I’ve explained before what “Q” theory is and why it is implausible and should have been abandoned by the Biblical Studies field decades ago (see Why Do We Still Believe in Q?). In short, we can prove conclusively that Luke used Matthew as a source (the evidence is extensive: see scholarship cited in On the Historicity of Jesus, pp. 269-70; for just a few examples, pp. 470-74). So every time Luke and Matthew share material not in Mark, we already have a sufficient explanation for it. We don’t need to imagine into existence some hypothetical third source. Nor if we did could we make any confident declarations as to what was or wasn’t in it. Mark could have used it. “M” could be material in it that Luke just didn’t use. “L” could be material in it that Matthew just didn’t use. It could have contained all kinds of things, like crucifixion narratives, that neither of them chose to use. And so on. So the theory is largely useless, as well as not well-evidenced.

But there is a recurring methodological travesty that keeps Q theory alive that is worth calling attention to: defenses of Q (which is short for “Quelle,” meaning “Source” in German) always hinge on some modern scholar claiming magical knowledge of the secret thoughts and intentions of ancient authors. Instead of going at history in an empirically logical way, where you look at what an author actually said and didn’t say, at the choices they actually made, and then infer from that what their thoughts and intentions are (which is called “evidence-based reasoning”), Q apologetics starts with presumptions as to what an author’s thoughts and intentions were—typically phrased as what they “must” have been, thus insisting on some sort of existential laws of the universe compelling ancient authors to do modern scholars’ bidding, as if by backwards psychic causation. And then they use that unevidenced presumption to invent new conclusions about the evidence, which then becomes “new evidence” that they use as a premise in another step of reasoning. This is backwards logic, and decidedly not empirical. It should never survive critique or review, in any empirical field of knowledge. But as this field has an unhealthy affinity for Christian apologetics, its standards are quite low when apologetical methods are on display.

The General Case

This backwards methodology appears elsewhere in Biblical studies, so it’s a trope, not particular to Q studies. Such as when it’s argued that Luke “would never” have left out the Great Omission, “therefore” that text must not have existed in his copy of Mark, “therefore” that material was interpolated. Empirical reasoning—actual logical reasoning—would work the other way around: Luke chose not to use that material from Mark (everything from Mark 6:45 to 8:26), ergo we are warranted in working out what reasons he might have had not to. Luke is, in other words, signaling his intentions and concerns by this very choice (on both theories see Michael Pattem’s “Luke’s Great Omission and his View of the Law,” in New Testament Studies). Yes, you could say it’s still “possible” that, instead, Mark lacked that material in Luke’s day, and it was interpolated by someone, and thus ended up in the copy of Mark used separately by Matthew, for example, but not the copy used by Luke—but that’s a possibiliter fallacy.

In actual fact, the presumption of an interpolation (especially of such extraordinary length) is always very improbable. Even granting abundant evidence that such things occurred a lot in Christian literature, it’s still less frequent than once in every two hundred verses. Which is why you need evidence for such a proposal (and quite good evidence), not just its “mere possibility.” It’s all the worse that not only does its prior probability thus tank this hypothesis, but the evidential probability does as well: since we can prove Luke knew Matthew, the jig is up. Even if Luke’s Mark lacked that material, Luke’s Matthew didn’t, so it is still the case that Luke saw that material and chose not to use it. So “interpolation” becomes entirely ineffective as an explanation for the Great Omission. The evidence supports “Luke chose not to use it” well over “Luke never saw it.” So much for that.

Thus, sound—as in, actually empirical and actually logical—approaches to the Great Omission accept the choice theory and abandon the interpolation theory. After all, Luke often leaves material out, including big chunks (like the “Little Omission” of Mark 9:41-10:12). For a particular example, Luke conspicuously chose to omit Mark 14:3-9 (on which see the commentary by Julie Smith). And yes (as Smith rightly points out) we might not be able to determine Luke’s reasoning; we can only hypothesize. But that’s at least the correct direction to go in: from the evidence we have, to a conclusion about Luke’s thinking—not the other way around. And Smith does a good empirical job of honing in on what that thinking might be in this case. For example, she points out that that story in Mark 14 contains an apostolic feature repeated six times in Mark—and all six of those were omitted by Luke. That’s a tell. This is how you accumulate evidence for a conclusion about what someone’s intentions and thinking were.

