The following article and discussion have been occasioned by a paper published in the journal BRISMES (British Society for Middle Eastern Studies) by Rashad Khalifa called "Computer Analysis of the Quran Text" (1988) which lists 57 mathematical patters of 19 in the Qur'an. This proves, the author claims, that "(1) the Qur'an was not written by humans," and "(2) that the Qur'an has been perfectly preserved" (ibid, p. 7). In response, AbdulraHman [his spelling; common spelling is Abdul-Rahman] Lomax prepared the following article disputing the methodology of the procedure and its results. Following the article are two email discussions elaborating on the discussion. None of these documents mentions the Bahá'í Faith.

For further reading, see the detailed article The Mysterious 19 in the Quran: A Critical Evaluation; for a sympathetic exposition of the "19 miracle," see Mathematical Miracle of the Quran.

---

All Rights Reserved. Permission to reproduce this document for free distribution is granted, provided that this notice is preserved and the document is reproduced in its entirety. Quotation in whole or in part for the purpose of discussion in the Internet newsgroup soc.religion.islam is explicitly allowed. Contact the publisher at the above address or by e-mail (marjan@crl.com) for further information or permission.

---

Contents

1. Introduction
2. What is the Qur'an?
3. What, precisely, is counted?
4. Is the data filtered before presentation?
5. Is the data verifiable?
6. Summary

1. Introduction.

This document is planted as a seed; it is hoped that it will grow into an authoritative discussion of the problems involved in the study of what has been asserted to be a "numerical miracle" in the Qur'an. The editor of this FAQ, and the author of the initial draft, is Abdul-Rahman Lomax (marjan@crl.com). Corrections and comments are invited; permission to incorporate them into the next draft will be assumed unless it is denied.

Because the initial person to report the kind of phenomena we will be discussing was Rashad Khalifa, late of Tucson, Arizona, USA, and he also made numerous claims considered heretical by many, the subject is tainted with suspicion. However, Khalifa claimed that the "miracle" was objectively verifiable. As such, it should stand regardless of the truth or falsity of the remainder of his claims. Allah, the Most High, has said in his book (49:6) "... If a wicked person comes to you with news [nabaa'], ascertain the truth, lest you harm people in ignorance...."

Further, it remains to be demonstrated that the "miracle" of the nineteen, or the other numerical phenomena claimed by Khalifa and others, if valid, would prove that Khalifa was correct in his other claims, such as the wholesale rejection of hadith. A full discussion of all this is outside the scope of this document, which will not discuss the character and the other claims of Rashad Khalifa; rather we will attempt to define the problems involved in looking at the Qur'an itself to determine if the claims of a "numerical miracle" are justified. Here is what Khalifa said in the preface to his book, a book which was intended to be definitive and irrefutable, *Qur'an, Visual Presentation of the Miracle*:

"There now exists physical evidence for a message from God to the world. This marks the advent of a new era in religion; an era where FAITH is no longer needed. There is no need to "believe," when one "knows." People of the past generations were required to believe in God, and uphold His commandments ON FAITH. With the advent of the physical evidence reported in this book, we no longer believe that God exists; we KNOW that God exists. Such knowledge is ascertained through God's final scripture, Quran, wherein overwhelming physical evidence has been encoded. "Employing the ultimate in scientific proof, namely, mathematics, the evidence comes in the form of an extremely intricate code. Thus, every word, indeed every letter in Quran is placed in accordance with a mathematical design that is clearly beyond human ability. [...]

"Not only does the evidence prove the authenticity and perfect preservation of the Qur'an, but it also confirms the miracles of previous messengers [....]

"[...] Upon reviewing the evidence here, and examining the appropriate narrations, the reader will be as positively certain as an eyewitness."

It must be noted that Khalifa, without explicitly acknowledging it, later denied his quoted affirmation of the "authenticity and perfect preservation" of the Qur'an he reproduced in his book. Certain of his counts in *Visual Presentation* were erroneous, and he eventually claimed that 9:128-129 were not authentically Qur'anic, referring to these verses as the "false verses."

This document will not attempt to conclude whether or not Khalifa was correct in his conclusions; rather it will examine how we might investigate the problem. In this investigation, certain errors or anomalies in the work of Khalifa will be mentioned. We have seen objections that this is "focusing on the errors of a dead man" instead, presumably, of looking to see if there is an actual miracle. Lomax's response is that these anomalies are mentioned as examples of the pitfalls involved in these studies. He agrees that the errors of Khalifa do not prove that there is no miracle.

