Jesus and the Beelzebul Controversy: A Devilish Synoptic Puzzle

Lately, I can’t stop thinking about the Synoptic Problem. This is my third article on the subject, so if you’re not sure what the Synoptic Problem is, I suggest that you first check out an earlier post I wrote on it.

In my second article on the subject, I looked at the Parable of the Talents/Minas, which is found in Matthew and Luke, but not in Mark. Under the standard two-source theory, the source of such material is a hypothetical document called Q. According to noted Q skeptic Mark Goodacre, Luke copied that parable and other double tradition passages from Matthew. My own analysis suggests a third solution supported by a small but increasing number of scholars: that Matthew copied Luke — to be precise, an earlier version of Luke we can call proto-Luke. We’ll call this view “Matthean Posteriority” for convenience.¹ 

The problem of the “minor agreement” passages

Although the two-source hypothesis (Q hypothesis) enjoys widespread acceptance, there are a number of well-known passages that pose a difficulty. The problem is this: because Luke and Matthew are supposedly unaware of what the other has written, it is highly unlikely that they should ever make exactly the same changes to material from Mark. And usually they don’t.

But sometimes they do. These so-called “minor agreements” (one scholar has counted over 700 instances!) have Luke and Matthew agreeing with each other against Mark, and they can’t always be explained by coincidence or manuscript corruption.

Synoptic Problem — Minor Agreements 1

The passage in Mark 3:19b–35, which has Jesus accused of being in league with Beelzebul, is one of the most problematic examples. This section, a series of pericopes related by topic as I will explain below, includes Jesus’ rejection by his family, the Beelzebul accusation, and several brief teachings about exorcism and blasphemy.

Both Luke and Matthew copy this passage, with several interesting differences:

  1. They both relocate it to a scene not found in Mark, in which a mute or mute/blind demoniac undergoes exorcism. The relevant passages are Luke 11:14-28 and Matt. 12:22-50. Matthew also has a second, highly abbreviated version in 9:32-34.
  2. They both reword the response Jesus gives to his accusers, and their changes largely agree with each other, though both retain some of Mark’s wording.
  3. They both include extra teachings on exorcism (Lk 11:19-20, Mt 27-28), opposition to Jesus (Lk 11:23, Mt 12:30), and unclean spirits (Lk 24-26, Mt 11:43-45) that are nearly word-for-word identical, and in the same relative sequence.

Revised Q hypothesis to the rescue

In order to explain these passages without discarding the otherwise helpful Q hypothesis, many scholars hold that there must be a few places where Mark and Q contain similar content. Thus, Luke and Matthew combine both sources when they get to these passages, and the material they get from Q supplies this agreement against Mark. It’s not a bad theory.

Synopsis of the Beelzebul Controversy (click for PDF)

Synopsis of the Beelzebul Controversy (click for PDF)

I wanted to see how well this explanation holds up, so I copied the text of Mark 3:19b–35 into a table alongside Luke 11:14–28, Matthew 12:22–50, and some other parallel texts. Correspondences between the various Synoptic Gospels are marked in colour-coded text. At first glance, the complexity of the linkages is intimidating, but some patterns emerge when the text is laid out this way. Let’s put ourselves in the shoes of the evangelists who penned Luke and Matthew, and see if we can trace the process by which they worked.

Let’s assume that Q has some overlap with Mark. Luke is sitting there with Mark and Q in front of him, copying from Mark chapter 3 but adding material from a similar Q passage as well as his own contributions. Matthew, completely independently of Luke, is doing the same.

We should expect Luke’s choice of Markan material to copy to differ somewhat from Matthew’s. Thus, we should expect to find (1) shared Mark-Luke material that Matthew doesn’t have, and (2) shared Mark-Matthew material that Luke doesn’t have. And when we look at Matthew 12, sure enough, there are plenty of words and phrases shared with Mark but not Luke.

But the converse isn’t true: not a single word borrowed from Mark 3 into Luke 11 is missing from Matthew 12. Not one. The odds of that happening are very small.