Many theories have been proposed for Luke’s Great Omission (for just some of them see, for example, this Stack Exchange; and examples in Pattem, above). If I were to start exploring a hypothesis myself, it would be that Luke needed to cut material to add his own, given the limited length of a standard scroll at the time; and the material he cut exactly corresponds to two of three repetitious cycles of material in Mark (see On the Historicity of Jesus, pp. 412-13), which is a typical Markan practice (to employ triplets and threes, often to emphasize specific points; ibid.). That this finding (that that material exactly corresponds to two cycles in a sequence of three in Mark) is independent of Luke’s Omission (it was discovered and proved without reference to it) makes for an unlikely coincidence. It looks to me that Luke saw Mark repeating the same sequence of events three times (albeit each time with different stories expressing the same ideas), and saw an obvious economy in just using one of them (and conspicuously, the first of them). The rest could go. So they did. This is how you use evidence to reach a conclusion.

In Q apologetics we get the same kind of backwards logic (for some examples, see this Stack Exchange). We are told Luke “would never” change something in Matthew. He “would never” reverse elements of the nativity narrative or rewrite even the shared material in his own words; he “would never” move things around into a different order; he “would never” change the genealogy; he “would never” change the story of how Judas died; he “would never” leave out Matthew’s story of a mighty angel descending to open the tomb; he “would never” only sometimes prefer Matthew’s rewording of Mark; he “would never” simplify Matthew’s text (e.g. turn his “blessed are the poor in spirit” into just “blessed are the poor”); and so on. So he “can’t” have been using Matthew every time; yet Luke and Matthew remain largely verbatim even in those sections, therefore there “must” have been a third document, conveniently lost and somehow completely unknown to anyone of the time. 

Correct logic would go the other way around. Instead of claiming magical knowledge of what authors “would never” do, you would look for evidence of what authors actually do, and then build your generalizations from those actual particulars. For example, in general background, we know rewriting a source in your own words was actually the most preferred method of composition taught in the schools Luke and Matthew had to have attended, and so it cannot be the case that Luke “would never” do that. Likewise, across ancient literature, we have countless examples of later versions of a text being simplified, therefore it cannot be the case that Luke “would never” simplify a source text. And, in internal context, we have many examples of how Luke uses sources—because we have one of them: Mark.

So. What do you think happens when you look to see if Luke changes Mark? Gosh. So much for the Q apologists. Luke often changes shit. You can see some examples in this sample sheet, which also evoke obvious causal motives for Luke’s choices; many more examples were surveyed long ago by Maurice Goguel in “Luke and Mark: With a Discussion of Streeter’s Theory,” in The Harvard Theological Review. There are many more. Does Luke ever rewrite Mark in his own words? Yes, he does (cf. Luke 8:19–21 and Mark 3:31–35; Luke 13:6-9 and Mark 11:12-14; Luke 21:20-21 and Mark 13:14-15; etc.). Does he ever outright change Mark into something different? Yes, he does—just research all the ways Luke contradicts Mark: for instance, he relocates the resurrection appearances from Galilee to Jerusalem, reverses Mark’s ending from the women telling no one to the women telling everyone, completely changes what Jesus says on the cross, expands the narrative of the two criminals executed beside him, and so on. Does he ever reorder the events in Mark? Yes, he does (cf. Luke 4:16-30 and Mark 6:1-6; Luke 22:24-27 and Mark 10:41-45; Luke 8 and Mark 4; and many more examples). Luke will even do both (reorder events from Mark and completely change them: cf. Luke 7:36-50 and Mark 14:3-9). Does he ever simplify the text of Mark? Yes, he does (cf. Luke 8:43-44 and Mark 5:25-26; Luke 9:47-43 and Mark 9:21-24; Luke 17:1-2 and Mark 9:42; Luke 22:31-34 and Mark 14:27-31; etc.). Does he ever omit seemingly important material from Mark? Yes, he does (as we just saw). And does Luke ever prefer Matthew’s rewording of Mark, a clear give-away that he knows and is using Matthew? Yes, he does (so many times that they have their own name: the so-called “Minor” Agreements; see Goodacre, “Too Good to Be Q”). In fact, Luke at one time quotes Matthew’s Nativity text verbatim (“and you will call his name Jesus,” cf. Luke 1:31-32 and Matthew 1:21), which adds even more to the case. Meanwhile, all like behavior can be found throughout Matthew’s use of Mark, too, proving all this behavior was normal for authors generally, not some peculiar affectation.

