20 January 2022

Reproducibility and imperfection

Prove all things; hold fast that which is good.
— Paul the Apostle (I Thessalonians 5:21)

Enthusiastic believers like to praise the great order in the Bible and recall the high number of evidence of its inspired nature. Their enthusiasm is indeed remarkable, but it is very annoying if it turns out that some of their statements are in fact false! The Bible indeed contains many remarkable facts but let us be a bit skeptical.

For example, a very frequently cited work by Geisler and Nix A General Introduction to the Bible (Moody Press, Chicago, 1986) claims that after 1000 years it is not usual that the books of the Old Testament do not change because of copying errors (p. 263). “Of the 166 words in Isaiah 53, there are only 17 letters in question. Ten of these letters are simply a matter of spelling, which does not affect the sense. Four more letters are minor stylistic changes, such as conjunctions. The three remaining letters comprise the word LIGHT, which is added in verse 11 and which does not affect the meaning greatly. Furthermore, this word is supported by the Septuagint (LXX). Thus, in one chapter of 166 words, there is only one word (three letters) in question after a thousand years of transmission – and this word does not significantly change the meaning of the passage.” On the other hand, Benner recently checked these statements on his web page and reported his results: “While I have reviewed several passages of the book of Isaiah to compare the text of the Great Isaiah scroll found in the Dead Sea Caves with the Masoretic text, I decided to put the above quotation to the test. I began with verse 1 of chapter 53 and found that it did not take long to find 17 letters that varied from the Isaiah scroll and the Masoretic text. In just the first 3 verses of chapter 53, a total of 23 words in the Masoretic text and 24 words in the Great Isaiah scroll, I found 19 letters that were different between the two texts.”

Such contradictions ruin all enthusiasm and make believers seem quite ridiculous. But there must be dry facts that can always be presented to anyone as evidence. Digital versions of the Bible texts are already available for free, and anyone can reproduce certain checks on his or her own.

The bad news is that we don't have the original text of the Bible. Just for the New Testament there are more than 5,800 Greek manuscripts and they differ in several parts – in fact, most manuscripts are just fragments. There is some discussion if the differences can be considered major and doctrinal, or they are just minor, but the main questions of Christian theology can be clearly answered. In near 20,000 copies the New Testament has been preserved in translations to Latin and other ancient languages in more manuscripts than any other ancient work of literature. A great introduction to this topic can be found on Wikipedia.

But what to do if the manuscripts differ? Clearly, there is a need to find the most plausible text, based on the estimated dates of the manuscripts and their similarity. Such a project is Novum Testamentum Graece, which is a Bible edition that contains, first, the most probable text that was used by the first Christians, and second, comments on possible parts of the text that may be different in some early manuscripts. Some modern studies like the Wuppertal project work with a high number of manuscripts (eventually: with all of them) assuming that the original text is something similar but maybe not completely the same. For some details I refer to the paper by de Vries and Karrer Early Christian quotations and the textual history of the Septuagint: A summary of the Wuppertal research project and introduction to the volume (Textual History and the Reception of Scripture in Early Christianity, Society of Biblical Literature, 2013).

The Wuppertal database is a remarkable collection of quotations. According to de Vries and Karrer it lists 449 quotations of 357 different verses from the LXX in 389 New Testament verses (without the disputed quotations, allusions and innumerable echoes). These are amazing numbers, and after reading their paper I contacted Dr. Karrer to get access to the online database in 2018. Dr. Karrer kindly sent me the link to a nice user interface of a flexible version of the Greek Bible – this, however, is not working now, and it's a shame but because of that I cannot point to something reproducible.

During the last years I have been collecting the quotations on my own, by using mostly mechanical methods. Another leader of the Wuppertal project, Dr. Kreuzer warned me that unbiased research needs to assume that we don't have the original text of the Septuagint. Only plausible texts exist. So, when I start working with the Sword library, in particular with the texts LXX and SBLGNT, I am already biased to a plausible text, and such research is clearly inaccurate. On the other hand, if we are satisfied with fuzzy matches (that is, not just literal matches), a plausible text can be acceptable. That's why I finally decided to use these two plausible sources – the research on them is reproducible for anyone.

