1 ‘Narrative criticism’ does not characteristically claim that synoptic interrelations should not be investigated at all—only that source-critical investigation should not be made foundational for all other forms of gospel scholarship.

2 Augustine is usually credited with the view that Mark was dependent on Matthew as his ‘follower and summarizer’ (pedisequus et breviator)—the so-called ‘Augustinian hypothesis’. In fact, Augustine changed his mind in the course of writing De Consensu Evangelistarum, concluding that Mark was more probably dependent on both Matthew and Luke, and thereby anticipating the so-called ‘Griesbach hypothesis’ (see de cons. evang. i.2.4; iv.10.11). More significant than either theory is the fact that Augustine advocated literary dependence at all, in opposition to the dominant tradition of independent authorship. John Kloppenborg's suggestion that Augustine does not actually envisage literary dependence is unlikely, given the use of breviator and the vacillation between one theory and another; see his Excavating Q: The History and Setting of the Sayings Gospel (Minneapolis: Fortress, 2000) 38n.

3 Thus Lindemann, A. writes: ‘[E]s bleibt die Frage, ob eine umfassende literarische Analyse und theologische Auslegung der Logienquelle, die der Analyse und Interpretation der synoptischen Evangelien vergleichbar wäre, wirklich möglich ist’ (‘Die Logienquelle Q: Fragen an eine gut begründete Hypothese’, The Sayings Source Q and the Historical Jesus [ed. Lindemann, A; Leuven: Leuven University, 2001] 3–26, here 26)Google Scholar.

4 According to John Kloppenborg, ‘[n]o volume of support for a hypothesis will ever turn it into a fact’, for ‘hypotheses are our ways of configuring and accounting for data…’ (Excavating Q, 3; italics original). In practice Kloppenborg usually treats Q as though it were a fact. If the arguments for the Q hypothesis are as strong as he believes them to be, he is not wrong to do so.

5 Streeter, B. H., The Four Gospels: A Study of Origins (London: Macmillan, 1930) 183–4Google Scholar; Kümmel, W. G., Introduction to the New Testament (London: SCM, 1975) 63–4Google Scholar. Streeter devotes five paragraphs to the independence of Matthew and Luke and the consequent existence of Q; Kümmel, one. Tuckett, Christopher bases his case for the existence of Q on four ‘traditional arguments’ which do not go significantly beyond Streeter (Q and the History of Earliest Christianity: Studies on Q [Edinburgh: T. & T. Clark, 1999] 7–11)Google Scholar. Davies, W. D. and Allison, D. C. devote fourteen lines to the argument for Q rather than Luke's use of Matthew, and merely repeat Streeter's familiar claims (The Gospel according to Saint Matthew [ICC; 3 vols.; Edinburgh: T. & T. Clark, 1988–97] 1.116)Google Scholar.

6 The usual nomenclature—the ‘Farrer hypothesis’, the ‘Farrer–Goulder hypothesis’—should be abandoned, for two reasons. First, the L/M possibility is a concern not just of this or that scholar but of anyone seriously wishing to assess Q itself. Second, the case for Luke's use of Matthew is best articulated not by Farrer or Goulder but by Goodacre, Mark, whose book, The Case against Q (Harrisburg: Trinity, 2002)Google Scholar is fundamental to the renewed debate. See also Goodacre, Mark and Perrin, Nicholas, ed., Questioning Q (London: SPCK, 2004)Google Scholar.

7 Kloppenborg, Excavating Q, 30.

8 For a succinct defence of the need for literary explanations, see Kloppenborg, Excavating Q, 18. Kloppenborg appeals to (1) ‘[s]trong verbal agreements…between each pair of Gospels’, and (2) ‘striking agreements in the sequence of pericopae’, especially significant given that ‘the relative ordering of most of the Synoptic pericopae is not intrinsically determined by their content’.

9 I hope to provide a fuller discussion of these matters in Chapters 3 and 4 of a book provisionally entitled, Receiving Jesus: Gospel Writing in Canonical Perspective, scheduled for publication in 2011.

10 Robinson, James M., Hoffmann, Paul and Kloppenborg, John S., The Critical Edition of Q (Minneapolis: Fortress; Leuven: Peeters, 2000)Google Scholar.

11 The Critical Edition (pp. 6–7) assigns the following items to Q: Q 3.1a ἐν δέ (?), 3.2b {Ἰωάννη}, {ἐν τῇ ἐρήμῳ} (?), 3a πᾶσα..ἡ.. πɛρί{χωρ}ο… τοῦ Ἰορδάνου, 3b {κηρύσσων} (?), 3.4 the Isaiah citation (?).

