1 B. H. Streeter, The Four Gospels: A Study of Origins. Treating of the Manuscript Tradition, Sources, Authorship, and Dates (London: Macmillan, 1924) 183.

2 The output of the IQP includes J. M. Robinson, P. Hoffmann and J. S. Kloppenborg, The Critical Edition of Q (Leuven: Peeters, 2000), and twelve volumes of the Documenta Q database (Leuven: Peeters, 1996–). This database is projected to include thirty-one volumes – with a combined total of well over 10,000 pages.

3 Streeter, Four Gospels, 183.

4 The assessment of relative primitivity depends on subjective judgement. Such judgements are rarely clear-cut – as the many volumes of the Documenta Q database testify. Taken in overview, however, the proposal that Luke is always more primitive than Matthew, or vice versa, is not convincing. This means that the phenomenon of alternating primitivity requires some form of explanation. The most robust proposal to date is that, as Streeter proposes, Matthew and Luke share access to an earlier source. For another perspective on Alternating Primitivity, see F. Watson, Gospel Writing: A Canonical Perspective (Grand Rapids: Eerdmans, 2014) 162.

5 D. R. Catchpole, The Quest for Q (Edinburgh: T&T Clark, 1993) 6.

6 C. M. Tuckett, Q and the History of Early Christianity: Studies on Q (Edinburgh: T&T Clark, 1996) 10. Similarly, W. D. Davies and D. C. Allison, Matthew 1–7 (ICC; Edinburgh: T&T Clark, 1988) 116; and R. H. Stein, Studying the Synoptic Gospels: Origin and Interpretation (Grand Rapids: Baker Academic, 2001²) 112.

7 Streeter, Four Gospels, 183.

8 Tuckett, Q, 4. The ‘various reasons’ to which Tuckett refers are not articulated. Similarly, Stein, Synoptic Gospels, 76, states that Matthew's use of Luke faces ‘insurmountable problems’. Instead of stating what these are, Stein hedges with, ‘[Matthew's use of Luke] is seldom argued today and will not be discussed at length’ (p. 99). Ultimately, only the (irrelevant) occurrence of Alternating Primitivity is cited before Stein concludes: ‘when all the arguments are considered together, the conclusion seems reasonably certain that Matthew and Luke did not know each other’ (p. 121).

9 M. Goodacre, The Synoptic Problem: A Way Through the Maze (London: T. & T. Clark International, 2001) 108. A helpful digest of fourteen scholars, sensible or otherwise, who explore the case for Matthew's use of Luke is provided by R. K. MacEwen, Matthean Posteriority: An exploration of Matthew's use of Mark and Luke as a solution to the Synoptic Problem (LNTS; London: Bloomsbury, 2015) 7–24. Of particular interest are: West, H. P. Jr, ‘A Primitive Version of Luke in the Composition of Matthew’, NTS 14 (1967–8) 75–95 CrossRefGoogle Scholar; Huggins, R. V., ‘Matthean Posteriority: A Preliminary Proposal’ NT 34 (1992) 1–22 Google Scholar; M. Hengel, The Four Gospels and the One Gospel of Jesus Christ: An Investigation of the Collection and Origin of the Canonical Gospels (Harrisburg: Trinity, 2000); and E. Powell, The Myth of the Lost Gospel (Las Vegas: Symposium, 2006).

10 Hengel, Four Gospels, 171, outlines the essential elements of this arrangement: ‘Certainly the existence of “Q”, whatever is to be understood by that, cannot be ruled out from the start. Even if we can be certain that Matthew as a rule follows Mark and has largely used him, and we conjecture with good reason that he also took over material from Luke, the sum total of his sources remains as unknown to us as the πολλοί in Luke 1.1.’ A similar arrangement is advocated by von Dobschütz, E., ‘Matthäus als Rabbi und Katechet’, ZNW 27 (1928) 338–48CrossRefGoogle Scholar and Aurelius, E., ‘Gottesvolk und Außenseiter: Eine geheime Beziehung Lukas–Matthäus’, NTS 47 (2001) 428–41CrossRefGoogle Scholar.

