The Holtzmann-Gundry Solution to the Synoptic Problem (Three Source Hypothesis)

I’ve been immersing myself in the Synoptic Problem of late. Based on the limited studies I’ve done before, I’ve always gravitated towards the four source theory (Mark, Q, L, M). Although I have had my curiosity aroused by the Farrer-Goulder-Goodacre theory of Mark–>Matthew–>Luke (eliminating the need for Q). And I have to say that Mark Goodacre has, at least, sown seeds of doubt in my mind about Q.

For those who think the Synoptic Problem is dull, they should read Mark Goodacre’s Is the Synoptic Problem Tedious? Any way, I’m gravitating towards a solution to the Synoptic Problem that is not necessarily new, but I think an underrated one in the discussion. Here’s what I think …

1. Marcan priority.
2. Double tradition is Q-lite and other synoptic traditions like non-Marcan narrative material  (OST).
3. Matthew used Mark, Q-lite, OST, and M.
4. Luke used Mark, Q-lite, OST, L, and Matthew .

(Diagram from Stephen Carlson’s Synoptic Problem Website).

The Synoptic Problem is so complicated and challenging that it can make the most determined researcher want to put his or her head in the sand to somehow escape from the frustration, or else incline them to take a yoga class in the hope that attaining nirvana might bring release from the intense intellectual suffering and provide closure to their curiosity. Such respite has eluded every Synoptic researcher apart from the most self-assured.

Skepticism aside, I wish to optimistically put forward a tentative solution for the Synoptic Problem. Nearly half a century ago, E.P. Sanders proffered his opinion on future for the Synoptic Problem: “I rather suspect that when and if a new view of the Synoptic problem becomes accepted, it will be more flexible and complicated than the tidy two-document hypothesis. With all due respect for scientific preference for the simpler view, the evidence seems to require a more complicated one.”[1] I think that is indeed the case. While I think that the two (four) source hypothesis is basically correct, there are several features of the theory that need to be tweaked in order to accommodate to what I believe to be the slightly more complex nature of the problem, specifically, that Luke used Mark, Q, and Matthew.

I call the position the Holtzmann-Gundry hypothesis (though it has also been called the “three source hypothesis”). It is named after the German scholar H.J. Holtzmann its first proponent and American scholar Robert H. Gundry its most well-known proponent. In a nutshell this view is (1) Marcan priority; (2) Matthew used Mark; (3) Luke used Mark, Q, and other traditions; and (4) At a later point, Luke has incorporated Matthew into his own work.

To give a brief history of this position, H.J. Holtzmann argued for Marcan priority and postulated the existence of Q (or L as he called it called for “logia”) in his celebrated Die synoptischen Evangelien,[2] but he modified his view in a journal article published fifteen years later that tacitly suggested that Luke also used Matthew.[3] Holtzmann afterward declared in his Einleitung that he had come to change his mind on three areas of the Synoptic Problem:[4]

1)       Not everything in Matthew and Luke that goes beyond Mark can be accommodated in the sayings collection. At times elements from the collection are more highly modified by Luke than they are by Matthew. It is possible that it contained narrative sketches that formed undetachable frames for the sayings of Lord [In der Spruchsammlung lässt sich nicht Alles unterbringen, was Mt und Lc von Redegehalt über Mc hinaus darbieten; ihre Elemente sind zuweilen von Lc noch mehr überarbeitet als von Mt; sie enthielt möglicher Weise auch skizzenhafte Erzählungen als Umrahmungen davon unabtrennbarer Herrnssprüche].

2)       Other sources for Matthew are unprovable; but such a question remains open for Luke [Weitere Quellen sind für Mt unerweislich; dagegen bleibt für Lc die Frage nach solchen offen].

3)       Luke knew Mark and Matthew even if he might not have used them [Lc hat neben Mc auch Mt, wenn nicht benutzt, so doch gekannt].

4)     Consequently, at least most of the motives for distinguishing between an Ur-Mark and Mark are eliminated [In Folge dessen kommen wenigstens die Meisten Motive zur Unterscheidung eines Urmarcus von Mc in Wegfall].[5]

Eduard Simons, a student of Holtzmann, argued for this hypothesis more fully  in his  Hat der dritte Evangelist den kanonischen Matthäus benutzt? (Bonn, 1880), but was basically ignored, the direction of his own career didn’t help either, working as he did in practical theology rather than New Testament.

