9-letterwords (the text length of the OT C 952,000 letters C is over 90% of that needed [table 3] for 9-lettersearch words to be very common). Ten-letter and longer words or phrases wouldbe much rarer, but there are also many more such word combinations. With enough searching, investigatorsare bound to find some. We don'tknow how much searching they did.

Atthis point it is possible to see a real problem with the idea that this"Bible Code" is the work of God or even a very clever author. Such a divine or human author who is constructing the text of the Bible from scratch ought to be ableto insert within it a meaningful coded text of almost any length. (A good experiment would be to tryconstructing such a text yourself!)[18]Yet the longest coded words or phrases (not regular text or redivided regulartext words) that occur in the Bible Code are "Captivity of Toledano" (12 letters)[19]and "Armageddon Asad Holocaust" (13 letters).[20] Getting long words in the regulartext is much easier, as we saw with our Gettysburg example. Yet the longest examples used byDrosnin in the Bible Code are"to shut up the words and seal the book until the end" (24 lettersstraight out of Daniel 12)[21]and "the writing of God engraved on the tablets" (21 letters straightfrom Exodus 32).[22]