11 According to Streeter and supporters of the 2DH, Matthew and Luke drew independently on Mark and Q. According to the FH, Matthew used Mark, and then Luke used both Matthew and Mark. Because of the well-established and convincing arguments for Markan priority the Griesbach Hypothesis (GH) is not considered here. Arguments against the FH will, however, also apply to the GH inasmuch as both propose Luke's use of Matthew.

12 MacEwen, Matthean Posteriority, 51, advocates SVA≥4 as a measure preferable to ‘word to word’ agreement because it takes account of word order, as well as word selection.

13 J. S. Kloppenborg, Q the Earliest Gospel: An Introduction to the Original Stories and Sayings of Jesus (Louisville/London: Westminster John Knox, 2008) 51 notes that Matthew and Luke agree verbatim for slightly more than 50 per cent of their Q words.

14 Kloppenborg, J. S., ‘Variation in the Reproduction of the Double Tradition and an Oral Q?’, ETL 83 (2007) 73–4Google Scholar (emphasis original). Kloppenborg makes this point in support of a written, rather than an oral, Q. However, as MacEwen, Matthean Posteriority, 64, notes, this data may prove more than Kloppenborg intends. The puzzle posed by high-agreement passages is also noted, in passing, by Downing, F. G., ‘Redaction Criticism: Josephus' Antiquities and the Synoptic Gospels (ii)’, JSNT 9 (1980) 29–48, esp. 33Google Scholar.

15 This is also the implication of the very detailed statistical study, A. Abakuks, The Synoptic Problem and Statistics (London: Chapman and Hall/CRC, 2014).

16 J. S. Kloppenborg Verbin, Excavating Q: The History and Setting of the Sayings Gospel (Minneapolis: Fortress, 2000) 63 summarises Morgenthaler's data to show agreement of more than 60 per cent for 41 per cent of the Double Tradition, and of less than 19 per cent for a further 8.2 per cent.

17 This is not to suggest an absence of ‘Mid DT passages’. The immediate challenge, however, is to explain the most extreme phenomena. Success here should contribute mechanisms for explaining more intermediate examples.

18 So, Dunn, J. D. G., ‘Altering the Default Setting: Re-envisaging the Early Transmission of the Jesus Tradition’, NTS 49 (2003) 139–75CrossRefGoogle Scholar, esp. 163–5. Also, S. E. Young, Jesus Tradition in the Apostolic Fathers (WUNT2 311; Tübingen: Mohr Siebeck, 2011) 34, 134–9.

19 For example, U. Luz, ‘Sermon on the Mount/Plain: Reconstruction of Qmt and Qlk’, SBL 1983 Seminar Papers (ed. K. H. Richards; SBLASP 22; Chicago: Scholars Press, 1983) 473–9. See also Kloppenborg Verbin, Excavating, 104–11, esp. 109. Cf. also texts listed and discussed in Hengel, Four Gospels, 305–6 n. 671.

20 Synopses 2 and 3, and my accompanying discussion in ‘An Extant Instance of “Q”’, scheduled for publication in NTS 62.3 (July 2016).

21 Streeter, Four Gospels, 183.

22 Tuckett, Q, 4, as quoted above.

23 A concrete example of this phenomenon is provided in the companion article, ‘An Extant Instance of “Q”’ (Synopses 2 and 3, accompanied by n. 45).

24 This point is made with force by R. A. Derrenbacker, Jr, Ancient Compositional Practices and the Synoptic Problem (BETL 186; Leuven: Peeters, 2005).

