|Alan Garrow Didache|
the problem page
Once Markan Priority is accepted three simple solutions to the Synoptic Problem become available:
1. Luke used Matthew;
2. Matthew used Luke; or
3. Matthew and Luke both independently used Q.
The adoption of Option 1 requires the demonstration of its superiority to Option 2.
The adoption of Option 3 requires the elimination of both Option 1 and Option 2.
Regrettably, almost every attempt to solve the Synoptic Problem over the past 200 years has assumed the non-viability of Option 2 – but without direct investigation.
The recent resurgence of interest in the case for Matthew’s use of Luke (Huggins, Powell, Hengel, MacEwen, and Garrow) has brought this embarrassing omission to light.
This paper asks the long overdue question: Why not Matthew’s use of Luke?
Videos of previous BNTC Seminar Papers are available here:
BNTC 2015 - Making Jesus' Sayings
BNTC 2017 - The Didache: Key to the Acts-Galatians Conundrum
BNTC 2018 - Reflections on the $1,000 Challenge
In 1947 Norwegian explorer, Thor Heyerdahl, attempted to prove the viability of his theory that ancient Peruvian people had traversed vast tracks of ocean to colonise Polynesia. Using only the technology of the time, Heyerdahl and five companions sailed a balsa wood raft for 101 days over 4,300 miles across the Pacific Ocean until they made landfall in French Polynesia. All returned safely.
The Kon-Tiki movie charts the extraordinary challenges faced by Heyerdahl and his crew - but he proved his theory was physically possible. Could the supporters of the various Synoptic Hypotheses do the same?
I put this question to highlight the physical challenges faced by every Synoptic Hypothesis - and, most particularly, those faced by the Farrer Hypothesis. We have no example of anyone using a text in the way that Luke is required to use Matthew under this theory. Using only the technology of the time, could the supporters of this theory demonstrate that Luke's handling of Matthew is physically possible?
This poster serves as an abstract for a paper due to be delivered at this year's British New Testament Conference. The text version is available here.
If you enjoy spot the difference:
In a recent Facebook exchange a well-informed conversation partner made an observation along the lines: isn't it remarkable that supporters of the Farrer Hypothesis, Two Document Hypothesis and Matthew Conflator Hypothesis (FH, 2DH and MCH) can all look at the same data and confidently read it as pointing to their favoured solution to the Synoptic Problem - and not their opponents'? I'd like to try to explain how this happens using the illustration of Brexit.
The case for Brexit is clear. A country wishing to have control over its own destiny cannot afford to be governed by other people in another country, etc. - it's obvious!
The case for the UK remaining in the European Union is clear. Our economic systems are inextricably linked with that of our European partners. To rip away from those systems would be economically disastrous, etc. - it's obvious!
Depending on whether you find the political or the economic argument more compelling, you choose your side in the debate. Having made that decision this then provides the lens through which all further data is viewed. Those who disagree with you seem increasingly blind and obtuse. 'Why can you not see the negative effects of your position!' - we scream as the divide gets wider. At the same time, the weaknesses of our own position go through a process of minimisation; they pale into insignificance beside the monstrous problems with the other side of the argument.
Something similar has happened, I think, in the debate between supporters of the 2DH and the FH. The former are usually open about the fact that their hypothesis doesn't perfectly resolve every bit of data - but they stick with it because they are so convinced that Luke's use of Mathew is impossible. Farrer supporters, on the other hand, are frustrated by the fact that their opponents continue to wave away evidence that, in any other context, would be taken as a clear indication of direct copying between Matthew and Luke.
So long as we are forced to choose between binary options, 2DH or FH, Leave or Remain, there is a tendency to focus on one piece of data, form a position on that basis, and then read all the other data through that lens - constantly highlighting the difficulties of the opposite view while seeking to minimise or explain away the problems in our own back yard.
In the case of Brexit, unfortunately, we can't wind back the clock. If we could, then we might be able to take different decisions to avoid the mess we're in now.
In the case of the Synoptic Problem, however, we do have that luxury. In my article "Streeter's 'Other' Synoptic Solution: The Matthew Conflated Hypothesis" I wind back to some false decisions made by BH Streeter and scholars of his generation - which have been repeated (and thereby consolidated) ever since. Streeter made two critical logical errors. 1) He assumed that arguments against Luke's use of Matthew amounted to arguments against Matthew's use of Luke. 2) He assumed that evidence that Matthew and Luke shared a source or sources (other than Mark) amounted to evidence that Matthew and Luke had no direct contact with one another. Both these assumptions are false. We will never escape the Synoptic Problem's binary bind until we reckon with these simple, unavoidable errors.