So if that’s how Luke treats Mark as a source, we can expect that’s how he’d treat Matthew as a source. Legitimate evidence-based reasoning thus gets us exactly opposite conclusions to the backwards logic of Q apologists. Authors were creative. They did not think or act like Q apologists irrationally insist. And this conclusion is based on evidence; unlike Q apologetics, whose premises are maintained only by ignoring all pertinent evidence.

The MacDonald Case

We find this backwards logic (and claims to “magical knowledge” about what authors “would never” do) even in the otherwise more credible approach of Dennis MacDonald, whose version of Q theory has the merits that it admits that Luke used Matthew as a source, that Q wasn’t a sayings source but a full Gospel (a complete narrative with an agenda and argument through-line, and not a random hodgepodge like the Gospel of Thomas), and that it followed the same literary conventions as the other Gospels (it was composed in Greek, used the sequenced pericope method of assembly, and emulated other literature for its content—in particular, the Greek text of the book of Deuteronomy). All of that is provably true (however much biblical scholars want to deny them, the actual evidence is extensive and not honestly dismissible). But none of that actually entails his (or any) Q hypothesis. Because all of that just sounds like…Matthew. Why then do we need to imagine a Q? We have “Q”!

Hence, the alternative explanation for all this same data is that Matthew “is” Q. After all, we already know Matthew extensively employed the Moses narratives (including Deuteronomy) to expand and rewrite the story of Jesus as inherited from Mark (the whole Nativity narrative; placing a Sermon on the Mount; even the choice of inserting Five Great Discourses to evoke the five Mosaic books of the Torah—all evince this plan, as does a great deal else). So we don’t need to look anywhere else for where all this emulation of Deuteronomy came from. And we can plainly see all the rest here already (Matthew is a full Gospel, following recognized literary conventions, that routinely emulated other literature for its content, e.g. not just the Torah, but the book of Daniel was raided for inspiration and material—from the role of inserting Magi into the Nativity, to the complete rewrite of the empty tomb narrative, on which see Proving History, pp. 199-204).

So there really isn’t any case to be made for MacDonald’s theory of Q either. It’s at least a more credible theory, but credibility cannot be substituted for evidence, and there just is no evidence for his theory. And that’s by the only pertinent definition of evidence there is: observable facts that are more likely on the theory they are evidence for than on any other explanation of how they came about (see my article on defining evidence in SHERM). For example, MacDonald has an elaborate theory of how Papias “mentions” Q, but it is a convoluted mess that relies on contentious suppositions, and then ignores all the incongruences between his theory and the actual data, or else dismisses it with apologetics rather than logic (like “explaining away” inconvenient evidence rather than presenting actual evidence those excuses are true, or that their being true would even help his case). 

MacDonald wants Eusebius’s quote of Papias claiming “Matthew set in order the logia,” meaning the sayings or sentences from or about Jesus, “in the Hebrew dialect and each interpreted them as he was able” to mean Papias knew of multiple different Greek texts supposedly being translated from that Hebrew original. MacDonald admits there can’t have been any Hebrew original, so Papias’s theory is defunct. But it’s Papias’s belief that there was more than one Greek Gospel that he thought derived from the same Hebrew source text that is pertinent to MacDonald’s argument. This would be congruous with Q. But it would also be congruous with any general lore about any Gospels with overlapping material, like Luke, or the Gospel of Peter, or anything else.