25 R. A. Derrenbacker, Jr, ‘The “External and Psychological Conditions under Which the Synoptic Gospels were Written”: Ancient Compositional Practices and the Synoptic Problem’, New Studies in the Synoptic Problem: Oxford Conference, April 2008. Essays in Honour of Christopher M. Tuckett (ed. P. Foster, A. Gregory, J. S. Kloppenborg, J. Verheyden; Leuven: Peeters, 2011) 435–58, esp. 437. Cf. also Downing, F. G., ‘Redaction Criticism: Josephus' Antiquities and the Synoptic Gospels (i)’, JSNT 8 (1980) 46–65, esp. 48, 49Google Scholar.

26 Derrenbacker, ‘External Conditions’, 441.

27 Derrenbacker, ‘External Conditions’, 440–1.

28 Derrenbacker, ‘External Conditions’, 440 lists various exemplars alongside Josephus including Diodorus Siculus, Strabo and Arrian of Nicomedia. Downing, ‘Josephus (ii)’, 30 notes that Josephus acts along the lines of accepted convention.

29 Derrenbacker, Ancient Compositional, 213, notes that, ‘Luke follows a procedure much like that of Josephus', cf. also Downing, ‘Josephus (ii)’, 29. On the apparent counter-example of the so-called Lukan transpositions, see Derrenbacker, Ancient Compositional, 214–15.

30 The complexities required of the hypotheses in view are considered below. See also Derrenbacker, ‘External Conditions’, 444.

31 The ‘accident reconstruction’ model for evaluating competing hypotheses is one that tends to favour consistency over inconsistency. That is to say, a reconstruction that allows a given player to act similarly in response to similar situations is generally favoured over one that requires that player to react differently in similar situations.

32 Compelling presentations of the problem of ‘unpicking’ are offered by Downing, F. G., ‘Towards the Rehabilitation of Q’, NTS 11 (1965) 169–81CrossRefGoogle Scholar; Downing, F. G., ‘Compositional Conventions and the Synoptic Problem’, JBL 107 (1988) 69–85 Google Scholar; and Downing, ‘Josephus (i)’ and ‘Josephus (ii)’. The problem of ‘unpicking’ is recognised, but unconvincingly dealt with, by A. M. Farrer, ‘On Dispensing with Q’, Studies in the Gospels: Essays in Memory of R. H. Lightfoot (ed. D. E. Nineham; Oxford: Blackwell, 1957) 55–88, esp. 66–85.

33 Watson, Gospel Writing, 163–216 attempts to show, by contrast, that Luke's use of Matthew is ‘simple and intelligible’ (p. 163). There are, however, three particular problems with Watson's approach. First, there is too little explanation of why Luke's proposed (complex) treatment of Matthew is so very unlike his use of Mark. Second, those who oppose Luke's use of Matthew are characterised as suggesting that Luke's order is ‘chaos’, ‘rubble’ or a ‘confused miscellany’ (pp. 170, 173, 215). This is not the case. What is claimed is that if Luke was using Matthew then his handling of Matthean material is implausibly complex. Third, Watson's conscious decision not to consider Matthew's use of Luke (p. 137) leaves his position vulnerable to the possibility that Matthew's gathering of dispersed Lukan material might be more plausible than the reverse.

34 For a sustained examination of the FH, see Derrenbacker, Ancient Compositional, 190–203.

35 Derrenbacker ‘External Conditions’, 443.

36 Derrenbacker, Ancient Compositional, 253.

37 Derrenbacker, Ancient Compositional, 254.

38 Derrenbacker, Ancient Compositional, 255.

39 Derrenbacker, Ancient Compositional, 213 (emphasis added).

40 The contrast between Luke's treatment of Mark and Q + ‘L’ is downplayed in Derrenbacker, Ancient Compositional, 212–15. However, Derrenbacker, ‘External Conditions’, 444 is more straightforward in noting that, under the 2DH, Luke is sometimes required to act as a conflator.

41 How Luke actually treats sources other than Mark is not, according to the MCH, open to scrutiny. What may be said, however, is that the MCH, unlike the 2DH, does not require Luke to treat his additional sources in a way that is inconsistent with his treatment of Mark.