12 According to the Critical Edition, the following items may be assigned to Q from the baptism accounts: Q3.[[21]] {Ἰησου}, {βαπτισθɛ}, νɛῳχθη, {ο}, {οὐρανο} (p. 18). Q3.[[22]] {καὶ}, {καταβ…ν} (?), {τό πνɛῦμα}, το (?), {ὡς πɛριστɛράν} (?), ἐπ' {αὐτόν}, {καὶ ϕωνή} (?), [(ουραν)] (?), {ὁ υἱός μου ὁ ἀγαπητός, ἐν} (?), {ɛὐδόκησα} (?). These items are sufficient to ensure a connection between the messianic preaching (Q 3.16b–17) and the temptations (Q 4.1–13). The Critical Edition also repeatedly suggests that phraseology apparently drawn from Mark or from Matthean or Lukan redaction may actually reflect identical wording in Q.

13 So Lindemann, for whom no coherent opening for Q can be extracted from Matt 3–4 = Luke 3–4; Q opened instead with the beatitudes (‘Die Logienquelle Q’, 3–26).

14 See Lambrecht, Jan ‘John the Baptist and Jesus in Mark 1.1–15: Markan Redaction of Q?’, NTS 38 (1992) 357–84CrossRefGoogle Scholar.

15 Streeter, The Four Gospels, 183.

16 Perhaps Luke felt that the Baptist's initial reluctance to baptize Jesus was incompatible with the predestined relationship between them, as portrayed in Luke 1? Or perhaps he failed to understand Jesus’ cryptic reference to ‘fulfilling all righteousness’? Or perhaps the Matthean passage simply seemed superfluous to him? The possibilities could be multiplied.

17 Contrast David Catchpole's discussion of the opening narrative sequence, where the existence of Q is made to hang on the unlikelihood that Lukan wording is dependent on Matthean (The Quest for Q [Edinburgh: T. & T. Clark, 1993] 7–16). Why, for example, would Luke omit the Matthean ɛἰς μɛτάνοιαν from the double baptism saying, given that he introduces this very phrase into his rendering of Mark 2.17 in Luke 5.32 (10)? It is assumed here that a writer who introduces a phrase into one context would be ‘inconsistent’ if he failed to reproduce it in another, where it appears in one though not the other of his sources. But that is to operate with a mechanistic model of ‘consistency’. Ad hoc arguments of this kind do nothing to demonstrate the likelihood of Q, in the absence of an overarching reconstruction of the redational procedure entailed in the alternative hypotheses.

18 Kümmel, Introduction to the New Testament, 64.

19 For the ancient use of notebooks (codices) and wax tablets for preparatory work, see Alexander, Loveday, ‘Ancient Book Production’, The Gospels for All Christians (ed. Bauckham, R.; Grand Rapids, MI: Eerdmans: 1998) 71–105 (82–4)Google Scholar.

20 Compare the analysis of Mark A. Matson, ‘Luke's Rewriting of the Sermon on the Mount’, Questioning Q (ed. Goodacre and Perrin) 43–70 (50–61).

21 It is true that, ‘[i]f Luke were proved to have used Matthew, who in turn used Mark, the origin of the non-Marcan material in Matthew would still need clarification’ (Catchpole, Quest for Q, 2). Yet, if one postulated a pre-Matthean sayings source on that basis, it would be utterly misleading to call it ‘Q’.

22 Goodacre, Case against Q, 61.

23 See the detailed discussion of this point in Goodacre, Case against Q, 133–51.

24 The pressure to identify one version of a saying as ‘more primitive’ than another reflects the assumption that the favoured version may approximate to the very words Jesus uttered. For a theoretically sophisticated critique of this view of ‘authenticity’, see Jens Schröter, ‘Die Frage nach dem historischen Jesus und der Charakter historischer Erkenntnis’, Sayings Source Q (ed. A Lindemann) 207–54. According to Schröter: ‘[D]ie Quellen der Vergangenheit enthalten nicht die Tatsachen und Ereignisse, sondern Deutungen von diesen… Die Vorstellung, es könne einen Zugang zu einer hinter diesen Interpretationen liegenden Wirklichkeit geben, wird damit grundsätzlich obsolet’ (229). Thus, ‘[d]ie Vorstellung des “wirklichen” Jesus hinter den Quellen erweist sich dabei als obsolet, die Jesusfrage ist mithin umzuformulieren in diejenige nach einem an die Quellen gebundenen Entwurf des erinnerten Jesus als Inhalt des sozialen Gedächtnisses des Urchristentums’ (233; italics original).

25 ‘It is…a matter of empirical observation that Matthew transposed Marcan passages. From this derives the possibility that in the case of disagreements, Matthew may be secondary in his setting of Q’ (Kloppenborg, J., The Formation of Q: Trajectories in Ancient Wisdom Collections [Harrisburg: Trinity, 2nd ed. 1999] 42–51Google Scholar, 72; see also Excavating Q, 88–91). For a critique of the assumption that Luke keeps closer to Markan sequence than Matthew, see Goodacre, Case against Q, 86–90.