Since we don’t have Papias, we cannot actually say (despite MacDonald having an elaborate argument for saying it) that Papias did not know of Luke merely because Eusebius only preserved direct quotes from him regarding Matthew and Mark—both of which completely inaccurate, and thus which in that case might not even refer to the Gospels we know. Earlier in the same chapter where these appear, Eusebius relates an account of the Gospel of Luke without attribution (likewise for John), and it has been noted this may well derive from Papias. Eusebius, in History of the Church 3.39.14, says that what follows (his quotations from Papias about Mark and Matthew) are “other” (sic) statements Papias made about the Gospels, which “add” (sic) to what Papias had already said about the Gospels that Eusebius had “set forth before” (sic), clearly implying what Eusebius related about the other Gospels earlier, which was there unsourced, was in fact a paraphrase of Papias, and therefore here Eusebius is just adding two additional quotes only about Mark and Matthew because they add important details Eusebius wanted to include (for whatever reason is unclear, as they are not in Eusebius’s preceding account of the four Gospels, yet immediately follow Eusebius’s declaration that Papias was an unreliable idiot; but nevertheless Eusebius declares them “necessary” to add, as if by afterthought).

In fact the Greek that Eusebius quotes from Papias is not precise enough to know if he was even referring to translations at all. Without context, we can’t tell (and even Eusebius himself does not clarify). Because hêrmêneusen can simply mean “interpret” in the usual sense (“explain, expound, write about”), and thus refer to exegetes, not variant Gospel translations all labeled “Matthew” (as MacDonald’s theory requires). And even if it means “translate,” Papias could here simply be referring to our Matthew and Luke. The words Papias uses, which Eusebius selected to quote and MacDonald attempts to make much of as similar to what we find in Luke, could in fact even be an apologetic rejoinder to those who may have been quoting the preface of Luke against the reliability of Mark and Matthew, thus explaining overlaps in terminology between that preface and Papias, as well as the purpose of even saying these things in the first place—and for Eusebius to add them here.

In the end, MacDonald’s theory about Papias attesting to Q rests on a mountain of conjectures and suppositions, none of which can be established as fact, and some of which are quite doubtable. At worst, they are a stretch; at best, they are a wash against competing theories. This is not a valid empirical edifice on which to build a premise from which to argue “for Q.” Yet this is typical of the broken methodologies used to defend Q generally. Quite simply, this is just more apologetics in defense of a pre-determined conclusion, not objective historical reasoning. And the same holds for MacDonald’s entire case for his version of Q, which largely rests on a single conjecture about his ability to magically “know” that when Luke simplified a saying in Matthew that’s not in Mark, this “means” Luke is consulting a text that says something more like what Luke is saying, which Matthew had embellished, and therefore this “proves” Q is not Matthew. But as we just saw, no such principle is valid. Luke often does the same thing to Mark. So it cannot be argued that when Luke does it to Matthew that this means he’s then getting such material from somewhere else. 

It’s notable that when I pointed this out to MacDonald recently, he got quite angry and sought to browbeat the point away without ever actually responding to it. Which does typify an apologist defending a dogma rather than a historian trying to ascertain the truth independent of their hopes and desires. MacDonald is an excellent scholar, whose discoveries regarding the text of the Gospels are generally more than defensible, and total folly to neglect; but this appears to be a rare blindspot for him. He is too invested in his theory to let it go. “Simpler” does not mean “more primitive” as Q apologists insist, and MacDonald is still too sold on that old fallacious equivocation to concede the point. All Q apologetics operates by circular argument like this: it will be assumed that Q would be written a certain way, and when Luke’s version agrees with that assumption, this will then be used as “proof” that Q was indeed written that way—and therefore existed. Round and round it goes.

This is not logical. Yet MacDonald deploys the same errors in defense of his own project. For instance, he will offer many other criteria he has invented by which to “prove” Luke must be using a source other than Matthew, but they all rest on elaborate systems of presuppositions or weakly evidenced assertions, such as about what “themes” Q emphasized that Luke didn’t, or what words “are” or “are not” Lukan in style, or whether authors “never” pick sayings out of narratives and add them to other sayings or change their narrative context, and the like, which all ultimately end up circular. The conclusion is reached only by assuming it already in one of the premises. “Q had x theme, and Luke didn’t; Luke mentions x theme; therefore Luke must there be quoting Q.” Spot the fallacy.