This brings us to the first difficulty:

2. What is the Qur'an?

Is the Qur'an a particular written text? Since a copy necessarily differs in some points (large or small, or even very small) from an original, is the Qur'an something that can be copied? Or is it an ideal, a form (sura) which the copies follow with greater or lesser degrees of perfection?

Traditionally, the consensus of the reciters has had greater authority than any particular written text.

Khalifa generally used the received text of the Qur'an known as Hafs, written in the Egyptian style. There are other received texts, which differ in certain respects from the text he used.

For example, in Warsh (another common version), the first verse of the Qur'an, called the invocation, is not given a number; rather the first numbered verse begins with "al-Hamdu lillah" (Praise belongs to Allah). There are other differences as well, essentially matters of spelling. Normally this is not a problem, but if one is counting letters, words, or numbered verses, as Khalifa did in claiming miracles, these differences become important.

But suppose that the Hafs version showed the phenomena Khalifa asserted and the others did not? Would this not simply show that Hafs was correct and the others wrong?

Theoretically this might be true; however, Khalifa did not always follow the received Hafs version. In particular, he modified spelling in at least two places: at 7:69 he changed a Sad to a Sin, and at 68:1 he spelled out the initial letter, thus adding two extra letters. It could be said that he was following pronunciation, but in numerous instances, he emphasized that it was the written Qur'an that was being studied and counted, not the pronunciation. He justified the change at 7:69 by referring to the Tashkent Qur'an, which is perhaps the oldest extant copy, and the change at 68:1 by claiming that it was spelled out in the "original text." He never specified what, exactly, this "original text" was, or where it could be found. (The copy of the Tashkent Qur'an available to Lomax does not extend to Sura 68, and Khalifa did not mention it in this connection.)

So, perhaps Khalifa is referring to the Tashkent Qur'an, as modified by removing the "false verses." But the Tashkent Qur'an does not match the Hafs which he normally counts, in many, many respects. For example, 3:37 in all the current received versions (as far as Lomax knows) has the phrase "inna 'llAh," and it is missing from the Tashkent Qur'an (which, in context has very little effect on meaning: it is only a phrase of emphasis). This, of course, would affect Khalifa's count of the word "Allah," which is crucial to his theory.

However, we could start with what is, to be sure, a widely accepted text of the Qur'an, the Egyptian Hafs which Khalifa generally used. If we could find substantial evidence of a numerical code in that text, then it *might* be possible to search for anomalies in the text, to see if some modification of the text, preferably but not necessarily with some authority from another text or hadith, makes the pattern initially found more complete. But if we can pick and choose from the various sources, there appears the second difficulty.

3. What, precisely, is counted?

How can we tell the difference between a genuine miracle and a numerical pattern which is created by manipulating the data or the method of analysis? Such manipulation can occur, for example, by choosing among different texts or definitions of what is being counted, or by choosing particular ways of analyzing the data over other ways which do not show the desired pattern.

I will give an example of each of these, from Khalifa's work, *Qur'an, the Final Testament*, the current edition of his translation, p. 625-626:

"We find that "The Quran" is mentioned 58 times in the Quran. However, verse 10:15 refers to 'a Quran other than this,' and therefore cannot be counted. Thus, 'this Quran' is mentioned in the Quran 57 times, 19x3. The suras where the word 'Quran,' in all its grammatical forms, is mentioned are 38, 19x2. The sum of numbers assigned to the suras and verses where the word 'Quran,' in all its grammatical forms occurs, is 4408, 19x232."

Looking in the Kassis concordance, I find 70 occurrences of Qur'an. Of course, this includes "all grammatical forms." Checking the number of different suras in which the word is mentioned, it is, indeed 38.

The mention at 10:15 is discarded because of reasoning regarding its meaning. It appears that we are not counting words, but meanings, and this opens a whole can of worms. If it is meaning which is being counted, then we are faced with all the places where another word is used to mean the Qur'an, including where it is clear that it is specifically "this Qur'an." In fact, at 10:15, contrary to Khalifa's assertion, the real Qur'an is mentioned, using the relative pronoun "hadha," "this." Further, at this verse, the word "Qur'an" is in the genetive indefinite form (qur'anin), so it is difficult to understand why it was included, in the first place, in the count of "The Qur'an," which would be "al-Qur'an."

