Luke's Doublets and the Synoptic Problem

Abstract The Synoptic Gospels contain a significant number of so-called doublets, i.e. sayings or narratives which appear twice in one and the same Gospel. Since the nineteenth century these doublets have functioned as a classical argument in favour of the existence of Q. Focusing on treatments of Luke's doublets within the contemporary rivalry between the Farrer hypothesis and the two-document hypothesis, the present article contributes to a not-Q-biased discussion of the evidence. While adherents of the two-document hypothesis should not overestimate the force of doublet-based arguments, defenders of the Farrer hypothesis should pay greater attention to the creation and elimination of doublets as part of Luke's alleged redactional activity.


Introduction
As an 'eternal' question of New Testament research, 1 the Synoptic problem constantly generates numerous publications almost each and every year. Like their forerunners in the nineteenth century and before, present New Testament scholars need to address simultaneously multiple literary phenomena if they wish to create a helpful proposal regarding synoptic relationships or to defend an older one convincingly. Although never boring, 2 at times such attempts may lead to moments of frustration 3 because no hypothesis regarding the Synoptic problem answers all tantalising questions in a satisfactory way. 2 Despite the famous statement of R. E. Brown, An Introduction to the New Testament (ABRL; New York: Doubleday, 1997) 111: 'Although much scholarly attention and even passion has been devoted to this problem, most readers of the NT find the issue complex, irrelevant to their interests and boringa fact that causes me to be succinct in my treatment.' 3 M. F. Bird, The Gospel of the Lord: How the Early Church Wrote the Story of Jesus (Grand Rapids: Eerdmans, 2014) 155, writes: 'Indeed, the Synoptic problem is so complicated and challenging that it can make the most determined researcher want to escape from the frustration somehow. Such respite has eluded every Synoptic researcher apart from the most self-assured.' 4 Regarding the two-document hypothesis, the neo-Griesbach hypothesis and the Farrer hypothesis none other than J. S. Kloppenborg ('Conceptual Stakes in the Synoptic Problem', Gospel Interpretation and the Q-Hypothesis In a certain respect, the synoptic 'maze' bears even a Hydraean characteristic: 5 if a proposed hypothesis allows one to contextualise a particular feature of the synoptic data, two other features remain without an equally convincing explanation. The present article focuses on one such fascinating phenomenon, the so-called 'doublets', 6 and traces its impact on synoptic studies across the sometimes fiercely contested boundaries of different synoptic hypotheses. It does not provide a complete history of scholarship but analyses the most important doublet-based arguments in order to foster a more nuanced and methodologically sound treatment of doublets in synoptic research. Because the assessment of Luke's redactional strategies is one of the most important issues at stake in the debate between advocates of the two-source or two-document hypothesis (still 'the most widely held and accepted solution') 7 and proponents of the Farrer hypothesis ('the current major rival proposal to the two-source theory'), 8 special focus will be laid on the presence of doublets in the Gospel of Luke.

Conjoined Twins? Doublets and the Q-Hypothesis
Christian Gottlob Wilke (1788-1854) 9 and Christian Hermann Weisse (1801-66), 10 in their landmark studies both published in 1838, argued for Markan priority and thus challenged (ed. M. Müller and H. Omerzu; LNTS 2018; London/New York/Oxford/New Delhi/Sydney: T. & T. Clark, 2018) 12-42, at 13) recently declared: 'I am not of the view that definitive arguments for or against any of these hypotheses or variations on these are in fact possible. At best, we can propose compositional scenarios that account for most of the data, most of the time, by what we think of as the "most probable" explanations' (emphasis original). In his contribution to the same volume, S. Alkier ('Sad Sources: Observations from the History of Theology on the Origins and Contours of the Synoptic Problem', Gospel Interpretation and the Q-Hypothesis, 43-77, at 76-7) rightly urges synoptic studies to overcome 'unscholarly confidence' and proposes ten hermeneutical theses for a more balanced and pluralistic approach. 5  the traditional view which regarded the Gospel of Matthew as the oldest of the three Synoptic Gospels 11 but disagreed on the origin of the so-called double-tradition material shared (only) by Matthew and Luke.