When the errors of the Streeter generation are acknowledged, the log-jam eases. A third option, which preserves all the benefits of the 2DH and also the benefits of the FH, comes into view. It is no longer necessary to choose an option that is 'not as bad' as the binary alternative, it is possible to choose an option that resolves all the data while allowing every player, Mark, Luke and Matthew, to behave in a consistent manner. Mark wrote first, Luke incorporated Mark in blocks, Matthew used Mark as a spine and conflated in related material from Luke and other sources (some of which were also known and used by Luke) to create his thematically arranged gospel.
That, then, is my explanation of why three scholars can look at the same data with such variable results. The first two are caught in a binary contest where both sides have significant strengths as well as significant weaknesses. When, however, the basis of that binary set-up is challenged, a third way of looking at things can come into view.
In scholarship we have the freedom to retrace our steps and untangle the mistakes of our predecessors. Most unfortunately, we don't have the same privilege in politics.
Tobias Hägerland's recent article, "Editorial Fatigue and the Existence of Q", NTS (2019), 190-206, makes an important contribution to the debate by calling into question Mark Goodacre's assertion that there are instances where Luke shows fatigue in his handling of Matthew, but none where Matthew shows fatigue in his handing of Luke. Having called into question cases where Luke supposedly shows fatigue in relation to Matthew, Hägerland goes on to conclude:
Against Goodacre's claim that editorial fatigue does not occur in Matthew's handing of the double tradition, I have developed Allison and Davies' suggestion that Matt 3.7-12 provides an example of 'imperfect editing' of Matthew's source material, which in this case cannot be limited to the Markan source. When compared to Luke's handling of the same material, Matthew can be seen to exhibit fatigue. As Matthew's use of Luke remains unlikely on other grounds, it appears most reasonable to postulate Q as Matthew's source here.
Thus, Hägerland claims to have found an occasion when the differences between Matthew and Luke are best explained by Matthew working from Luke but failing to be entirely consistent in his editing of Luke. This might seem like a good time to propose that Matthew used Luke directly. Nevertheless, Hägerland steers away from this conclusion on the basis that: 'Matthew's use of Luke remains unlikely on other grounds'. When these 'other grounds' are taken into account, Hägerland suggests, the only way to explain the data is to propose that Matthew and Luke both used Q.
This line of reasoning would be legitimate if the 'other grounds' for believing that Matthew did not use Luke were either self-evident or well known. Truly, they are not. The assumption that Matthew could not have used Luke was based, in the initial instance, on an appeal to Alternating Primitivity. However, as demonstrated in Video 1, Alternating Primitivity is entirely irrelevant to the question. Since Streeter made the assumption that Matthew's use of Luke may safely be ignored other scholars have followed suit. For example, Robert Stein, in Studying the Synoptic Gospels (2001) assures the reader that Matthew's use of Luke faces 'insurmountable problems' (p. 76) - but the only problem he cites is Alternating Primitivity. Similarly, Chris Tuckett, in Q and the History of Early Christianity (1996) muses: 'For various reasons ... Matthean dependence on Luke is hardly ever advocated, though one sometimes wonders why given the tendency of many to believe that Luke's versions is very often more original.' (p. 4). Thus, Tuckett shows a flicker of awareness, but not enough to break his stride. Paul Foster's objections to Matthew's use of Luke, in, "Is it Possible to Dispense with Q?" NTS (2003) also include Alternating Primitivity. In addition, Foster cites Matthew's omission of Special Luke material. This would only be problem, however, if it could be shown that there is some element of Special Luke that Matthew would have been bound include. Foster's third and final objection is pleasingly ironic. He notes that there are occasions when Matthew appears ignorant of Luke’s additions to Mark. If Matthew were to duplicate Luke's additions to Mark this would create a very real problem for the Two Document Hypothesis. It might even be worth trying to give a name to such a phenomenon. How about 'Minor Agreements'?
A more creative attempt to find a reason for rejecting the case for Matthew's use of Luke was offered by F Gerald Downing in, 'Plausibility, Probability, and Synoptic Hypotheses', ETL (2017). Downing's approach rests on the idea that ancient authors would have scoured their sources to find sequences of thirty letters or more where those sources agreed verbatim and, having done so, they would have felt compelled to include these phrases (no matter how incidental) into their own narrative regardless of the impact on the style, theology, economy of their own creation. Downing finds the fact that Matthew fails to comply with this supposed norm a compelling reason to suggest that he could not have used both Mark and Luke. That Downing resorts to such an extraordinary approach might be considered revealing in itself.
Regular readers of this blog will be familiar with Mark Goodacre's reasons for rejecting Matthew's use of Luke, which are reviewed here, and also in Rob MacEwen's report of Goodacre's paper at SBL Denver in 2018. Perhaps most important among them is his identification of instances where Luke shows fatigue in his handling of Matthew (but not the reverse). This, however, is precisely the argument that Hägerland's article is concerned to undermine.
