And thanks for the phrase “gerrymandering the hypothesis”. I’ve seen it done, but never knew what to call it.

To supplement what Mr. Carrier said, even in this modern world, if you type any 7 words from any writer that is not about something completely generic (i.e. “Today GE rose in after hour trading”) into google using quotes, the only return will be from that writer. Not always of course, but it is so far more likely that it can said that no one can accidentally write the exact same seven words without copying (hat tip to.. better yet type “plagiarism seems desperately foolhardy” into Google, only 4 words, but even then there is only one return (don’t forget the quotes) and that is where I learned this fact – and my example).

Though after I post this I guess this means that there will soon be two returns 🙂 But I can’t think of a better way to demonstrate the strength of Mr. Carrier’s argument. Multiple fragments that can be sourced to a modern work either shows that the modern work used this exact document as its source, or…..

A side note (since it’s not the same thing exactly but pertains):

When I worked in the ETS (Electronic Text Service) of the Columbia University library, where one of my jobs was helping patrons (usually faculty and grad students) use and develop resources for scanning and data-analyzing texts, I decided to see how hard it was to fake an author’s style by scanning one author into a database and creating a vocabulary list (every word they ever used in one of their novels), then trying to write an essay using only vocabulary on that list (and thus nothing I would normally use myself that wasn’t on that list). Holy cow was that hard. It required a lot of effort. And felt very strange. I cannot imagine how much harder it would have been if instead of their entire novel’s vocabulary, I could only use one single chapter’s vocabulary, and instead of restricting myself only to the vocabulary list for that one chapter, I had to include several three or four word phrases as well from a list of all such that author used.

Obviously it would be easy to fabricate seven half sentences that way. But then, that’s why it was easy for the forgers. The odds of this happening by accident though, shoo, not bettable odds. If I completed the essay following this artificial rule, then I could randomly sample it and get results like this. But only because I had thus forced it to. By design. To instead write a chapter in my own style, randomly select seven sequential half-sentences from it, and expect every single word and several multi-word phrases in that selection to match by chance the vocab list for some other author’s chapter? Even allowing two words off, I doubt I’d succeed at that using another one of my own chapters! I’m no poker pro, but I play poker well enough to know I wouldn’t bet any significant money on that.

It’s true that if the sample were smaller, the coincidence becomes more likely (and if larger, the converse).

But this is too many words together.

Also, if we used writings as vast as Crossan and Lewis. But these are Gospels. Gospels are shorter than even a single chapter in Crossan or Lewis.

So, first, the comparison populations are small (Gospels are short chapters). It’s much less likely I can assemble a chapter by accident using only the words in someone else’s chapter. Unless I am doing that deliberately. Hence the conclusion.

Third, we aren’t randomly picking words. We are picking a cluster of words. So we have to explain their clustering. It’s like putting one microscope onto a pattern, and nearly everything inside the viewfinder matches a different pattern—and everything that doesn’t, just happens to match exactly what gives this papyrus market value.

Fourth, every word is identical (except for single letter changes necessitated by changing male to female, another instance of what makes this item financially valuable). It’s not just cognates or conjugations or declensions. They are direct lifts. And several are not single words but whole phrases of three or four words in the exact spelling and order.

We could actually test this. We could grab random “clusters” of the same number of words together in the canonical Gospel of John and see how often every word and several phrases in that cluster is in the Gospel of Mark, identically but perhaps for a change of gender (excluding instances, of course, where we conclude John is quoting Mark—which is why we conclude that!). I guarantee you, even if we find one, it will be rare. It will require a remarkable coincidence to get it. And that’s the argument the forgery advocates are making.

You can compare all of this yourself because Grondin’s Interlinear is online. You don’t need to read Coptic to check the symbol strings. Look at or for convenience print out Bernhard’s image I show above that shows all the underlined words in GJW and gives the line numbers in Grondin’s. You can sit that by your screen and go through item by item and see the words are identical in Grondin’s. (Maybe it’s less obvious for speakers of English why this is especially unlikely for an inflected language like Coptic.)

Look at the very first example: the first underline exact lift in line 1 in GJW from Grondin’s 49:36. Not just a word. But a three word phrase. Exact same order. Letter for letter identical. Search Grondin’s for “49:36” to see where it matches. Exact. The same happens case after case: another exactly matching three word phrase, then an exactly matching four word phrase, and so on; eventually just some isolated identical words, but again inflected identically except when the gender has to be changed (requiring the change of a single letter), exactly as required to give this papyrus modern market value. All in one single tiny conjoined selection.

And then the point about all the error-matches. But that’s just an additional point.

Although you’ve given me an idea of a sentence to add: the point about the only deviations all conveniently matching what makes this papyrus financially valuable on the modern market. I think I’ll also replace “astronomical” since you are right that can be vague enough to be hyperbole. The odds are quite low. That’s enough to say.

That you randomly grab just seven lines, and not even the whole lines but just fragments of them, and nearly every single word or phrase in that tiny snippet comes from the Gospel of Thomas? The odds are surely astronomical. If we were to suppose this random sample were typical (as her argument requires), then we’d have the bizarre situation where 95% of all the words and phrases in the entire GJW match GThom! In other words, that they almost never used any other words or phrases in their whole Gospel, than words or phrases in that other Gospel. Come on.

Written in the expansive way you present here, this is not as strange as you might think. We’re talking about a small portion of text being compared to a larger portion of text. The vast majority of words used in any text, in any language, are going to be the common but necessary words of that language – pronouns, articles (if any, some languages uses prefixes or suffixes and not separate words), helping verbs, etc. Many of the rest will vary from subject to subject, but not from text to text. (E.g. “epistle” and “epistolary” are very common words in entire genres of writing, and very rare elsewhere.)

If I grabbed 20-40 random words (in 7 to 10 groups of 1 to 6 words in each group, stipulating that when the words are in a group they must come from the same line, and when they are not they must come from a different line than any other group) from Crossan and compared them with an entire chapter of CS Lewis’ (very different) writing about Christianity and its gods, I would certainly find most of the words composing the first in the text of the second. Moreover, it would be inevitable that 2-4 word phrases from the first would be found in the second. Here, we aren’t distinguishing “it is” from “book binding” (where the definition of “binding” selected for translation absolutely depends on the adjacent “book”). I take the double underlined passages to be more like the latter, but nothing in your original assertion (that the odds are “astronomical” that the words of the GJW fragment would largely be found in the GoT) characterizes the “phrases”.

To be sure, I’m convinced by your full presentation of the evidence. This introduction to your argument, however, made you sound credulous.