Likewise, declarations as to what is or is not “Lukan” in style are extremely fragile, and often not well supported. It is simply not the case that even a hapax legomena, a unique word or phrase, “must” derive from someone else; ancient authors chose to employ their own hapax legomena quite often. So you need more evidence than that. In this and every other respect, the insistence that authors “never” do things will, when pressed, be admitted false, but then that admission will be forgotten when drawing an inference later on down the line. In every case, instead of comparing one explanation against the best available competing explanation for every detail of concern, and showing with evidence that one has a higher probability, we get ad hoc intuitions based on very little evidence for what a scholar “feels in their gut” would be the idiom Luke would use or the thing he would say or do, and when Luke doesn’t behave according to their gut feeling, this becomes “evidence” that Luke is quoting someone else. But this isn’t how evidence works. It substitutes hypothesis for fact, intuition for reason; rather than establishing the hypothesis is true on a strong basis of evidence. It is leaping from possibility to probability. It is assuming a generalization never has exceptions that must be tested for first.

Given the deeply suspect nature of his approach, MacDonald would need to prove, first, that his criteria work on some other project, before declaring they can work here (they work in textual criticism, but the way he adapts that to this is precisely what needs to be validated). I suggested they be applied to Luke’s use of Mark (because I strongly suspect that when he does this, his method will be proved ineffective), and he dismissed the idea as for various reasons impossible—which is an excuse to not do what has to be done to legitimate an otherwise suspect method. And even that would not suffice. Because then MacDonald needs to find some objective method of demonstrating that any given case actually meets those criteria. Gut feelings and intuitions will never cut it here. They are unexaminable and unanalyzable and thus useless methodologically. All one’s feelings and intuitions do is restate what one assumes, rather than defending it as correct (see Advice on Probabilistic Reasoning). That’s why we need to see how and why the conclusion logically follows from the premises. If you can’t produce that, you really shouldn’t be confident in your conclusions. To be fair, this is the behavior of most Q apologists in the field (or indeed, even most biblical historians altogether). It is not a defect peculiar to MacDonald.

To give you an example (one that came up in the recent Goodacre-MacDonald debate about Q on MythVision), consider the famous Beatitudes. A very common argument from Q apologists, repeated by MacDonald, is that when we see Matthew have Jesus say “blessed are the poor in spirit,” and in Luke this becomes simply “blessed are the poor,” this is evidence of Luke’s text being “more primitive” and therefore “closer to Q,” so Matthew must have added the “in spirit” bit (for whatever reason), and this “therefore” proves Q existed. If you found that argument already a bit dubious, you’re starting to get a feel for what I mean. If you want to maintain the premise that Luke would not drop the “in spirit” part and therefore “must” have gotten his text from Q and not Matthew, you first need to compare that assumption against the best alternative explanation: that Luke had a good reason to drop that bit when he took up this text from Matthew. 

And here we see why it is so crucial to admit that, even if there was a Q, Luke still knew and used Matthew (as MacDonald does, setting him apart from many Q apologists who fear exactly the consequence of that admission I am about to relate): this means Luke saw “poor in spirit” in Matthew’s version and still chose to omit “in the spirit.” This is harder to explain for MacDonald. Because the moment you come up with any reason why Luke would prefer the original to Matthew’s adaptation, you have just come up with a reason why Luke would change Matthew to what he wanted himself. In other words, you just blew up your own theory, by providing a perfectly good explanation already as to why Luke would drop that word and just stick with “the poor.”

You cannot try to avoid this by saying, “Well, Luke just usually preferred Q to Matthew’s revisions,” because that hypothesis is decisively refuted: quite a lot of the time, Luke actually likes Matthew’s revisions to things. And if you try to say, “Well, we can’t really know that Luke chose Matthew over Q at all, because when Luke agrees with Matthew entire, we don’t know what Q may have said differently,” because then you’ve just exposed the circular argument in your own position: if your assumption is that when Luke disagrees with Matthew in Q material Luke is following Q, and when Luke agrees with Matthew in Q material Luke is still following Q, your conclusion is now already in the premise. To see what I mean, ask yourself how you would falsify the conclusion that Luke never preferred Matthew to Q. Hm. Right.