Synoptic Problem — Minor Agreements (Beelzebul)

To explain this, Q theorists must add another wrinkle to the hypothesis: Luke completely set Mark aside and copied only from Q for this passage. That way, there is no Mark-Luke material that excludes Matthew, because there is no Mark-Luke material at all.

That’s still problematic, because Luke has plenty of verbatim agreement with Mark, only every word of it is shared with Matthew. So for the Q hypothesis to be correct, (1) Luke must be uncharacteristically ignoring Mark, and (2) the Q-Mark overlap must involve significant verbatim agreement, which means that either Mark copied Q or Q copied Mark; so we’ve added a new level of complexity to the Q hypothesis just to account for Luke’s odd behaviour.

In fact, the characteristic way that Mark assembles his narrative almost necessitates that Q, if it existed, was based on Mark.² Otherwise, it would not share so many of Mark’s unique traits as well as his wording. Let me explain:

It has often been observed that, aside from the Passion narrative, Mark’s story is generally not in any discernible chronological order. His passages frequently consist of what K. L. Schmidt described almost a century ago as “pearls on a string” — loose collections of parables, sayings, and other Jesus traditions that are arranged by topic or theme.³ However, Mark shows great sophistication in how he orders his material; he is especially fond of nested chiastic structures — e.g. stories that start with topic “A”, move on to “B”, and then conclude with “A” again.

The Beelzebub Controversy is composed precisely this way.⁴ It consists of the following independent snippets. They have been combined into what seems to be a chronological sequence of events but is actually a careful literary arrangement based on theme.

  1. Jesus is at home, surrounded by crowds, and his family tries to restrain him, thinking he’s insane. (3.19b–21)
  2. Scribes from Jerusalem accuse Jesus of being possessed by Beelzebul to cast out demons. Jesus refutes them with a parable about how kingdoms and houses cannot be divided against themselves. (Note that this is not preceded by an actual exorcism. The link with #1 is the theme of “accusation of demon possession”.)
  3. Jesus gives a parable on binding a strong man in order to plunder his property. (This is related to the topic of exorcism but not really a response to the Beelzebul accusation.)
  4. Jesus gives a teaching about blasphemy and forgiveness in response to unnamed accusers who say Jesus is possessed by an unclean spirit. (The topic again is “accusation of spirit possession”.)
  5. Jesus is at home, surrounded by crowds, when his family shows up, wanting to see him. Jesus responds with the teaching that those who do God’s will are his family.

These elements don’t seem to originate from a single story. Instead, they describe situations, parables, and teachings associated by common themes, arranged in a “Markan sandwich” that begins and ends with Jesus at home, first being rejected by his family and then rejecting them himself in favour of those who follow God.

Is it likely that Q not only used the same peculiar literary technique as Mark, but happened to include the same scenarios and parables in the same passage and in precisely the same order? No; if Q existed, it must have been copying Mark and adding material (like the exorcism of the demoniac) to strengthen the narrative. But this contradicts the standard description of Q as a “collection of sayings” with no narrative!

The Blind and Mute Man Possessed by Devils (Le démoniaque aveugle et muet) by James Tissot

The Blind and Mute Man Possessed by Devils (Le démoniaque aveugle et muet) by James Tissot

Do we really need Q?

Adela Yarbro Collins, in the Hermeneia commentary on Mark, inadvertently lends support for a Q-less solution when she writes:

The observation that the author of Matthew combined Mark’s version of this passage with another that Luke follows exclusively indicates the existence of a parallel passage in Q. (pp. 227–228)

In other words, “Matthew combined Mark with a text that resembles Luke”. Which is indistinguishable from saying that Matthew appears to have combined Mark and Luke.

Let’s examine the possibility that, contrary to the Q hypothesis, Luke followed Mark here as usual, and Matthew copied from both Mark and Luke. For starters, Luke’s changes make a lot of sense. We can observe that Luke is trying to turn Mark’s string of pearls into a more cohesive sermon. He doesn’t like the theme of Jesus’ family calling him crazy or the abrupt start of the Beelzebul accusation, so he puts the passage into an entirely new context. It now starts with Jesus performing an exorcism. Matthew copies Luke’s new intro, using much of the same wording, and embellishing the description of the demoniac so he is blind as well as mute.