Wilke explained the presence of the double-tradition material by Matthew's dependency on Mark and Luke. 12 Although this Matthean posteriority solution never became a major player in the field of synoptic studies, it has recently witnessed some sort of revival thanks to detailed monographs published by Alan Garrow 13 and Robert MacEwen. 14 Weisse, however, attributed the double-tradition material to a second source (independently used by Matthew and Luke alongside Mark) and pointed to 'a new kind of evidence', i.e. the doublets ('Doubletten'). 15 This proposal contained all of the crucial elements of what was later labelled the 'two-document hypothesis'. In a later work, Weisse defined doublets as 'the appearance of one and the same pithy saying at different places in one and the same Gospel' and produced a list of doublets in the Synoptic Gospels. 16 Although many such lists have been compiled since Weisse, 17 no definitive list of doublets exists. 18 Most scholars, however, agree on the approximate numbers of doublets in the Synoptic Gospels (about twenty in Matthew, about ten in Luke, one or two in Mark).
Closely linked to the 'birth' of the two-source hypothesis, the argument from doublets has remained a classical argument for the existence of Q ever since, 19 and has even been called 'the decisive evidence for the existence of a common, written source of Matthew    19 Van Oyen, 'Doublets', 285: 'The phenomenon of sayings doublets has been and continues to be of great importance for the study of the synoptic problem. Again and again, the value of the argument has been repeated by representatives of the two-source hypothesis, and lists of Matthean and Lukan doublets have been published by every generation of scholars'. and Luke'. 20 More recently, however, prominent Q-scholars have articulated hesitations about the force of the argument from doublets. According to Paul Foster, it 'is certainly true that by themselves the doublets are incapable of providing definitive proof of either the existence of Q, or the veracity of the two-source theory'. 21 Christopher Tuckett even calls the argument from doublets 'perhaps one of the weakest arguments for the existence of a Q source'. 22 Maybe this 'radical shift' 23 in the study of doublets is best illustrated by a comparison of two of the most influential books on the two-document hypothesis: while in Paul Wernle's classic Die synoptische Frage from 1899 (credited with giving the two-document hypothesis its final shape) doublets are regarded as 'main argument' ('Hauptargument') for the two-document hypothesis, 24 John Kloppenborg Verbin's Excavating Q from 2000 (probably the most detailed and balanced defence of the twodocument hypothesis ever published) mentions the argument from doublets only in a single footnote and describes it 'as an instructive subset of the argument from order'. 25 Nevertheless, such an important study as Harry T. Fleddermann's reconstruction of and commentary on Q still uses the doublets as one out of four basic arguments for the existence of Q. 26  Tiwald, 29 express similar convictions. Even some scholars who remain unconvinced by the postulate of Luke's and Matthew's complete independence stop short of dispensing with Q precisely because of the doublets. 30 Paul Foster, too, complements his statement quoted above with an assessment clearly in favour of the two-document hypothesis: 'However, on balance, it appears that the two-source theory explains the presence of doublets in the synoptic tradition in the most plausible manner.' 31 Robert Morgenthaler formulates the doublets' role within competing synoptic hypotheses in a much less diplomatic way: 'Any comprehensive and detailed analysis of the phenomenon of doublets is given a wide berth by opponents of the Q-hypothesis (Butler, Farrer, Farmer). They do have good reasons for doing so.' 32

Doublets and the Farrer Hypothesis
Morgenthaler's statement is provocative indeed. But is it true? Obviously, the three scholars mentioned by Morgenthaler (Butler, Farrer, Farmer) represent very different synoptic hypotheses which were received very differently within most recent research.
Basil Christopher Butler's defence of the so-called Augustinian hypothesis (Mark used Matthew, Luke used both of them) did not convince many New Testament scholars, 33 and even the much more influential revival of the Griesbach hypothesis (Luke used Matthew while Mark used both Matthew and Luke) by William Farmer and his followers has clearly passed its prime. 34 Therefore only one rival of the Q hypothesisthe solution proposed by Austin Farrer 35seems to be alive and well. 36 However, doublets have received surprisingly little attention in the camp of the Farrer hypothesis. In his ground-breaking The Case Against Q, Mark Goodacre mentions the argument from doublets only in a footnote and states that it 'does not appear to be used as an argument for Q in any of the recent literature'. 37 This non-discussion seems to be echoed 29 34 According to Foster ('Farrer Hypothesis', 86), 'support for this theory appears to have quite literally "died out"'. 35 The hypothesis that explains the double-tradition material not by the assumption of a source independently used by Matthew and Luke alongside Mark ('Q'), but by Luke's use of Mark and Matthew is widely known as the This remark perfectly illustrates how the phenomenon of doublets is to be explained in the perspective of the Farrer hypothesis in its pure form: all of Matthew's doublets are the product of 'a rather didactic writer', 43 while the doublets in Luke witness the use of two sourcesbut not of Mark and Q, as the two source hypothesis would have it, but of Mark and Matthew. Both of these assumptions have been challenged by defenders of the Q hypothesis and it is worth taking a closer look at two arguments against the Farrer hypothesis that make explicit use of the doublets.