However, Simons’ approach was noted and affirmed by Edward Y. Hincks in the US in the late nineteenth century.[6] Hincks noted the main rival hypotheses of the time including the Ur-Mark theory (H.J. Holtzmann) and Mark-used-Matthew’s-Logia theory (B. Weiss) and rejected them both as problematic and inadequate explanations. Instead, Hincks found Simons’ argument compelling since Matthew’s Gospel circulated widely, Luke was written later so Luke could have had access to Matthew, and the divergences between Matthew and Luke can be accounted for without requiring Luke’s independence of Matthew.[7] Furthermore, for Hincks, Luke’s use of Matthew explains the Luke-Matthew agreements against Mark. He also posits that Mark and the Logia source (Q) may have had independent reports of the same account that were used by Luke and Matthew. This is said to explain the contours of the mission discourses and the accusation of Jesus’ alliance with Satan (i.e., the Mark-Q overlaps).[8]

Robert Gundry has been the most successful advocate for a three source approach although his views have largely gone unnoticed among those outside the field of Synoptic Problem studies. Gundry has written two major commentaries on Matthew and Mark and also explored the presence of so-called Matthean “foreign bodies” in Luke.[9] According to Gundry, both Matthew and Luke used Mark and non-Marcan tradition held in common. While this sounds like the standard two-source theory, Gundry adds:

Because Mattheanisms occasionally appear as foreign bodies in Luke, we also have to think of Luke’s using Matthew as an overlay on his primary sources. That Matthew’s gospel did not provide one of those sources is shown by the disarrangement of Matthean material we would otherwise have to suppose. But that the Matthean foreign bodies come from our present Gospel of Matthew, not from an earlier source, is shown by their conforming to Matthew’s distinctive diction, style, and theology as evident elsewhere and by their frequently depending for their point on Matthew’s context (often stemming from Mark), whereas in Luke they lack contextual point. Therefore we need not free over the numerous minor agreements of Matthew and Luke that do not fall into the category of Mattheanism … They, too, may represent a Matthean overlay in Luke, though apart from the Matthean foreign bodies we would not have known so with any confidence.[10]

Gundry still remains convinced of the existence of Q on the grounds that “Luke’s borrowing wholesale from Matthew” remains problematic because it leads to “Luke’s equally wholesale disarrangement of Matthew’s materials.”[11] The postulation of Q does not prevent Gundry from arguing that Luke used Matthew because it accounts for the presence of a number of Matthean foreign bodies in Luke.

Ron Price has also argued for this position and even has a webpage dedicated to it (see here). See the below diagram:

In sum:

1. I think Marcan priority is, as Mark Goodacre puts it, the “cornerstone” of the Synoptic Problem. So much of the data makes sense on the premise of Marcan priority.

2. I think Q is still a valid postulation given that it accounts for the fluctuating primitivity between Matthew and Luke on several sayings and there is no need to revert to Luke’s wholesale use of Matthew which leaves us wondering why he broke up Matthew’s speeches quite so abruptly and artlessly.

3. But Luke probably used Matthew as an “overlay” or as a Nebenquelle (tertiary source) at a later stage in the production of his Gospels as this accounts for minor agreements (esp. in the passion narrative [like Mark 14:65/Matt 26:67-68/Luke 22:64]) and gives us a further way of explaining the Mark-Q overlaps.

[1] E.P. Sanders, Tendencies of the Synoptic Tradition (SNTSMS 9; Cambridge: CUP, 1969), 279.

[2] H.J. Holtzmann, Die synoptischen Evangelien: Ihr Ursprung und geschichtlicher Charakter (Leipzig: W. Engelmann, 1863).

[3] H.J. Holtzmann, “Zur synoptischen Frage,” Jahrbücher für protestantische Theologie 4 (1878), 533-54.

[4] Some like William Farmer (Synoptic Problem, 20, 40, 47) think that Holtzmann was influenced by his doctoral student Eduard Simons’ dissertation Hat der dritte Evangelist den kanonischen Matthäus benutzt? (Bonn: Carl Georgi, 1880) in coming to this conclusion. Though Edward C. Hobbs (“A Quarter-Century Without ‘Q’,” Perkins School of Theology Journal 33.4 [1980], 11) points out that Holtzmann more probably influenced Simons instead because Holtzmann published such a view two years before Simons’ thesis was submitted and taught Simons for a number of years earlier.