So, where does this leave us? Worn out in the search for those 'other grounds' or not yet quite fatigued?
I’d like to use this final Calendar entry to express my admiration and gratitude for Mark Goodacre. Mark may not take this as much of a compliment, but he is my kind of scholar. I admire his willingness to challenge the prevailing consensus, I enjoy the ease and clarity of his written prose, and I very much appreciate his willingness to present 'live' rather than reciting a full script. The aspect of his scholarship for which I am most grateful, however, is his willingness to engage. By engaging with the $1,000 Challenge, being willing to respond to my paper at BNTC 2018, and then presenting the paper, ‘Why not Matthew’s use of Luke?’ at SBL Denver 2018, Mark has done more than anyone to alert the wider scholarly community to the fact that an alternative solution to the Synoptic Problem is a topic of current scholarly debate. And, now that scholarly attention is focussed in this direction, Mark Goodacre’s SBL paper offers an important opportunity to learn whether this route is worthy of further attention and exploration.
Mark Goodacre’s paper ‘Why not Matthew’s use of Luke?’ presents such an opportunity because of the calibre of its presenter. Mark has spent the majority of his distinguished scholarly career promoting and defending the Farrer Hypothesis. This means he is superbly equipped to find the worst flaw in the case for Matthew's use of Luke – as well as being highly motivated to do so. In short, if Mark Goodacre cannot find the flaw in the case for Matthew’s use of Luke, then it is doubtful that anyone can. With this in mind it is worth noticing what is missing from the abstract for his SBL paper:
Why Not Matthew's Use of Luke?
It may be that, in the months between writing this abstract and actually delivering the paper, Mark Goodacre will have identified some more substantial reason for rejecting the case for Matthew’s use of Luke. Taking the abstract as it stands, however, note what is missing:
First, there is no suggestion that Mt3rd requires multiple, improbable levels of coincidence – the weakness that most seriously undermines the 2DH.
Second, there is no suggestion that Mt3rd requires Matthew to treat one of his sources in a way that is unique in the period, mechanically/physically exceptionally demanding, and out of tune with the way Matthew treats his other sources – the type of weakness that most seriously undermines the FH.
What remains instead are four phenomena that are all open to more than one interpretation; they are all capable of being explained if Matthew used Luke.
If this is how things continue to stand after SBL session S18-347 (Sunday 18th November, at 4.00pm in Room 302 (Street Level) – Convention Centre (CC)), then a new era of enquiry will have been thrown wide open. And if that does indeed happen, then anyone seeking a solution to the Synoptic Problem more satisfying than those repeatedly discussed to date, will have yet another reason to thank Mark Goodacre.
If you've not seen all the previous entries in the Calendar, I recommend some particular highlights:
Day 4: Richard Bauckham answers the question: How did you come to think that Matthew probably used Luke?
Day 11: Ron Huggins tells the story behind his article 'Matthean Posteriority: a preliminary proposal', NovT (1992)
Day 21: Tim Murray offers his reaction to the Garrow/Goodacre debate at BNTC 2018.
Supporters of the Two Document Hypothesis do not claim that their hypothesis is flawless:
"[E]ven though I have (unashamedly?!) sought in this paper to argue that the weaknesses of the 2DH are possibly less than those of other competing hypotheses today, I hope that I have shown that this theory too is open to questioning. It would be a brave, even foolhardy, person who claimed absolute certainty for the correctness of his/her viewpoint."
“No hypothesis is without its difficulties, and for any of the existing Synoptic hypotheses there are sets of data which the hypothesis does not explain very well”
When pressed to be more specific, 2DH supporters generally point to the problem of the Minor Agreements – places where Matthew and Luke agree against Mark. For example, Gerald Downing, in ‘Plausibility, Probability, and the Synoptic Hypotheses’ ETL (2017) 313-337, devotes considerable attention to attempting to explain the most famous of all the Minor Agreements: Matthew and Luke have: ‘Prophesy … who is it that struck you?’, whereas Mark has only ‘Prophesy!’ (Mark 14.65//Luke 22.65//Matt 26.68).
In my view, however, the Minor Agreements – though certainly problematic for the 2DH – are not its worst weakness. A much more severe problem is posed by those passages where Matthew and Luke agree almost exactly (and where there is no direct parallel in Mark). The problem, in a nutshell, is that ancient writers tended not to copy their sources verbatim. Matthew’s treatment of Mark is unusual in this regard inasmuch as Matthew sometimes copies Mark with a high degree of verbatim faithfulness. Luke’s treatment of Mark is, however, more conventional – he tends to paraphrase, rather than copy Mark word for word. We can see in some detail what happens when one author paraphrases his sources while another is more faithful, by looking at the extent to which Matthew and Luke exactly agree in their treatment of Mark. The answer is that they never come close to both copying Mark in exactly the same way for a whole paragraph. The closest they come is in The Parable of the Fig Tree (Matt 24.32-36//Mark 13.28.32//Luke 21.29.33) where 37% of the 86 words fall within Strings of Verbatim Agreement of 4 words or more. The next closest example is, ‘If any man would come after me …’ (Matt 16.24-28//Mark 8.34-9.1//Luke 9.23-27), where 29% of the 110 words fall within strings of verbatim agreement of 4 words or more. Beyond this there is no other example where more than 20% of a given paragraph falls within Strings of Verbatim Agreement of 4 words or more that are identical in Matthew, Mark and Luke.