By contrast to this circular reasoning, where you just assume when Luke drops or changes something from Matthew he isn’t using Matthew (and then use examples of when he does that as evidence that Luke isn’t using Matthew), let’s look at what an evidence-based method would do with this same information. Goodacre lays it out: Luke is more concerned about income disparity than the other evangelists. Not only does he add one of the most elaborate and colorful parables on the subject (that of the Rich Man and Lazarus) as well as the conspicuous declaration of Zaccheus (Luke 19:5-9) and the Magnificat besides (in which “the low will be made high, the hungry filled, and the rich sent away empty”), Luke doubles the amount of material about “the poor” throughout his entire Gospel (at least 8 distinct mentions, to the 4 we find in Matthew and 3 in Mark), and specifically (and uniquely) has Jesus say of himself, “The Spirit of the Lord is upon me, because He did anoint me to proclaim good news to the poor” (Luke 4:18), which is a direct description of what he then is doing in Luke 6:20 when he says, “Blessed are the poor.” Luke also retains the tale of the Widow’s Mite from Mark (while Matthew dropped it), and makes “the poor” a focus of Matthew’s Parable of the Banquet: where Matthew has the Lord direct his servants to invite “anyone,” Luke instead has the Lord first say “the poor, the crippled, the blind, and the lame,” and only then when all have come in, to go invite whoever is left: cf. Luke 14:15-25 and Matthew 22:1-14. It therefore would actually make no sense for Luke to suddenly throw in the self-defeating qualifier “poor in spirit” when the time came to have Jesus preach the point, and every sense for Luke to prefer the more direct statement of simply “the poor.” That is entirely in keeping with his entire practice and intent, multiply evidenced all throughout his Gospel. 

This isn’t something we are “intuiting” about Luke; this is something we have a whole bucket of evidence establishing is the case about Luke. The Q apologist has no rebuttal to this. It is simply obvious that Luke saw the text in Matthew, recognized it was not at all suited to Luke’s program, and thus dropped the unsuited word. There is no case to be made here that Luke saw his preferred version in some other text. Maybe he did. But you simply can’t argue it’s likely. It’s more likely Luke is just redacting Matthew exactly in line with Luke’s well-demonstrated design. (And for those who might wonder why this is so much a focus of Luke, I’d call attention to his declared upper-class audience in his preface, and as evinced by his more florid and elite style and even genre of composition: Luke is deliberately writing to wealthy members of the church, and thus is taking the opportunity to really drive home the point he most wants to make to them. Hence you’ll see that theme continued in Acts.)

Conclusion

This is what happens over and over again with every “example” that is supposed to prove any theory of Q (MacDonald’s or any other). Sometimes the only way to get to their argument is to adopt a huge edifice of assumptions, none of which are empirically proven, and some of which are dubious or outright disprovable. Sometimes the only way to get to their argument is to adopt a circular presumption, by which you interpret what an author does as evincing a reliance on Q, and then use that as evidence the author is evincing a reliance on Q. But worse, all of the time, the best alternative hypothesis is never being properly compared with the Q hypothesis. Rather than sincerely and ardently trying to disprove Q and failing (the only way to ever validly prove anything), they evade exactly that method and engage in verification fallacies instead, where they “see” everything as conforming to their theory—and then use everything as evidence for their theory—without correctly taking into account the fact that each of those things may well have as good or even better an explanation. Of course, already, prior probability cannot favor Q, as the “Luke redacted Mark and Matthew” hypothesis contains fewer assumptions, all of them in evidence (we have Mark and Matthew, and we can prove Luke used them); Q does not. So you really need good evidence for Q. And there just isn’t any. And as long as historians keep using illogical and backwards and unvalidated methodologies, they’ll fail to admit this.

§

To comment use Add Comment field at bottom or click a Reply box next to a comment. See Comments & Moderation Policy.

Discover more from Richard Carrier Blogs

Subscribe now to keep reading and get access to the full archive.

Continue reading