I find 52 occurrences of "al-Qur'an." What was Khalifa counting? He wrote:

"Two other grammatical forms of the word 'Quran' occur in 12 verses. These include the word 'Quranun' and the word 'Quranahu.' One of these occurrences, in 13:31 [,] refers to 'another Quran' that [would] cause the mountains to crumble. Another occurrence, in 41:44, refers to 'a non-Arabic Quran.' These two occurrences, therefore, are excluded. Table 23 shows a list of the suras and verses where the word 'Quran,' in all its grammatical forms, occurs."

Table 23 agrees with the Kassis concordance if 10:15, 13:31, and 41:44 are added back in. Note, once again, that some words were excluded because their meaning does not meet some standard. Here is the list of all the forms other than "al-Qur'an," organized by the form of the word (note that there are four forms, not two):

Qur'anin (genetive): 10:15*, 10:61, 15:1.
Qur'anan (accusative): 12:2, 13:31*, 17:106, 20:113, 38:28, 41:3, 41:44*, 42:7, 43:3, 72:1.
Qur'anun (nominative): 36:69, 56:77, 85:21.
Qur'anahu (verb + pronoun): 75:17, 75:18.

The forms excluded by Khalifa on the basis of meaning other than *the* Qur'an are marked with asterisks. However, since some of the other occurrences could also be referring to other than our Qur'an (a few of them are ambiguous), we are no longer looking at purely objective facts. In particular, the verbs are really a different word ("Recite" instead of "Recitation"), but they are counted in the second and third of Khalifa's statistics.

Khalifa did not specify, in his latest edition, what other two forms were included in his count of 57 for "the Qur'an." We know that it was not "Quranahu" and "Qur'anun," because he mentioned those as part of the "other forms." So that leaves "Quranin" and "Quranan." Lomax finds no way to reconcile the counts with the data.

For those who do not know Arabic, the three indefinite forms, distinguished only by the termination at the end, "un," "an," or "in," are only different because they are being used as the subject of the sentence, the object of the verb, or genetively (the object of a prepostion or an indicator of possession). The meaning of the word itself is not changed; only its place in the sentence changes between these three terminations.

The point is not only that there are hidden manipulations of the data going on, unstated premises, and the like, but also that these choices are arbitrary. If one looks at Khalifa's counting of other words, different standards are applied in each case. On close examinatin, "word" is not a precisely defined term. The only unifying thread is that methods of counting are chosen which lead to a multiple of 19.

This leads us to the third difficulty.

4. Is the data filtered before presentation?

How can we distinguish between selective presentation of data and a truly signifigant pattern? This is similar to the difficulty discussed in 3., but, the description is from a different, statistically-based, point of view.

Suppose we have ten, or a hundred, or a thousand statistics from the Qur'an which are multiples of 19. Does this prove that there is a "numerical miracle?" From a book even less complex than the Qur'an, it would be possible to generate more counts than there are atoms in the universe, and, presuming that the data was random, on the average, one out of 19 of these counts would be divisible by 19. One could start generating counts of different things in the book, and collect the ones which are divisible by 19. The size of such a list is limited only by the persistence of the one searching for divisible counts.

Only in one publication, to date, from Khalifa's followers, is there even a small start toward answering this question, and the analysis presented there was seriously flawed. It will be given here as an example of how easy it is to be misled.

From *Beyond Probability, God's Message in Mathematics*, by Abdullah Arik, Series I: The Opening Statement of the Quran (The Basmalah) [sic]*:

Arik presents a series of "Facts" in which he takes the letters and words of the Bismallah (bsm allh alrhmn alrhym) and generates numbers with them. 8 of these facts use the following form:

A number is generated by writing the numbers 1, 2, 3, and 4, interspersed with numbers derived from the corresponding words in the Bismillah. The first fact in this form asserted by Arik, called "Fact 2," is "the sequence number of each word in the Basmalah followed by the number of letters in it." This is:

1 3 2 4 3 6 4 6 = 19 x 19 x 36686
^   ^   ^   ^

He then asks the question, "what is the probability (chances) for the Basmalah's mathematical composition to occur by coincidence? Can we compute this probability? If we can, how? Based on our assumption of coincidental occurrence, we can treat each number in Facts 2-9 as a random number."