Luke's 'Elimination' of Doublets as a Challenge to the Farrer Hypothesis
Michael Bird's study on gospel origins provides us with a first illustrative example. Bird challenges the Farrer hypothesis by stating that 'if Luke used Matthew, then we are wondering why Luke kept four of these Matthean doublets, eliminated five by dropping the Markan version, and then created five more by augmenting units he had inherited from Matthew and Mark', 44 and judging that 'Luke's dual elimination and formulation of doublets' is not the most convincing explanation of the data. 45 This argument may be countered from the perspective of the Farrer hypothesis in a twofold way. To begin with, the surprisingly low number of doublets in Bird's argument only nine doublets in Matthew compared to twenty-two in Hawkins' listreveals that Bird deals only with those doublets which constitute of a 'Markan half' and a 'double tradition half' but completely ignores the bigger part of Matthew's doublets that seem to contain material from one and the same tradition. In Matthew, we find doublets of Markan material (e.g. Mark 1.15 in Matt 4.17b, but also in Matt 3.2b), of double-tradition material (e.g. Matt 3.7b parallels Luke 3.7b but reappears in Matt 23.33; Matt 3.10b parallels Luke 3.9b but reappears in Matt 7.19) and even of Sondergut material (e.g. Matt 9.13a // 12.7a; Matt 16.19 // 18.18). In other words, Bird's argument rests on a flawed set of data (i.e. only a small part of the evidence) and betrays a logical erroronly doublets which are identified as source doublets from the outset are used in support of a certain source hypothesis. However, this petitio principii is not uncommon in discussions of the Synoptic problem: it is visible in Joel Marcus' case against the Griesbach hypothesis, 46 it is the formative principle behind Harry T. Fleddermann's lists of doublets, 47 and it is evident in Paul Foster's argument that 'the doublets, especially those examples that have been labelled as source doublets (and in particular "the double doublets"), are an important aspect of the cumulative evidence in favour of the two-source theory'. 48 On a methodological level this argumentative fuzziness points to a crucial aspect of any discussion of the phenomenon of doublets: in order to clarify the synoptic relationships, it is extremely desirable to distinguish 'source doublets' from 'redactional doublets'but it is very difficult to draw this distinction without implicitly presupposing a certain solution to the Synoptic problem.
Yet another problem of Bird's argument emerges if one takes a closer look at Luke's redactional activity and especially at those five doublets which Farrer's Luke, according to Bird, 'eliminated … by dropping the Markan version': To sum up, Bird's argument, that 'Luke's dual elimination and formulation of doublets' seems difficult to explain on the Farrer hypothesis, raises two objections. First, Bird treats only a sub-group of doublets (so-called 'source doublets'), which leads to circular reasoning. Second, even if Bird's starting point is accepted, Luke's elimination of some of Matthew's doublets is not that awkward once Luke's redaction of Mark is taken into account.

Luke's Reception of Matthew's Non-Markan Doublet Halves as a Challenge to the Farrer Hypothesis
No less than four decades before Bird's study an even more extensive argument against the Farrer hypothesis (and its explanation of doublets in Matthew and Luke) was developed by Morgenthaler in his already quoted Statistische Synopse. After a detailed discussion of every single doublet in the Synoptic Gospels (twenty-eight pages!) and the creation of three (although not identical!) lists of doublets, 53 Morgenthaler presents a series of objections against the Farrer hypothesis based on the phenomenon of doublets.