Mark 3:19b–21: Then [Jesus] went home; and the crowd came together again, so that they could not even eat. When his family heard it, they went out to restrain him, for people were saying, “He has gone out of his mind.”

Luke 11:14: Now he was casting out a demon that was mute; when the demon had gone out, the one who had been mute spoke, and the crowds were amazed.

Matt. 12:23a: Then they brought to him a demoniac who was blind and mute; and he cured him, so that the one who had been mute could speak and see. All the crowds were amazed and said, “Can this be the Son of David?”

Next, Luke copies Mark’s accusations of collusion with Beelzebul, but changes the accusers to be members of the crowd (that just watched the exorcism) rather than “scribes from Jerusalem”. Matthew makes the crowd more appreciative and pins the accusation on the Pharisees, who are usually his go-to antagonists.

Mark 3:22: And the scribes who came down from Jerusalem said, “He has Beelzebul, and by the ruler of the demons he casts out demons.”

Luke 11:15: But some from among [the crowd] said, “By Beelzebul the ruler of the demons he casts out the demons.”

Matt. 12:23b-24: But when the Pharisees heard it, they said, “It is only by Beelzebul, the ruler of the demons, that this fellow casts out the demons.”

Luke takes Mark’s response parable about divided kingdoms and houses. He expands it with new exorcism teachings (vv. 19–20). Matthew copies Luke’s additions but adds several expressions from Mark that Luke did not use, particularly one about Satan casting out Satan.

Mark 3:23-26: And he called them to him, and spoke to them in parables, “How can Satan cast out Satan? If a kingdom is divided against itself, that kingdom cannot stand. And if a house is divided against itself, that house will not be able to stand. And if Satan has risen up against himself and is divided, he cannot stand, but his end has come.

Luke 11:17-18: But he knew what they were thinking and said to them, “Every kingdom divided against itself becomes desolate, and house falls on house. If Satan also is divided against himself, how will his kingdom stand? … Now if I cast out the demons by Beelzebul, by whom do your sons cast them out? Therefore they will be your judges. But if it is by the finger of God that I cast out the demons, then the kingdom of God has come to you.

Matt. 12:25-28: He knew what they were thinking and said to them, “Every kingdom divided against itself becomes desolate, and every city or house divided against itself will not stand. If Satan casts out Satan, he is divided against himself; how then will his kingdom stand? And if I cast out demons by Beelzebul, by whom do your own sons cast them out? Therefore they will be your judges. But if it is by the spirit of God that I cast out demons, then the kingdom of God has come to you.

Luke completely rewrites Mark’s parable about the strong man so it works better with his expanded teaching about casting out demons. Matthew discards Luke’s version and uses Mark’s.

Luke inserts a statement by Jesus that “whoever is not with me is against me…”, which is based on Mark 9:40. Matthew includes the same verse, word for word, in the same location.

Luke doesn’t think the blasphemy discourse is relevant here, so he skips it, inserting it later on in Luke 12:10. Matthew keeps Mark’s version mostly intact. (It has some changes that mirror Luke’s in 12:10.)

Matthew insert two discourses here that Mark doesn’t have: “The Tree and Its Fruit”, and “The Sign of Jonah”. Luke has the same pericopes, but they are located elsewhere.

Apparently inspired by Mark’s reference to unclean spirits, Luke inserts a fairly long discourse about the behaviour of unclean spirits. Matthew copies it word for word, adding to the end a warning aimed at “this evil generation” that connects it to his “Sign of Jonah” pericope.

Luke omits Mark’s closing about Jesus’ family looking for him (relocating it to 8:19). Instead, he rewrites the ending so that someone from the crowd blesses Jesus’ mother — a blessing Jesus deflects onto those who do God’s will. Matthew keeps Mark’s closing, but makes it explicitly the disciples whom Jesus considers family.

The end result: Luke turns Mark’s string of demon-possession-related accusations and discourses into a sermon on exorcism and God’s kingdom. Matthew has kept Luke’s sermon mostly intact but includes more content of Mark’s.