He proceeds to generate all possible eight-digit numbers which satisfy the criteria that the first, third, fifth, and seventh numbers are 1, 2, 3, and 4, respectively, and the number is divisible by 19, finding 527 such numbers, and he compares this with the 100 million possible eight-digit numbers, concluding that, "We can say that the probability of the occurrence of the mathematical phenomenon [,] described in Fact 2, is 189,753 to 1. [sic]"

He meant "1 in 189,753." However, actually, the probability is 1 in 19. The discrepancy between these two figures is explained by the fact that he took his exprimental condition (1a2b3c4d) and treated these fixed numbers as if they were random variables. Thus only one out of 10,000 eight digit numbers is included in the experiment, from the outset; since one in 19 of these numbers could be expected to be divisible by 19 (this is the normal case), the theoretical prediction of the result of his (erroneous) calculation would be in in 190,000. The remaining discrepancy is basically round-off error caused by his method.

Moreover, the entire analysis is incorrect. Since the statistics given are selected out of a much larger body of statistics, it is certainly true that it is not a coincidence that they are divisible by 19. They were *selected* that way. Unless it were shown that the sample was unbiased, that statistics were not examined and discarded (because they were not divisible by 19), the numbers given in Arik's work prove nothing more than the persistence of Arik and his sources. None of Khalifa's followers have done the kind of global analysis necessary to convert a list of interesting numbers into a statistical proof.

There is a fourth difficulty which applies in some cases:

5. Is the data verifiable?

As an example, Khalifa reported counts of the letters known as "initial letters," which prefix some of the chapters of the Qur'an. Uniformly, he presents these counts as being divisible by 19. It is easy to miss, however, that he does not always count or combine letter counts in the same way to produce a total. This is another example of arbitrary counting criteria.

However, with the Suras which contain alif as an initial letter, he always uses the same overall pattern: he adds up the count of all the letters which initial the chapter, within the chapter. He reports all of these counts as divisible by 19. Since there are 13 chapters with alif as an initial letter, this, if true, is strong evidence for the existence of a pattern in the text. From random data, to find a method of analysis which would produce this kind of pattern would require examining an estimated 4 x 10^16 statistics. (This is 4 followed by 16 zeros.) The difference between this statistic and those reported by Arik is that a single counting method, which can be stated in a few words, is applied precisely to all examples (at least, all examples containing alif), whereas the counting or calculation method changes with each statistic in Arik's work.

However, counts of alif are impossible to verify. In *Visual Presentation,* Khalifa presented verse-by-verse counts of alif. His counts do not match his own published text of the Qur'an. It is apparent, from this, that he is counting hamza as alif. But he does not always count hamza: for example, both 3:158 and 30:51 contain the same word, la'in, which contains a hamza (represented by the apostrophe); but Khalifa counts 4 alifs in 3:158, apparently including this hamza, but only 6 alifs in 51:30, excluding the same hamza. This is particularly odd in light of the fact that Khalifa strongly blasted his critics for denying that the Bismillah contains only 19 letters because they pointed to similar unwritten letters.

(Hamza was not written in the earliest Qur'ans.)

Particularly because of the history of Khalifa's counts of alif (they changed radically over the years: the alif count in Baqara increased from 4502 to 4592), Lomax concludes that errors in counting other initial letters forced Khalifa to reanalyze his alif counting in order to keep the total counts at multiples of 19. He re-analyzed until he believed he had found a method of counting that produced the "miraculous" numbers, but he did not go so far as to apply the new criteria to all cases; he stopped as soon as he had the results he wanted. Neither did he state explicitly his criteria for counting.

Further, since there are known errors in his counts where there are no hamzas to manipulate, the statistics cannot be accepted, even if the alif counts were correct: He counts 16 alifs in 13:41, where there are only 15 and no extra hamzas, and he misses a lam at 30:21 (He counts 7; there are 8).

By no means has this discussion mentioned every error which Khalifa is known to have made; only one or a few examples of each type of error has been given.

6. Summary.

It is the opinion of Lomax that it will never be possible to prove that there is no "numerical miracle" in the Qur'an; however, it can be said that, until the questions raised in this document are addressed and answered, it remains to be demonstrated that this kind of "miracle" exists.