Like Goulder on the other side of the debate, Morgenthaler starts with the observation that most of Matthew's doublets appear to be products of Matthew's 'redactional' duplication. In contrast to Goulder, however, who assumed it to be 'perfectly believable that what he [i.e. Matthew] has done eight times with total repetitions, and seven times with partial ones, is an indication of his general manner in writing the rest', 54 Morgenthaler insists that eight of Matthew's doublets do not fit into this scheme that easily (the 'Markan half' is given first): To take one example, 'The Cross Saying' appears (1) in Matt 16.24b (εἴ τις θέλει ὀπίσω μου ἐλθεῖν, ἀπαρνησάσθω ἑαυτὸν καὶ ἀράτω τὸν σταυρὸν αὐτοῦ καὶ ἀκολουθείτω μοι) in a version almost identical with Mark 8.34 (εἴ τις θέλει ὀπίσω μου ἀκολουθεῖν, ἀπαρνησάσθω ἑαυτὸν καὶ ἀράτω τὸν σταυρὸν αὐτοῦ καὶ ἀκολουθείτω μοι). Moreover, it can be found (2) in a slightly different version in Matt 10.38 (καὶ ὃς οὐ λαμβάνει τὸν σταυρὸν αὐτοῦ καὶ ἀκολουθεῖ ὀπίσω μου, οὐκ ἔστιν μου ἄξιος). For this non-Markan version of the saying, Luke 14.27 offers (3) a parallel (ὅστις οὐ βαστάζει τὸν σταυρὸν ἑαυτοῦ καὶ ἔρχεται ὀπίσω μου, οὐ δύναται εἶναί μου μαθητής). 55 Regarding the second characteristic ('certain degree of difference' between the doublet halves) Morgenthaler notes that none of Matthew's doublets with high verbatim agreement between both halves can be found in Luke and calls it 'improbable' that Luke, on the Farrer hypothesis, would have eliminated every single one of these doublets with high verbatim agreement. 56 As for the third characteristic ('non-Markan half with parallel in Luke'), Morgenthaler emphasises that these non-Markan halves 'never' appear in the same context as in Matthew, that all of them can be found within the great insertion Luke 9.51-18.14 (with the sole exception of Luke 19.26) and that all of them fit snugly within their immediate contexts. Taken together, these observations form an impressive pattern leading to the conclusion that the non-Markan halves of these eight doublets stem from a common source ('Q') 57 used by Matthew and Luke besides Mark. According to Morgenthaler, 'no other explanation makes sense'. 58 However, Morgenthaler's arguments require careful scrutiny. To begin with, his observation that Matthew's doublets with high verbatim agreement are missing in Luke while eight Matthean doublets with more different halves are included into the third Gospel gives the impression of a compelling argument indeed: how should it be explained that Luke consistently avoided a certain category of Matthean doublets but just as consistently took over another category of doublets? Did Luke check the tradition-historical background of Matthew's doublets to separate them like Cinderella's lentils? Of course, this would be absurd. But Morgenthaler's own assessment of Luke's attitude towards doublets unwillingly points to an alternative solution: 'No, Luke is not afraid of doublets. He rather wants to be a witness. He is avoiding the twofold incorporation of one and the same incident or the duplication of one and the same element of tradition. If, however, he found a logion both in Q and in Mark, yet in different context and in different form, he was happy to include both versions of it.' 59 If one simply changes 'Q' to 'Matthew', Morgenthaler's statement leads to a plausible Farrer-based explanation of the evidence.
Similarly, the argument from the placement of non-Markan doublet halves in Luke loses some of its force if one studies the individual texts carefully. For Morgenthaler, the appearance of all of Luke's non-Markan doublet halves (except Luke 19.26) in Luke 9.51-18.14 is 'the most strange piece of evidence'. 60 But it can hardly be called 'strange' to find certain elements of the double tradition precisely at that place where most of the double tradition in Luke's Gospel appears. On the Farrer hypothesis, Luke makes use of non-Markan material he finds in Matthew's discourses to build up his central section. One can certainly challenge this concept in generalbut it is not surprising at all that Farrer's Luke moves Matthew's non-Markan doublet halves to Luke 9.51-18.14.
For the very same reason, Morgenthaler's claim that these non-Markan doublet halves 'never' appear in the same context as in Matthew does not lead to an independent argument against Luke's redaction of Matthew: in most cases the 'Matthean' context simply does not exist in Luke's framework. This is most evident with regard to the four doublet halves in Matt 10.33, 38, 39, 40 (see items (b)-(e) in the list above). Matthew's mission discourse (Matt 10) has no equivalent in Luke, who, on the Farrer hypothesis, integrated Mark's mission discourse in Luke 9 but used some of Matthew's additions to build his second mission discourse in Luke 10 and moved yet other of Matthew's additions to other places of his Gospel.
The situation is similar in Matt 5.32 (see item (f)) and Matt 20.16 (see item (g))neither the Sermon on the Mount nor the parable of the workers in the vineyard (note the inclusion Matt 19.30; 20.16) is part of Luke's Gospel.