[5] H.J. Holtzmann, Lehrbuch der historisch-kritischen Einleitung in das Neue Testament (Freiburg: Paul Siebeck, 1892), 2:350 trans. with adjustments from Stoldt, History and Criticism of the Marcan Hypothesis, 92.

[6] Edward Y. Hincks, “The Probable Use of the First Gospel by Luke,” JBL 10 (1891), 92-106.

[7] On the Matthew–Luke divergences, Hincks argues (“Use of the First Gospel by Luke,” 96-97): (1) Luke used the Logia source (i.e., Q) differently; (2) In the infancy narrative, Luke shows the influence of Matthew’s genealogy but otherwise follows different sources; (3) Luke uses Mark different to Matthew because Luke esteemed Mark more than Matthew; (4) Luke’s omission of material from Matthew, but if they were in the Logia source where Matthew found them, the problem of omission would still remain; (5) Luke does not use Matthew’s pragmatic way of referencing the Old Testament because of Luke’s independent “doctrinal conception”; and (6) Luke used different sources for his resurrection accounts.

[8] Hincks, “Use of the First gospel by Luke,” 99-101.

[9] Robert H. Gundry, Matthew: A Commentary on His Handbook for a Mixed Church under Persecution (2nd ed.; Grand Rapids, MI: Baker, 1994); idem, Mark: A Commentary on His Apology for the Cross (Grand Rapids, MI: Baker, 1993); idem, “Matthean Foreign Bodies in Agreements of Luke with Matthew Against Mark: Evidence That Luke Used Matthew,” in The Four Gospels 1992, eds. F. van Segbroeck et al (FS F. Neirynck; BETL 100; Leuven: Leuven University Press, 1992), 2:1466-95; idem, “A Rejoinder on Matthean Foreign Bodies in Luke 10,25-28,” ETL 71 (1995), 139-50; idem, “The Refusal of Matthean Foreign Bodies To Be Exercised from Luke 9,22; 10,25-28,” ETL 75 (1999), 104-22.

[10] Gundry, Matthew, 5.

[11] Gundry, Matthew, xvi.


"“In addition, I would aver that inspiration covers a wider suite of human processes that ..."

Plenary Verbal Inspiration and its Problems
"Boettner might not appreciate this, but his analysis of inspiration sounds very like that of ..."

The Case for the Dynamic View ..."
"Honestly, I believe it’s a losing battle.. trump and the Christian evangelical agenda has soured ..."

Response to Christina Rees on the ..."
"Used to attend a church that was way ahead of things, woman 'priest', advocated CND ..."

Response to Christina Rees on the ..."

Browse Our Archives

What Are Your Thoughts?leave a comment
  • Thanks for the enjoyable post, Mike, and especially for the reference to that blog post of mine — I’d completely forgotten that but on re-reading it, I agree with everything I say! 🙂

    On Sanders’s prophecy, it’s worth bearing in mind his own shift in opinion twenty years later, in Studying the Synoptic Gospels, co-authored with Margaret Davies in 1989, where he accepts Goulder’s hypothesis with the important modification that he does think Luke has other sources too, a modification with which I agree and for which I have argued too.

    One of the issues here is the reticence of the scholars of yesteryear to think seriously about issues of memory and oral tradition in the way that they configure the problem. Gundry’s half-way house between Farrer and the 2ST is symptomatic of this — he is thinking in purely literary terms as a means of configuring his solution.

    If you don’t mind my saying so, I think you are falling prey to the same issue by finding Q a solution to issues of “alternating primitivity”. The 2ST projects every variation onto a textual base and does not take seriously what Luke himself tells us, that he was working with both oral and literary sources (Luke 1.1-4).

    The main difficulty, though, with the Holtzmann-Simons-Gundry approach is that once Luke’s knowledge of Matthew is (rightly) conceded, there is no need for Q. The high verbatim agreement between Matthew and Luke means that we are not dealing with later, secondary overlay, but direct copying by Luke of Matthew — Q actually causes problems for making sense of that high verbatim agreement.

    The issue of order is similar — it is still working with a notion of order as dictated by source constraints, which made sense in the 19th and early 20th centuries, but makes less sense now. And as I have often pointed out before, postulating a Q to explain Luke’s order only throws the problem back to Matthew’s order. In other words, the different ordering of the double tradition is a fact; at least one person has done some rearranging.

    Moreover, the remark about Luke breaking up Matthew’s speeches “artlessly” takes us back to the value judgements that I and others have criticized in the past. I suppose that I have a higher opinion of Luke’s art than you, Streeter, Kuemmel and others, but I would repeat that (a) Luke does the same with Mark’s speeches; (b) the value judgement is not shared by contemporary artists; (c) narrative-critical reflection on the alleged artless episodes in fact provides further pause.