26 The Mission Discourse of Matt 10 is omitted here, since it depends on diverse Markan and non-Markan materials.

27 According to Davies and Allison, in Matt 8–9 the evangelist has reorganized the Markan sequence to create three groups of three miracles stories concluding with a summary and words of Jesus (Matthew, 1.67, 102; 2.1–5 [Excursus V]). In Matt 8–13, five transpositions of Markan sequence may be identified (1.100–101). The analysis is greatly simplified once it is seen that Matthew here follows two distinct Markan sequences. The double sequence is noted by F. Neirynck, who distinguishes here between ‘Mk (A)’ and ‘Mk (B)’ (‘Matthew 4:23–5:2 and the Matthean Composition of 4:23–11:1’, The Interrelations of the Gospels [ed. David L. Dungan; Leuven: Leuven University, 1990] 23–46 [40–1]).

28 Kloppenborg also notes this common sequence, but adds further items to it—the parable of the lost sheep, sayings on forgiveness, the parable of the talents (Formation of Q, 73). The effect is to obscure the sharp divide in Luke between common sequence material (Luke 3.1–11.32 = Matt 3.1–12.50) and non-sequential common material (Luke 11.32–17.37 = Matt passim; exceptions at Luke 6.39–40; 11.2–4, 9–13).

29 In the same way, Matthew supplements the Markan sabbath controversy stories with sayings relating to the temple (Matt 12.5–7), and with the analogy of the sheep falling into a pit (Matt 12.11–12). Matt 12.27, 28 are closely connected to one another and to the context, as is indicated by the sequence ἐν τῷ Βɛɛζɛβούλ (v. 24), ἐν Βɛɛζɛβούλ, ἐν τίνι (v. 27), ἐν πνɛύματι θɛοῦ (v. 28). Matt 12.30, 32 are both variants of Markan sayings.

30 It is therefore not a problem for the L/M hypothesis that ‘there is no sign in Luke of the major addition Matthew makes to Mark at 16:16–19’ (Stanton, Graham, The Four Gospels and Jesus [Oxford: Oxford University, 1989] 88Google Scholar). Luke incorporates many Matthean additions to Mark but not all (e.g. Matt 12.11–12 = Luke 14.5; Matt 16.2–3 = Luke 12.54–56; Matt 17.19–20 = Luke 17.5–6).

31 Against Kloppenborg, the L/M hypothesis does not find it ‘difficult to account for the fact that Luke's placement of the double tradition differs almost entirely from that of Matthew’ (Excavating Q, 30). Once this placement is adequately investigated, its logic is straightforward.

32 This is acknowledged as a possibility in Critical Edition, 244. If Luke 11.27–28 is not drawn from Q, independent redaction has coincidentally attached parallel passages from different sources to ix Return of unclean spirit (Q).

33 The existence of a ‘doublet’ (Matt 9.32–34 = 12.22–24) might be taken as an indication that the evangelist draws on a non-Markan as well as a Markan version of this story. Matt 9.32–34 falls outside both the Markan and the Q sequences; on the other hand, it corresponds closely to the opening of Luke's (single) version of the story (Luke 11.14–16), and could therefore derive from Q. The majority of Matthew's doublets do not easily fit the ‘two source’ model, however, and the duplication is often the work of the evangelist himself. See the examples assembled in Hawkins, Sir John, Horae Synopticae (Oxford: Oxford University, 2nd ed. 1909) 82–99Google Scholar.

34 Neglect of this methodological point may be traced back to the very roots of the Q hypothesis. Q in its modern form originated in Weisse's extension to Luke of Schleiermacher's logia source, which had been derived exclusively from Matthew (Schleiermacher, F. D. E., ‘Ueber die Zeugnisse des Papias von unsern beiden ersten Evangelien’, ThStKr [1832] 735–68Google Scholar; Weisse, Christian Hermann, Die evangelische Geschichte kritisch und philosophisch bearbeitet [2 vols.; Leipzig: Breitkopf & Härtel, 1838] 1.84)Google Scholar. Weisse makes the simple observation that much of the material in Schleiermacher's logia recurs in Luke, who is therefore dependent on the same source as Matthew (1.84). What is at issue is therefore not the existence of the source (which Weisse like Schleiermacher believes is attested by Papias), but rather its scope. Support for the claim that two earlier sources underlie both Matthew and Luke is found in the ‘doublets’ in both (1.82–3). Thus Weisse is led to the conclusion that ‘[n]icht nur Marcus ist beiden gemeinschaftliche Quelle, sondern, unserer bestimmtesten Ueberzeugung nach, auch die Spruchsammlung des Matthäus’ (1.83). It is taken for granted that Luke is ‘völlig unabhängig’ of Matthew (1.56), and the importance of demonstrating this independence is overlooked.