42 Strictly speaking, it is not necessary to speculate about how Matthew achieved his compositional feats. All that is required is to note that his compositional techniques are consistent, and consistently different from Luke's conventional use of Mark.

43 Diatessaronic studies are very complex, a point illustrated with clarity and candour by W. L. Petersen, Tatian's Diatessaron: Its Creation, Dissemination, Significance, and History in Scholarship (Vigiliae Christianae Supplements 25; Atlanta: SBL, 1994) esp. 442–4. Illustrations derived from the Diatessaron must, therefore, be treated with appropriate caution.

44 Petersen, Tatian, 420, 427 argues that, in addition to the four canonical Gospels, Tatian also worked with Justin's harmony and extra-canonical traditions.

45 Given the difficulties associated with establishing the original language and wording of text, precision is not possible here. However, the broad observation of Petersen, Tatian, 369 is relevant: ‘Large portions of the Diatessaron's original text agreed verbatim with the text now found in the principal gospel manuscripts.’

46 Petersen, Tatian, 26–7 states that ‘the Diatessaron appears to have been a very subtle, word-by-word harmonization’. On pp. 398–401 he discusses the ‘very complex harmonization’ of Matt 28.1–7, Mark 16.1–7, Luke 24.1–9 and John 20.11–12.

47 A. A. Hobson, ‘The Diatessaron of Tatian and the Synoptic Problem’ (PhD Dissertation: University of Chicago, 1904) does not engage with the textual problems associated with the Diatessaron. However, as a broad observation, he is justified in noting that ‘the remoteness of the conflated elements from each other in the written sources, are practically unlimited’ (p. 262).

48 C. H. Roberts and T. C. Skeat, The Birth of the Codex (Oxford: OUP, 1983) 60 suggest that ‘if the first work to be written on a papyrus codex was a Gospel, it is easy to understand that the codex rapidly became the sole format for the Christian scriptures, given the authority that a Gospel would carry’.

49 Identifying Matthew as a consistent codex user does not necessarily imply a second-century date for this Gospel. Roberts and Skeat, The Birth, 45 note that ‘so universal is the Christian use of the codex in the second century that its introduction must date well before ad 100’. Matthew's use of codices does indicate, nonetheless, a late date relative to Luke. Hengel, Four Gospels, 186–205 further supports a late date for Matthew relative to Luke.

50 Streeter, Four Gospels, 183, see also 161.

51 Kloppenborg Verbin, Excavating, 29 (emphasis original).

52 For example, Davies and Allison, Matthew 1–7, 116 and Stein, Synoptic Gospels, 104, 112.

53 Huggins, ‘Matthean Posteriority’, 6 notes just five occasions.

54 This example is treated extensively by Huggins, ‘Matthean Posteriority’, 5–9. Kloppenborg, Excavating, 29 also uses the example of Matthew's and Luke's placement of the makarisms to illustrate his (contrasting) point. His example is flawed, however, in that it requires Matthew's makarisms to begin immediately after Matt 4.22 or 4.23, when in fact they occur after Matt 4.25. Watson, Gospel Writing, 148–54, explores the significance of Luke's Markan context for the Sermon from the point of view of Luke's use of Matthew.

55 Tuckett, Q, 32 (emphasis original). Similarly, Downing, ‘Towards Rehabilitation’, 175, and cf. n. 32 above.

56 Tuckett, Q, 33.

57 Tuckett, Q, 34.

58 This much is admitted by Downing, ‘Josephus (ii)’, 42. Stein, Synoptic Gospels, 126, 142 and Kloppenborg, Earliest Gospel, 36 are similarly candid.

59 Kloppenborg, Earliest Gospel, 36 describes this example as particularly problematic for the 2DH. He concludes: ‘None of the explanations offered for this coincidence is particularly compelling.’

60 The problem is not restricted to these much-discussed examples. E. P. Sanders and M. Davies, Studying the Synoptic Gospels (London: SCM, 1989) 67 note that ‘[t]here are virtually no triple tradition pericopes without such agreements’.