---

Follow-up emails culled from listservers

Date: Mon, 8 Apr 1996
From: Abdulrahman Lomax
Subject: Re: Current Status of 19-pattern

xxxx wrote: "Of course the historical textual issues you have raised are very legitimate and highly fascinating, but I am wondering if you have made a list of which discoveries of Khalifa's are suspect in your view and which ones still stand up to objective verification..."

I did write a long article examining one list of the "miracles," (which, by the way, did not include the kind of work Ali Fazely has done) and I found that, certainly, many of the "facts" are what I would call "interesting." But there is a fundamental problem which has not been addressed by any of the researchers promoting the "miracle":

In any collection of random numbers, one would expect to find some of the numbers to be a multiple of nineteen: this is without any ambiguity whatsoever as to how the numbers are determined, and it is not, by itself, an indication of any sort of pattern in the data. Only if a collection of numbers is chosen by a process which would be expected to be random, that is, which is not biased toward nineteen-divisibility, and substantially more than one out of nineteen numbers turn out to be divisible by nineteen, would one suspect the existence of a pattern.

Khalifa's data is not randomly chosen: it is fairly obvious that only nineteen-divisible statistics are presented, out of a much larger universe of possible statistics. To a certain degree, to non-Arabic speakers, this is not obvious, for the Arabic language, in the way it is written, is subject to a certain ambiguity in terms of exactly what letters are used or how one would define or divide words.

By choosing word definitions, for example, one may alter the count of a word until it comes out to be nineteen divisible.

So far, most Khalifa and Khalifite statistics are anecdotal: they are not part of a pattern that repeats across a universe of many examples. Sometimes the data is presented in a way that makes it appear that there are multiple examples (in fact, the whole 19 thing would make it appear this way), but what I mean by "a universe of multiple examples" is a series of statistics produced by examining all occurrences of a particular kind of pattern.

An example of this would be the counts of initial letters. The count of each initial letter within the sura that it initials would be an item of data of which there are many examples. Does this count come out a multiple of 19 more often than would be expected by chance?

The answer is no. What Khalifa has done is to combine these counts in various arbitrary ways; a consistent method of determining the total has not been used.

But, it may be objected, there are a series of suras which are initialled with alif and other letters. For each one of these suras, according to Khalifa's counts, the total occurrence of the initial letters within each sura is a multiple of nineteen. If true, this is a stunning statistic, not reasonably attributable to chance. However, it turns out that counting alif is not a simple matter, and we do not know how Khalifa did it. His published counts in Visual Presentation can be used to infer his technique, but the results are not consistent. Essentially, we have only his word that these counts are accurate. Further, if one looks at his published data over the years, as he discovered errors in the counting of lam, mim, he always changed his count of alif to compensate for these errors. This published series of inaccurate counts does not inspire confidence in the accuracy of his work.

Now, there has been a lot of work done after Khalifa's assassination, and not all of it involves the number nineteen. Milan Sulc, for example, has found many interesting phenomena with primes, prime and composite ranks, and so forth. However, it remains to be shown that these fascinating numbers are uniquely Qur'anic. Here, the problem is that it is possible to derive any number one wishes through some transformation from the Qur'anic text. Further, what is highly suspicious about Sulc's work is that it generally is based in verse counts; and what is not commonly appreciated is that the versification of the Qur'an is not a matter of consensus; there are many variations generally considered acceptable. But all the work has been done with the Egyptian system used by Khalifa; no serious attempt has been done to determine if the "miracles" apply as well to other systems of versification.

I wrote a document on the issues involved in showing a numerical miracle in the Qur'an. So far, no researcher in this field has even attempted to satisfy or, alternatively, show the inapplicability of the criteria described in that paper. I don't have it in front of me, but, as I recall, these are the criteria:

(1) The text being used must be specified. Believe it or not, there are variations in Qur'anic texts. If one may pick and choose, as Khalifa did, one may amplify the natural occurrence of nineteen-divisibility. An example of this is the count of Sad. Khalifa originally claimed perfection in this count, and pointed to 7:68 as a miracle, for if the word there had been spelled in its conventional spelling, there would have been a missing Sad. Then it was discovered elsewhere that he had missed a Sad. In Visual Presentation, he quietly changed the Sad to a Sin, the conventional spelling. He found manuscript evidence for this in the Tashkent Qur'an, which is certainly very old. However, the Tashkent Qur'an has other variations, including, at one place, a missing Allah. If there are not others, this would make the elimination of 9:128-129 unnecessary to make the count of "Allah" to come out as he claims.