In Matt 17.20 (see item (h)) things are a bit more complicated: Matthew answers the disciples' question about their incompetence as exorcists (Matt 17.19: 'Why where we not able to cast him out?'; cf. Mark 9.28) with the saying 'On Faith'. If the disciples had faith ὡς κόκκον σινάπεως, the Matthean Jesus explains, they would be able to dislocate mountains and surely every demon would listen to them (Matt 17.20: οὐδὲν Luke 19.26 (see item (a)) is definitely essential within Luke 19.11-27but this is a text that was heavily formed by Lukan redaction also in the perspective of Q-scholars. Thus the claim that non-Markan doublet halves of Matthew's doublets appear in Luke as 'encapsulated' ('eingemauert(e)') 68 elements of their surrounding pericopae remains unconvincing too.
Summing up the discussion of our second example of a doublet-based case against the Farrer hypothesis, it should be emphasised that Morgenthaler's Statistische Synopse is packed with valuable philological observations on the Synoptic problem in general and on the doublets in particular. It remains an indispensable tool for the student of synoptic relationships and is rightly referred to in every recent publication on doublets in Matthew and Luke. Exactly because of the lasting influence of Morgenthaler's monograph, however, it is very important to recognise that his attempt to create a decisive argument against the Farrer hypothesis based on the phenomenon of doublets is hardly convincing.

Conclusion
For obvious reasons, the history of research on doublets in the Synoptic Gospels is closely tied to the emergence of the two-document hypothesis. From Weisse onwards, a long tradition of scholarship has used the doublets as a classical argument for the existence of Q, culminating in the judgement of a very influential introduction to the New Testament that doublets provide 'the decisive evidence for the existence of a common, written source of Matthew and Luke'. 69 Present scholarship on Q, however, is divided on the issue: while some scholars (e.g. Ebner, Fleddermann, Foster, Schnelle, Tiwald) still consider the doublets a strong or even decisive argument for the existence of Q, others (e.g. Kloppenborg, Tuckett) are more hesitant on this point.
Since according the Farrer hypothesis Luke made use of Matthew and Mark, proponents of this solution to the Synoptic problem have no need of 'theoretical contortions' 70 to account for Luke's different list of doublets compared to Matthew's. On the one hand, Luke's redaction of Mark naturally leads to the elimination of some of Matthew's doublets, while, on the other hand, Farrer's Luke treats doublet material in just the same way as he treats double-tradition material in general. Therefore the argument from doublets is not to be considered an independent argument in favour of Q and against the Farrer hypothesis but more (1) 'an instructive subset of the argument from order' 71 and (2) an easily understandable illustration of Matthew's and Luke's redactional activity on the assumption of Q.
The present author aims not to ally with one or the other side of the debate. Just as the two-document hypothesis needs to make Matthew's redaction of Q plausible, 72 the crucial challenge for the hypothesis of Lukan posteriority is to explain Luke's redaction of Matthew in a convincing way. It is important, however, to understand that doublets are a part of that very challenge and not an independent proof that Luke's second source besides Mark could not have been Matthew. In any case, the doublets are worthy of studying more thoroughly by adherents of the Farrer hypothesis given that proponents of Q keep using them as an important argument in their favour.
These methodological considerations already allude to what is maybe the most important task for further research on the doublets. Because of its close connection to the development of the Q-hypothesis, research on doublets is partly Q-biased both in its terminology and its argumentation (for examples, see § 2.2 above), which sometimes leads to circular reasoning. It goes without saying that scholars should avoid terminology that presupposes a certain solution to the Synoptic problem (e.g. 'Q half' of a doublet) while they are arguing which solution is the right one. Furthermore, any discussion of the doublets should take into account all of the evidence; starting from reduced lists of doublets that contain only supposed 'source' doublets may lead to flawed results. In particular, it should never be forgotten that even on the Q-hypothesis most doublets in Matthew and Luke do not result from overlap of two sources (see the lists of Hawkins, Morgenthaler). Finally, the common distinction between 'source doublets' (both halves are taken over from different sources) and 'redactional doublets' (one half of the doublet is the evangelist's reduplication of the other half) is a desirable analytical tool. Drawing this distinction without implicitly presupposing a certain solution to the Synoptic problem remains, however, a daunting task.
Matthew's and Luke's rather different usage of doublets was evident throughout this article. This suggests that doublets illustrate different redactional interests and even different theological agendas. 73 Exploring this connection is beyond the scope of this article with its focus on source-critical considerations but is a much needed endeavour indeed. Foster correctly states that 'the doublets repay close attention since they cast light not only on source-critical questions, but perhaps more importantly on the compositional practices of the evangelists'. 74 Seen from this perspective, Fleddermann's imperative is certainly to be followed: 'No one should underestimate the importance of the doublets.' 75