Contrast this with the numerous occasions where 96-50% of the words shared by Matthew and Luke fall within Strings of Verbatim Agreement of 4 words or more (e.g. Matt 6.24//Luke 6.13; Matt 3.12//Luke 3.17; Matt 23.37-39//Luke 13.34-35; Matt 3.7-10//Luke 3.7-9; Matt 12.43-45//Lujke 11.24-26; Matt 11.25-27//Luke 10.21-22; Matt 24.45-51//Luke 12.41-46; Matt 6.22-23//Luke 11.34-36; Matt 7.7-11//Luke 11.9-13; Matt 13.33//Luke 13.20-21, and others).
What the 2DH requires, therefore, is that Luke had a particular desire to copy Q exceptionally closely – significantly more closely than he copied Mark. Moreover, this desire has to be mirrored by Matthew – who is required to have had the same attitude to the same passages. Moreover, Matthew and Luke are required coincidentally to have had access to virtually identical copies of Q – even though they were writing in times and places that prevented them from having any direct knowledge one of the other. Q is generally understood to be significantly more ancient than Matthew and Luke and is sometimes attributed to village scribes in rural Galilee. If this were indeed the case, it is necessary to suppose that Matthew and Luke both inherited versions of Q that had been virtually identically adapted (or left pristine) over the decades. Furthermore, it is necessary to suppose that Matthew and Luke both thought that the text of Q they had inherited was, for significant passages, incapable of further improvement (or they both happened to improve it in virtually identical ways). When, in addition, you notice that very high levels of verbatim agreement sometimes occur in the so-called Mark-Q overlap passages (e.g. The Beelzebul Controversy, Matt 12.22-30//Luke 11.14-23), then the credibility of the 2DH is stretched still further. That is to say, even though at least two distinctly different versions of such events were in circulation (as evidenced by Mark 3.22-27 and perhaps also by Matt 9.32-34; John 10.19-21 and Thomas 35) the copies of Q that came down to Luke and Matthew were both equally unaffected (or equally affected) by that environment of diversity.
If there were some really cast iron reason why Luke could not have used Matthew and Matthew could not have used Luke, then, of course, we would no choice but to accept these exceptional levels of coincidence (and the many further examples set out by, for example, Francis Watson in Gospel Writing(Eerdmans, 2013) 117-55). The difficulty is, however, that no-one has yet shown that Matthew could not have used Luke. (NB. As demonstrated in the full Matthew Conflator Hypothesis presentation, the standard arguments against Luke’s use of Matthew do not work in reverse). This means that we are not required to imagine (as supporters of the 2DH must), that Matthew and Luke both happened to inherit virtually identical versions of Q and happened to treat them virtually identically. Instead we can propose that passages where Matthew and Luke agree almost exactly are simply the product of Matthew copying Luke much as he also copied Mark.
For texts and charts relevant to this discussion, see Video 2.
Back in 1996 Mark Goodacre was prepared to allow that the Farrer Hypothesis, as defended by Michael Goulder, includes weaknesses as well as strengths:
Goulder makes ... numerous criticisms of traditional approaches which his opponents will ignore at their peril and, further, his thesis has many strong points. One virtue is the simplicity of his solution to the Synoptic Problem. Another is the explanatory power of his thesis: frequently Goulder sheds new light on famous difficulties.
Mark Goodacre's observation that Goulder's theory, 'demands a degree of complexity in the application which partly undermines its credibility' has been very frequently echoed in subsequent scholarship - albeit more forcefully and with greater attention to the realities of ancient compositional practices. Luke, under the Farrer Hypothesis is required to perform an extraordinarily complex set of operations on Matthew, while treating Mark in a very much more traditional and straightforward manner. All this might be forgivable if it were clearly the only way to avoid the serious problems encountered by the Two Document Hypothesis - but it is not. The difficulties faced by the Two Document Hypothesis and the difficulties faced by the Farrer Hypothesis are both resolved if we allow ourselves to contemplate a relatively simple alternative: Matthew used Luke.
Alan Garrow is Vicar of St Peter's Harrogate and a member of SIIBS at the University of Sheffield.