Of course, the question is if fuzzy matches are acceptable as scientific evidence. I think so. Let me begin my explanation from biology and then mechanics. It is clear that every living thing has flaws. Human beings inherit genetic errors from their ancestors and new errors could also be introduced in the moment of conception. Later, malfunctional organs introduce further difficulties in a human's life. But despite that, most living beings can live with these issues for a long time. Imperfection is not a reason to die immediately, but living organisms seem to be designed to go beyond such problems. To go towards mechanics, let me recall that the system of the human skeleton consists of 206 pieces, and a couple of bones form the arm (3 pieces) and in the hand (27 pieces). Can you draw a straight line with a pencil on a sheet of paper by exploiting your bone system? I mean: can you draw a perfect line? You may think that you can draw a completely straight line, but this is extremely difficult, at least from the mathematical point of view. Just one argument to highlight the problem: James Watt designed his famous 4-bar linkage that produces an approximately straight line – the concept is still used in modern cars in their suspensions. But to be honest, Watt's curve is just approximately straight (this cannot be fixed by changing the input parameters, that is, the lengths of the bars), still it is good enough for the everyday use, at least for a couple of years during the warranty of some parts of the car.

I assume that imperfection can also be assumed in several fields of science, including comparison of ancient manuscripts, without the loss of the most substantial results.

Let's go back to the Bible. In my second blog entry on this topic, I mentioned Matthew 1:23 which cites Isaiah 7:14, and in that case a simple fuzzy match of 8% difference is adequate, even if two literal matches can also be detected. Now my project identifies 234 passages in the New Testament that are, in my opinion, with no doubt, clear matches between LXX and SBLGNT. This number differs, of course, from the number reported by de Vries and Karrer, but now I am still in a stage where I can only check my database and cannot compare the results with the Wuppertal database. On the other hand, my condition for acceptance may be a bit stricter than that of de Vries and Karrer. I postpone this question for a later entry.

A difficult dilemma for the mechanization is to decide if a probable match is written on purpose or not. In my research I wanted to be as strict as possible: If the author in the New Testament does not indicate that the written text is a quotation, then I skipped the probable match. One example of such a questionable passage is Romans 9:20: But who are you, a human being, to talk back to God? “Shall what is formed say to the one who formed it, ‘Why did you make me like this?’” Even if this seems to be a deliberate quote written by Paul, there is no direct mention of the usage of the Old Testament. By using the tool bibref we can agree that the mechanical study confirms the match, however, we cannot explicitly claim that Paul remembered this passage from Isaiah 29:16 accidentally, or he just forgot to mention the source of the quotation.


Now, by clicking on the input bar and pressing ENTER, after some seconds you can see that Romans 9:20 seems a good candidate for citing Isaiah 29:16. The last lines of the outputs read as follows: A continuous literal match of 24 characters is remarkable (“μη ερει το πλασμα τω πλασαντι”, “shall what is formed say to the one who formed it”, see an Interlinear translation to check the meaning of the Greek words), and, at the end (“με εποιησας”, “did you make me”) there is a literal match of 12 characters.

Anyway, this is a kind of grey zone that cannot be decided unequivocally, in my opinion. On the other hand, there are still plenty of passages that are unequivocal, more than 200 of them, and they are more than enough to show the wonderful connection between the Septuagint and the Greek New Testament. Unfortunately, the example of Romans 9:20 warns us to use more work than just a mechanical search of the text. But mechanical searches can surely be a kind of pre-processing to find good candidates for obvious matches.

Also, we need to admit that a mechanical search may ignore some obvious matches. To learn more on this – and on how the mechanical algorithms work – please stay tuned!

Acknowledgment. My friend László Gyöngyösi kindly helped me improve the first version of this blog entry.


Continue reading…

See also a filtered list of the entries on topics GeoGebra, technical developments or internal references in the Bible.


Zoltán Kovács
Linzer Zentrum für Mathematik Didaktik
Johannes Kepler Universität
Altenberger Strasse 54
A-4040 Linz