(2) The rules and criteria for counting necessary to reproduce the count, and to make other similar counts, are not explicitly stated. For example, Edip Yuksel once wrote, defending a certain word count, that I should use the word definitions implicit in Abdul-Baqi's concordance. Fine. I obtained a copy, and used it to study all the words in 9:128-129. I found that using the form of the words in Abdul-Baqi, and checking the occurrences of those words for nineteen divisibility, out of eighteen words, one word's count was divisible by nineteen. This was true whether or not 9:128-129 were included or not; of course the word was different in each case (raHiymun with the verses and tawallaw without them.) This kind of comparative and consistent analysis is completely lacking from the Khalifite research, which consists entirely of anecdotes on one side of the question.

(3) Is the data filtered before presentation? The Khalifites provide reams of counts of things that turn out to be multiples of nineteen. But they do not publish all the numbers they found in their search for these phenomena. Rather, only the results which fit the theory are presented. This would be like trying to determine if all Irish are red-haired by counting how many Irish actually have red hair; to determine if a person is Irish, one would consider their appearance, naturally; so anyone without red hair is not counted. One then presents a list of fourteen thousand Irish persons with red hair as proof of the theory that all Irish persons are red haired. Anecdotes, no matter how many there are, do not prove any theory.

Do Qur'anic statistics show a bias toward 19-divisibility? No conclusive research has been done. Is there some other kind of numerical pattern in the Qur'an, sufficiently clear that one could determine whether or not 9:128-129 are part of the Qur'an or not? The research simply has not been done. However, I will say this: it is impossible to prove that there is NOT a pattern, unless the pattern is specified. Specific statement of the pattern in a way that can be verified is something which, so far, has not been done, nor have I ever seen any attempts to do this.

AbdulraHman Lomax

---

Date: Thu, 11 Dec 1997
From: Abdulrahman Lomax
Subject: Re: KORAN

as-salamu 'alaykum.

I have written extensively on the subject of Dr. Khalifa and his alleged discoveries. I knew him personally. There is a great deal of information on the web, most of it misleading for those who do not study the matter carefully. The matter has been debated at length in soc.religion.islam, and archives are searchable on dejanews (I get there by searching usenet on www.yahoo.com).

There are many aspects to the so-called "numerical miracle." Only some of them are directly related to the number 19. There are also claimed discoveries using prime and composite indices. It gets pretty complex.

No rigorous studies have been done, to my knowledge. It is notoriously difficult to determine after-the-fact probabilities of a "coincidence," for there is no clear standard for us to judge what events we might consider "miraculous" in advance.

So if the coincidence of two numbers seems amazing, how many *other* possible coincidences would also seem amazing? Without an estimate of this, it is actually impossible to determine a probability that a "miracle" will occur.

Nearly all the claimed discoveries are actually well within the bounds of what can and does occur by chance, and statements to the contrary have, so far, not checked out as true.

That is a summary of the matter, but I will give some details.

In the 1970s, Rashad Khalifa began studying the letter frequencies in the so-called initial letters. He was *looking* for unusual patterns, and he found them, as could be expected. His first "discoveries" did not involve the number 19 at all, and they were so weakly "miraculous" that these early discoveries are not mentioned at all in later publications.

Later, however, he noticed certain facts, and he also found that if he made certain special definitions and selections, other facts began to appear, and the bulk of these involve divisibility by the number 19. Since 19 is mentioned specifically in the Qur'an (74:30) in a context that can be read as a reference to some kind of miracle, this was very attractive, and he pursued it vigorously.

The facts are of various kinds. There are counts of letters in the suras involving initial letters and in a few other places, there are word counts, either in a particular sura or section or in the whole Qur'an, there are counts of the suras themselves.

I'll start with some of the most obvious facts. "Bismillah al-rahman al-rahiym," as it is traditionally written, is written with 19 letters. And there are 114 (6x19) suras in the Qur'an. There are *also* 114 occurrences of the invocation in the Qur'an (one is missing from Sura 9, and there is an extra one at 27:30).

Khalifa and his followers will also be quick to point out that if one starts counting suras at sura 9, the 19th sura would be 27, where the extra bismillah is found. Of course, there are many other arrangements which would also seem significant. Suppose it were in Sura 19? or 28 (28-9=19).