Possible solution to the minor agreements in the Beelzebul Controversy passages

Possible solution to the minor agreements in the Beelzebul Controversy passages

The case for Matthean Posteriority

Though the Q Hypothesis can be used to explain the Beelzebul Controversy passage, it suffers from a few weaknesses. It requires additional levels of complexity, strange Mark-Q overlap, and odd behaviour by Luke. That doesn’t mean it’s wrong — and given scholars’ preference for it, it must be taken seriously — but if we value parsimony, we should consider other possibilities that are simpler.

Matthean Posteriority seems to be just such an option. It accounts for all peculiarities of the texts using just the three documents we already know exist.

Why not the reverse — the position espoused by Goodacre that Luke copied Matthew? That’s certainly possible, but a few points tilt in the favour of Matthew copying Luke (or proto-Luke):

  1. Where Luke and Matthew share material, Luke seems to be more primitive. For example, take Luke 11:20 (“by the finger of God”) versus Matt. 12:28 (“by the spirit of God”). It is more likely that “finger” would be changed to “spirit” rather than the opposite.
  2. Matthew shows fatigue from copying Luke. Throughout his Gospel, he always uses the phrase “kingdom of Heaven” in place of “kingdom of God” — except in four places where he is copying another text and forgets to make the change. This is one of those places (Matt. 12:28).
  3. The lack of exclusive Mark-Luke material would mean that Luke completely ignored Mark for this one passage.

Ronald V. Huggins (1992, p. 3) has summarized the strengths of Matthean Posteriority as follows:

…Matthean Posteriority if ultimately defensible would be preferable to the two-source hypothesis in that (1) it easily accounts for the minor agreements… (2) it does not require the introduction of additional “entities” in the form of a hypothetical document to explain the existence of the double tradition, and (3) it avoids the problems caused by the generic ambiguity of Q.

By “generic ambiguity”, he refers to the problem of Q’s genre as I mentioned above. Some passages require it to be a narrative-less collection of sayings; others, like the Beelzebul Controversy, require it to be different kind of document.

The healing of the blind man, from the Rossano Gospels, 6th century

The healing of the blind man, from the Rossano Gospels, 6th century

All theories must be provisional.

Barring any remarkable manuscript discoveries — e.g. an early copy of the Gospels or any copy of Q — we cannot be too confident in any Gospel source theory. Our earliest complete manuscripts of Matthew, Mark and Luke date only to the fourth century. As far as the passages we looked at here, go, our earliest witness to Luke 11 is the fragmentary Papyrus 45 from the mid-third century. Just four verses from Matthew 12 (24-26, 32-33) survive from the third century in Papyrus 21, and even those contain two textual variants.

We don’t know what the earliest manuscripts looked like, and it’s entirely possible that earlier versions predated our canonical texts. Helmut Koester (2000, p. 214), for example, speculates that many of the minor agreements result from Matthew and Luke working from an earlier version of Mark. Kloppenborg (2000, p. 34) notes the likelihood that Luke and Matthew were not working off identical copies of Mark; and that, furthermore, we can be certain their versions of Mark differed from our own at least a little.

And, as I have already mentioned, our earliest attested version of Luke — known to its readers simply as the Evangelion — differed significantly from our canonical Luke (though not in the Beelzebul passage). Increasingly, biblical textual critics question whether “original texts” ever existed at all. William L. Peterson (1994, p. 136) wrote:

First and foremost among the problems is the difficulty of defining “original.” The Gospel of Mark illustrates the point. Is the “original” Mark the “Mark” found in our fourth-century and later manuscripts? Or is it the “Mark” recovered from the so-called “minor agreements” between Matthew and Luke? And which—if any—of the four extant endings of “Mark” is “original”?

For now, we are stuck with our fourth-century texts. The hypothesis that Matthew copied from Mark and Luke (or Proto-Luke/Evangelion) seems to fit well with the passages we’ve examined so far. I’m curious to see what further study turns up, as well the reaction of academia to recent proposals for Matthean Posteriority.