    On the Mark-Q overlaps, I don’t think that the Gundry approach helps a great deal. If anything, it tends to confuse the issue. The problem with the so-called Mark-Q overlap material is that it contradicts the assertion that Luke and Matthew never agree in major ways against Mark, something that is used to argue for the independence of Matthew and Luke, and so Q. But since Matthew and Luke do indeed agree in major ways against Mark, one of the key reasons for postulating Q is removed. Having Luke working with Q and Matthew here has no explanatory advantage.

    To illustrate the point rather than leaving it at the abstract level, Mark has “He will baptize you with the Holy Spirit”. Matthew and Luke have “He will baptize you with the Holy Spirit and fire. His winnowing fork is in his hand . . .” We gain nothing by suggesting that Q is finishing Mark’s sentences here if one already has a theory in which Matthew is redacting Mark. No, Matthew is the one finishing Mark’s sentences and he is copied by Luke.

    This comment is long enough to be turned into a blog post, I think!

  • OK, decided to go ahead and revise that comment and turn it into a blog post at

  • Ian Thomason

    G’day, Mike.

    Have you had a chance to read Ward Powers’ “The Progressive Publication of Matthew” yet? It too strides ancient paths, albeit with a slightly different tread.

    Hoo, roo.


    • Ian, I know about, but sadly not convinced by Ward’s case.

      • Ian Thomason

        G’day, Mike.

        I didn’t think you would be 😉 Whilst Ward’s position provides an interesting slant I still tilt towards Markan priority myself.

        Still, I’m rather glad that scholars are looking at the Synoptic Problem in new ways, and not simply tilting to the ‘assured results’ of previous generations.

        Hoo, roo.


  • danzac

    Mike, what do you think about your argument in light of the infancy narrative? The two infancy narratives are so different that I just can’t see how Luke would have been using Matthew. As I did that last chapter in my dissertation, I was less and less inclined to Gundry’s Matthew commentary (he keeps mentioning the changes Luke made to Matthew)

  • John C. Poirier


    I’d like to answer your question for Mike, although I follow Farrer rather than Holtzmann-Gundry. I think that asking *how* Luke could have *used* Matthew’s infancy narrative is an instance of asking the wrong question. The better question — which supporters of the 2DH seldom ask — is “How did it happen that two (supposedly) independent gospel writers *both* decided to add birth narratives to the beginnings of their gospels?” We must keep in mind that if neither Matthew nor Luke had any knowledge of each otehr’s gospel (per the 2DH), then neither of them had ever seen or heard of a gospel with an infancy narrative. We should then marvel at the fact that both gospel writers made the same unprecedented upgrade to the gospel genre. To my mind, that makes it far more likely that Luke knew Matthew. It’s easier to believe that Luke got the idea of an infancy narrative from Matthew than to think that he independently chanced upon the same improvement.

    If that still leaves you wanting an explanation for why Luke didn’t use Matthew’s infancy narrative, I would first say that we should suppose, just as Luke tells us, that Luke didn’t always agree with his sources. (Why write another gospel if Matthew got everything right?) And it should be noted that there are elements in Matthew’s infacy narrative that practically beg to be omitted. For example, how would Luke have reacted to Matthew’s implying that the birth of the Messiah was subject to astrological signs? To my mind, it would be more remarkable if Luke *retained* the story of the magi than to imagine that he was familiar with it but decided not to use it.

    Part of the continuing problem with how people think about the synoptic problem is that they place the burden of proof on anyone positing one evangelist’s *non*-use of another’s gospel, as if we should expect the evangelists to follow their sources slavishly. Yet those who suppose that the writer of the Fourth Gospel knew Mark or Luke apparently don’t find it problematic that he should omit so much from those gospels. Why this double standard?

  • Robert

    “Luke knew Mark and Matthew even if he might not have used them [Lc hat neben Mc auch Mt, wenn nicht benutzt, so doch gekannt].”

    By the way, HJH is only saying Luke may not have used Matthew, not “them” (Mk & Mt).

    The Two Source Theory only indicates what we are able to reliably hypothesize. It in no way denies a more complicated reality behind the hypothesis. I’m convinced of the more complicated reality, but I don’t think we are able to define as reliably exactly what the contours of what that more complicated reality were.