The first revelation was the beginning of sura 96. Khalifa pointed out that this is number 19 from the end of the Qur'an.

So far these statistics are not particularly controversial. Sometimes it is pointed out that the writing of the bismillah is idiosyncratic and that "really" it has more than nineteen letters in it, but had I been asked, long before I knew about Khalifa's work, how many letters there were in the Qur'anic bismillah, I would have had no hesitation to say, "19." But it is still important to note that there are other possible ways of looking at it, for in some of the other statistics, Khalifa and his followers do look at things in idiosyncratic and unusual ways in order to make the numbers come out "perfect."

Then we come to an interesting allegation: that every "word" in the invocation occurs in the Qur'an and exact multiple of 19 times. Here is where things get really slippery. First of all, the term "word" is not precisely defined in Arabic. Even in English, where we might well define a word as a unique collection of letters, separated by spaces, we would still have problems. If we are counting "word," would we count "words," "wordy," "wordless," and the like? By my definition, we would not, for "unique" and "separated by spaces" kill those possibilities. But in counting words in the Qur'an, Khalifa counted some of the various forms and excluded others.

Starting at the beginning, what is the first "word" in the invocation? Is it BSM? How many times does BSM occur in the Qur'an?

It is known that Khalifa used the 'Abdulbaqi concordance. (Sometimes people think he used the computer to count words, but it appears that this was more hype than reality. He is not known to have possessed an accurate computer-readable text of the Qur'an.)

Now, Abdulbaqi lists three occurrences. But, actually, there are 115. Abdulbaqi does not list all the initial bismillahs beyond the first, so there are 112 more. Here we encounter, for the first time, a real question as to "what is the Qur'an" when we attempt to count how many times a particular word occurs "in the Qur'an."

Neither one of these numbers is divisible by 19. What gives? Well, Khalifa did not count BSM. Rather, he counted some forms of ASM (he wrote it as ISM, which is more like how it is pronounced when preceded by bi-; but it is written alif-sin-miym). If I look in 'Abdul-Baqi, under the entry "ASM", sure enough, it says "19" for the number of mentions. But this excludes all the forms with a prefix or suffix, *including the very form that is in the invocation*! It also excludes the plural forms.

In English, if I were to count the number of times "name" occurred in a book, I would certainly count "his name." In Arabic, "his name" is written ASMH, ismuh; so we can see how difficulties arise. By making special rules for counting, one can manipulate the counts and select one which appears to show some remarkable trait. Where no trait is found, one simply goes on to look elsewhere. One out of every 19 numbers that one examines, chosen at random, will be divisible by 19. All it takes is patience, lots of patience and lots of time, to find remarkable statistics.

Is there *really* a miracle here? It is very difficult to *prove* that there is no miracle. What is the standard of proof? *It is undefined.*

Briefly, I will look at the other three "words" in the invocation. they are reported as ALLH, RHMN, and RHYM. Usually the Khalifites will add the vowels; I keep them out because it makes certain things clearer. (A in Arabic, the A that I am writing, is not a vowel, rather it carries vowels or modifies them and serves other purposes. In the original writing of the Qur'an, there were no vowels written, per se. )

If you look in Abdul-Baqi under ALLHu (nominative), you will find 980 mentions listed. ALLHa (accusative) has 592 and ALLHi 1126. (Abdulbaqi says 1125 for the latter; it is a tabulation error, he actually lists 1126).

Abdulbaqi is sometimes arbitrary in how he lists words. In this case, he has listed LLH, though it is written without the alif (A) together with ALLHi, since this combination is always genetive (it is li- plus allah).

Now, 592 + 980 + 1126 = 2698, which is 19 x 142. Bingo.

However, not only have we neglected the initial bismillahs other than the first (and more on that in a moment), which would add 112 mentions, but we have also omitted the other form of the name of God which is found in the Qur'an: ALLHM, allahumma, which means "O Allah." This occurs five times in the book.

And, as it turns out, Abdulbaqi also omitted 1:1 when counting "Allah," so the number in the so-called "numbered verses" turns out to be 2699. (Some Qur'ans number the first bismillah, some do not.) It took Khalifa some years to discover this, however. His published count of Allah in "Visual Presentation of the Miracle" had perhaps a dozen errors in it, even though it looked like a computer printout. When the smoke cleared, it appeared that the true count of Allah in the numbered verses, excluding "Allahumma" but including all other forms, was 2699.