  • Adela Yarbro Collins, Mark: A Commentary (Hermeneia: a Critical and Historical Commentary on the Bible), 2007.
  • Ronald V. Huggins, “Matthean Posteriority: A Preliminary Proposal”, Novum Testamentum 34/1, 1992.
  • William L. Petersen, “What Text Can New Testament Textual Criticism Ultimately Reach?” in New Testament Textual Criticism, Exegesis, and Early Church History, CBET 7, 1994.
  • Helmut Koester, Introduction to the New Testament, Volume 2: History and Literature of Early Christianity, 2nd Edition, 2000.
  • John S. Kloppenborg, “Is There a New Paradigm?”, Christology, Controversy, and Community: New Testament Essays in Honour of David R. Catchpole, SNT 99, 2000.


  1. Scholars who have recently written in support of Matthean Posteriority include Ronald V. Huggins (see bibliography), Alan Garrow, and Matthias Klinghardt.
  2. T.E. Floyd Honey, in “Did Mark Use Q?” (JBL 62/4, 1943), attempts to show that Mark copied Q. His reasoning seems suspect for me; for example, he says that Mark omitted Q’s introduction, the healing of the demoniac, and “placed the story in his own setting”. But Mark’s placement seems nearly random, coming after the calling of the Twelve, which has nothing to do with exorcism and no narrative relationship to the Beelzebul accusations.
  3. Schmidt explained the confusing and often contradictory Gospel narratives in this manner: “The oldest Jesus tradition is ‘Pericope Tradition,’ that is, a tradition of individual scenes and individual sayings, which for the most part lack definite chronological and topographical identification within the community where they have been transmitted. Much of that which appears chronological and topographical is just the framework into which the individual pictures have been inserted.” Schmidt, Karl Ludwig, The Place of the Gospels in the General History of Literature, Trans. Byron R. McCane, Columbia: University of South Carolina Press, 2002.
    Quoted in David E. Aune, The Blackwell Companion to the New Testament, 2010, p. 142.
  4. Adam Winn (The Purpose of Mark’s Gospel: An Early Christian Response to Roman Imperial Propaganda, p. 8) correctly observes that Mark 3:22-30 is a collection of sayings grouped topically. He notes: “Mark’s apparent lack of interest in causality is a strong indicator that he does not write with history as his primary purpose.”

17 thoughts on “Jesus and the Beelzebul Controversy: A Devilish Synoptic Puzzle

    • I think you’ve found the weakness of this line of argument. What Goodacre believes, for instnace, is that Luke Followed Matthew, and Matthew followed Luke plus another source. So I asked him on his blog, well isn’t Matthew’s other source Q? Doesn’t that mean simply that Q was a separate document that Matthew knew and that Luke knew only through Matthew. His reply, in brief, was no, because Q would be a document seen by Luke and Matthew. Some are too close to the problem and get tangled up in semantics.


  1. I have a question concerning this I rarely see on any discussion of the Q document theory: What if Q was not a literal document but an oral narrative?

    I mean, oral narratives were still quite common around this era even as a great many of them were being fixed into writing, but before the fix-into-writing happens there almost certainly still would be an oral version of the story. This would not be a “narrative” in the true sense as it could be clipped, rearranged, and modified to the audience at hand, and probably was not told all at once, and possibly not even known in its entirety by many people. Some of the “important” stories would have developed a fixed schema, while others would be more mutable. Several of the written narratives, then, would be an attempt to gather these oral tellings together into a form of narrative more suitable for a written document.

    I don’t know nearly enough about oral storytelling to make more conjecture than that, I just wonder why “passages pulled from a shared oral narrative” is not often suggested as a possibility for Q.


    • Some Gospel material (Q and otherwise) could have come from oral traditions, but from what I understand, Q itself (if it existed) has to be a written source because of the identical wording you get between Luke and Matthew. You only expect that high level of correspondence from literary copying.

      Contrast this with the Gospel of Thomas, which has numerous paraphrases of synoptic parables, and very well might result from someone writing down the parables from memory, perhaps after hearing Matthew or Luke read in church.


  2. I’ve previously heard Carrier say that there’s no need for the Q source, and that he thinks the case against it is strong enough that he doesn’t know why it’s still such a popular theory. It was said in another context, and being tangential, it was not explained. I think I now have some idea, at least a proxy of an idea, what that case is. So many thanks for that.