61 Streeter, Four Gospels, 295–331. See also, in still more exhaustive detail, F. Nierynck, The Minor Agreements of Matthew and Luke against Mark (Leuven: University Press, 1974).

62 Stein, Synoptic Gospels, 112 recognises that, for this reason, omissions are a relatively weak argument for suggesting that one text did not use another. He references the advice of Nineham, D. E., ‘Eyewitness Testimony and the Gospel Tradition’, JTS 9 (1958) 247Google Scholar, that ‘arguments from what we should have done to what they [the Evangelists] “must” have done have always to be treated with the greatest caution’.

63 Derrenbacker, ‘External Conditions’, 440: ‘the abbreviation/abridgement of original writings was related to [the] perceived burden that lengthy works placed upon the reader (and presumably the “publisher” and bookseller) … abbreviated works tended to eventually win the favor of the literary public, with the original texts fading into oblivion’. Matthew's frequent removal of detail from Mark's narratives suggests that he did indeed value concision (cf. MacEwen, Matthean Posteriority, 116).

64 For example, West, ‘Primitive Version’, 79–88; Powell, Myth, 50–7; Huggins, ‘Matthean Posteriority’, 14–15; and MacEwen, Matthean Posteriority, 103–17.

65 MacEwen, Matthean Posteriority, 110 cites Davies and Allison, Matthew 1–7, 405 in confirmation of this point.

66 Matthean Posteriority, 186.

67 If Matthew is acting according to the conventions employed by Josephus, then he is unlikely to have invented events without reference to some form of source. Cf. Downing, ‘Josephus (i)’, 55–6.

68 These motives may also explain Matthew's reworking of Luke's Genealogy to make Jesus’ line of descent run through Solomon rather than Nathan (cf. Huggins, ‘Matthean Posteriority’, 19). Downing ‘Josephus (ii)’, 34 indicates that such improvement is not unprecedented, ‘Josephus can vary his own genealogical lists; even if Luke received one, he might have done his own “research” to improve it’. Hengel, Four Gospels, 200–1 sees Matthew's account as being in deliberate opposition to Luke. See also Huggins, ‘Matthean Posteriority’, 17–21.

69 Huggins, ‘Matthean Posteriority’, 17 n. 38 lists all twelve parallels.

70 Tatian also struggled with these conflicting accounts. Petersen, Tatian, 60 writes: ‘According to bar Salibi, it appears that Tatian … despaired of harmonising the discrepancies among the resurrection accounts; this recalls Theodore bar Koni's remark … that Tatian stopped harmonising when he reached the resurrection accounts. This need not mean that Tatian abandoned creating the Diatessaron at that point; rather … that he ceased his harmonising, and presumably switched to presenting the accounts seriatim.’

71 Derrenbacker, ‘External Conditions’, 438: ‘in the case of Arrian, when an author is bringing two sources together, he will follow the accounts of both where they both agree. “But where they differ” Arrian states that he will “select the version [he] regard[s] as more trustworthy (πιστότερα) and also better worth telling (ἀξιαφηγητότερα)”’. See also Downing, ‘Josephus (i)’, 62: ‘If his sources conflict in a fairly straightforward fashion over some major matter, Josephus follows the older and fuller source.’ For Plutarch's similar practice see, Downing, ‘Compositional Conventions’, 81.

72 On the Resurrection Appearances, see also Huggins, ‘Matthean Posteriority’, 21–2.

73 Kloppenborg, Earliest Gospel, 21.

74 A concrete example of Matthew's conflation of Luke with Luke's own source to create a Low DT passage with Alternating Primitivity is provided in the companion article ‘An Extant Instance of “Q”’.

75 ‘Q’, with the addition of quotation marks, indicates any entity (other than Mark) that is shared by both Luke and Matthew. This is as distinct from Q, without quotation marks, used to denote the full reconstruction attempted by the International Q Project (IQP).