Khalifa's response, ultimately, was to find something that he could remove from the Qur'an, and he picked the most likely candidate, the two verses at the end of Sura 9. This reduces the count to 2698, and also eliminates some fancy footwork which was earlier necessary to avoid counting RHYM in those verses.

And the other two "words" are "RHYM" and "RHMN." They are always written Rahman and Rahim by the Khalifites, probably because it obscures the fact that the words which are *really* in the invocation are ALRHYM and ALRHMN.

ALRHMN occurs in the numbered verses of the Qur'an 57 times. There is no other form of RHMN in the Qur'an. This statistic is genuine and requires no manipulation, except the selection of "numbered verses."

ALRHYM occurs 95 times. RHYMA occurs 20 times. Abdulbaqi, again, is somewhat arbitrary in his classification, since forms without the definite article AL- are listed with ALRHYM. Khalifa lumped these together, and excluded the mention at 9:128 on the grounds that it referred to the Prophet, not Allah. Later he excluded it on the basis of a certain well-known hadith reported from Zayd ibn Thabit indicating that these two verses (9:128-129) were found *only* with Ibn Khuzayma.

With the exclusion, the number is 114. Note that plural forms are also excluded.

Khalifa also claimed to have found a pattern of 19 in the letter counts of the initial letters. The bulk of his work is utterly unverifiable; he used no known consistent method of counting the letter alif. In spite of his statements, regarding the bismillah, that one simply counted what was written, he sometimes counted alif where there was a hamza and sometimes not, and he did not do this consistently.

It's a mess. But most of his followers have never tried to actually verify his work, beyond counting, say, the letter Qaf in Sura Qaf (50). Don't bother: there are 57, and likewise there are 57 Qaf in Sura ash-Shura (42). To examine all these statistics in detail and to show how they have been selected from a larger universe of statistics which are *not* divisible by 19 is a complex task. The information looks very interesting when it is presented in isolation; when one starts to realize how many different ways there are to count these letters and words and so forth, one's level of amazement can decline a bit!

There is currently a web site which purports to show a statistical proof that the initial letter data is truly an amazing miracle. It is the first attempt by a Khalifite (albeit a heterodox one, being in opposition to most of Khalifa's followers and acknowledging that Khalifa made many mistakes) to actually apply statistical principles. I have yet to examine it in detail; from a superficial examination it appears that he has himself done some data selection; essentially, there are an unlimited number of ways to analyze the data; if one picks a way that produces an amplified occurrence of 19-divisibility, it is not surprising that one finds an amplified occurrence.... The writer of this page is Omar Farouq; I suspect that a search of soc.religion.islam on his name might come up with a post containing the URL.

    >I'm not sure if I can help you. I know these calculations which all
    >come out to the number 19, were done by a somewhat eccentric Iman in
    >Tucson where I used to live. I think his name was Khalid. He had kind
    >of a cult-following of Muslims who accepted only the Qur'an and not
    >hadith as authentic. As I recall he was eventually assasinated by
    >some right-wing Muslim group. There must be something on the web
    >about this. Do you want me to post this on Islam-L?

Khalifa is not the only one who has worked with this concept, but he is the best-known. Dr. Cesar Majul found some of the remarkable statistics; he later parted with Khalifa when Khalifa started making outrageous claims about himself and about other aspects of Islam. Khalifa did assert that "following any other source of religion other than the Qur'an" is explicity forbidden in the Qur'an; he also claimed to be the so-called Messenger of the Covenant (wa idh ahadna n-nabiyyiyn....). He was apparently assassinated by Furqan; there was one conviction for conspiracy in connection with it.

There are also a few sects, particularly among the African-American community, which claim special significance in the number 19 and point to the kinds of facts that Khalifa asserted. It is not clear to me that any of these claims predate Khalifa's work; if they do not, it is quite likely that they were taken from it. Khalifa was *very* popular and widely known for a while, since he was purportedly showing "scientific, mathematical evidence" that the Qur'an was perfectly preserved" and that it was impossible for it to have been of human origin.

Had his data involving the letter alif been accurate and verifiable, it might indeed have had some such significance. Unfortunately, it appears far more likely that his search for patterns was the source of the patterns that he found.