    While I can see the reason for introducing and testing the theory of an unattested document, I find the fact that what such a source might say is a black box to be highly problematic, if for no other reason than different scholars are going to come to the table with different ideas about what it says, which can make their work a bit like comparing apples and oranges. In other ways, it can be problematic, as you’ve said, to explain one passage we need to be narrative-less, so we’ll say that at one time, and for other passages we need it to be the opposite, but maybe we won’t shine too bright a spotlight on that?

    I really think that the 4th century texts we’re “stuck with” are the result of a couple hundred years of development, and that it is not the case that there was an autograph that was more or less faithfully copied. That’s an apologetic assumption, and really doesn’t fit with what else we know about the Greek canonical and non-canonical texts.

    Have to hand it to you, accomplishing this kind of synthesis requires one to keep a lot of very detailed info in their head at once. Hats off. My deconversion was only about 3 years ago and I feel like I’m playing catch-up, trying to arrange all this stuff into a more likely and reasonable story…


  3. Thanks for the comment, ephemerol. I agree with you on all counts.

    I suspect there is a certain attraction for some in postulating the Q source, because if it exists, it gives us a source for understanding early Christianity that potentially pre-dates the canonical Gospels. (For the same reason, some have wanted to see the Gospel of Thomas as predating the Synoptics, though this seems unlikely.) But you’re on pretty shaky ground when you start doing history based on a document no one has ever seen.

    Conversely, very conservative scholars apparently object to Q on theological grounds, it being incompatible with the traditional church view and supposed divine inspiration of the Gospels. (If they accept Q, do they have to acknowledge Q as inspired scripture? If so, how could God allow it to go missing?)


  4. Heyy thanks for the really incredible blogging on this site! I just discovered this place and I was immediately blown away by the depth of discussion on interesting topics, and the no-B.S. voice you use to navigate them. Also (since this is my first time commenting here), I need to praise the clarity of the writing, and the beautiful presentation of each page!

    I’m wondering if you’re familiar with Dennis MacDonald’s “Q+/Papias” synoptic solution? Do you have an opinion on it?


    • Thanks for the nice comment, Jeff. I confess, I haven’t read MacDonald’s book on his synoptic solution. When I get to it, I will see if any test cases are worth exploring here. Thanks for the suggestion!


  5. Might Q have not been orignally a separate document but rather an ‘annotated’, be it in the margins or between the lines, copy of Mark? Some of the differences between Luke and Matthew could then be caused by one author refusing to include/exclude notes they disagred with?


    • I think Q is just what Luke copied from Matthew or made up himself. The Q hypothesis is apologetics to explain the similarities between Matthew and Luke without having to explain why Luke would have rejected other parts of Matthew.

      Matthew makes a big deal about there being 14 generations from Abraham to David, David to the Exile, and from the Exile to Jesus but he left out four names in the second set and one must count one name twice or count the Exile as a generation to get 14 for each set as listed. Either is a good reason for Luke to reject Matthew’s genealogy.

      Perhaps Luke couldn’t bear the thought that God would allow all those Bethlehem babies to die while Jesus escaped, which is an explanation for the other nativity story.


  6. If we’re talking about a “proto-Luke”, could this be related to Irenaeus’ accusation that Marcion’s Canon included a “mutilated” Luke that lacked a genealogy or nativity?


  7. A very problematic element in the ‘Q as a collection of sayings’ theory is that it would include one of the most striking bits of narrative about Jesus, when he goes toe to toe with the devil, flying around to prominent locations. If Mark has, for argument’s sake actually found the Beelzebul story in Q, as some like Mack have suggested, that would be striking in itself because Q has hardly any names. Aside from Jesus and John, there are a handful of characters from the Hebrew scriptures and a (recent?) prophet called Zachariah killed between the altar and the sanctuary. As Q included Diabolos maybe it included Beelzebul as a participatory character. Maybe it included other direct encounters between Jesus and supernatural figures. In which case we might be looking at extensive euhemerisation by Mark, who in order to create a more naturalistic biography has only demons who are encountered within human characters not, as in the Temptation, flying around independently to take on Jesus. Mark’s account of the temptation would thus be a truncated account of the full blown version Luke and Matthew,


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s