The underlined words represent substantial addition to Mark by Matthew, material then paralleled in Luke, refuting the claim that such material never occurs. The same is true in the nearby story of the Temptation of Jesus. Mark's version (Mark 1.12-13) is only two verses long whereas Matthew (Matt. 4.1-11) and Luke (Luke 4.1-13) both have an extended story featuring a major dialogue between Jesus and the Satan with the three famous temptations and rebuttals. Once again, it will seem to the scholar assuming Marcan Priority without Q that the simple Marcan story has been elaborated by Matthew and copied by Luke. Or, to put it another way, Luke has here preferred to use Matthew's substantial modification of the Marcan story. The argument from Luke's lack of Matthew's modifications of Mark seems to be refuted by a simple glance at the Synopsis.
This material is of course familiar to scholars using the argument about Luke's lack of Matthew's additions to Mark, but the force of this evidence, and the fact that it contradicts one of the standard arguments is not felt. The primary reason for this [21] is that examples of this kind are placed in the special category of "Mark-Q overlap", those passages occurring in all three Synoptics in which Mark is not clearly the middle term, which blur the usually more straightforward distinction between "triple tradition" and "double tradition". The fallacy at the base of this argument for Q is that where Luke (on the assumption of Marcan Priority without Q) prefers the Matthean version of a pericope shared with Mark, this automatically goes into the category "Mark-Q overlap". And where Luke prefers the Marcan version of a pericope shared with Matthew, this is held to demonstrate his lack of knowledge of the Matthean versions of Marcan pericopae. I am afraid that Q sceptics are placed in a "no win" situation here. This argument for Q is one that sounds persuasive but which, on closer examination, is based on a fallacy.
(3) "The Q material appears in five large thematic discourses in Matthew but in an artistically inferior arrangement in Luke."
It has long been held that Luke's order constitutes the best evidence for the existence of Q. For while Matthew has Q material as part of his fine, artistic arrangement of Jesus' discourses into five major blocks, Luke has the same material scattered throughout his Gospel. This is held to be uncongenial for the thesis that Luke had read Matthew. R. H. Fuller, for example, writes:
"Matthew has tidily collected the Q material into great blocks. Luke, we must then suppose, has broken up this tidy arrangement and scattered the Q material without rhyme or reason all over his gospel - a case of unscrambling the egg with a vengeance!" [22]
One does not have to look far to find similar statements, praising Matthew's fine arrangement and denigrating Luke's. This is the context in which scholars arguing for Q tend to be particularly disparaging about Luke. "If Matthew is Luke's source," says Styler, "there seems to be no common-sense explanation for his order and procedure." [23] Or, if Luke read Matthew, Stanton says, "he has virtually demolished Matthew's carefully constructed discourses'. [24]
And Fitzmyer asks:
"Why would Luke have wanted to break up Matthew's sermons, especially the Sermon on the Mount, incorporating only a part of it into his Sermon on the Plain and scattering the rest of it in an unconnected form in the loose context of the travel account." [25]
Ultimately, such statements go back to the famous comments of B. H. Streeter, claiming that Luke would have been a "crank" to have acted in the way that a Q sceptical theory would demand, [26]
and of Holtzmann who asked why "Luke should so wantonly have broken up the great structures, and scattered the ruins to the four winds?" [27]
There are two fundamental problems with this pervasive view.
First, the above quotations all focus on Matthew's five famous "blocks" of teaching material, each of which is marked off with the Matthean formula, "When Jesus had finished . . ." (Matt. 7.28; 11.1; 13.53; 19.1; 26.1). But by emphasising the five Matthean blocks, a misleading impression is generated. The casual reader, who does not check the secondary literature by paying careful attention to an open Synopsis, might assume that all of the Q material appears in those five Matthean blocks. Fuller clearly states this - "Matthew has tidily collected the Q material into great blocks" [28]
and so does Stein; [29]
the others, by connecting Q material so closely with the five discourses imply a connection that is stronger than is actually the case.
The fact is that the double tradition material is often located in Matthew outside of his five major discourses, in Chapters 3-4 (John the Baptist; Temptation); 8.5-13 (Centurion's Boy); 8.19-22 (Foxes); 9.37-8 (Labourers); 11.2-27 (Messengers from John; Woes; Thanksgiving); 12.22-45 (Beelzebub; fruit; house swept clean); 22.1-10 (Banquet); and Chapter 23 (Woes against Pharisees). [30]
We should also remind ourselves that large parts of the five major blocks are made up of Marcan and other non-Q material. The third discourse (Matthew 13), for example, has only three verses of Q material (13.16-17, Blessings; 13.33, Leaven; cf. 13.31-2, Mustard Seed). And similarly the fourth (Matthew 18) has only four verses of Q material (18.12-13, Lost Sheep; 18.21-22, Seven Times Seventy). [31]
To put it mildly, the double tradition material is at least as widely distributed in Matthew as it is in Luke. We will do well to be wary of the assumption that Luke, on the Farrer theory, is primarily unravelling Matthew's five discourses when he is reproducing double tradition material. This is a major oversimplification of the evidence, and one which detracts from a proper appreciation of what might have been involved in a reworking of Matthew by Luke.
There is, furthermore, a second and more troubling element in the common perspective on Luke's ordering of the double tradition material. The idea that Luke's ordering of this double tradition material is "artistically inferior" to Matthew's is an important element in the discussion. It is stated most explicitly by Robert Stein:
"The thesis that Luke obtained the Q material from Matthew cannot explain why Luke would have rearranged this material in a totally different and 'artistically inferior' format." [32]
The same viewpoint is, as we have seen, common. Statements of the kind that Luke, if using Matthew, was a "crank", who acts "without rhyme or reason", who "demolishes" Matthew's fine discourses have been a key part of Streeter's legacy. Ultimately this amounts to nothing more than a somewhat dubious value judgement based on outmoded ideas of what Luke was attempting to do in composing his Gospel. Over seventy-five years have passed since Streeter spoke with all the confidence of an era of scholarship in which Luke, like the other evangelists, was essentially a "scissors and paste" editor. But scholarship has changed since then and we should avoid allowing Streeter's legacy to exercise an undue influence in an era in which contemporary literary criticism of Luke has tended to pronounce in favour of his artistic ability. [33]
As appreciation for Luke's literary ability and so for the narrative coherence of his Gospel intensifies, so too it will seem less necessary to appeal to the Q theory to explain the quirks of his order. Streeter's and others' statements, which take for granted a negative value judgement on Luke's order, will become difficult to sustain in the light of contemporary narrative-critical studies of Luke. One of the most important elements to emerge is the observation that Luke's narrative is constructed on the principle of creating a plausible, biographical account in which special attention is paid to movement and sequence. There would be little place in such a narrative for the kind of excessively long monologue that is Matthew's speciality. As Luke Johnson observes, for example, Luke's narrative is "essentially linear, moving the reader from one event to another . . . Instead of inserting great blocks of discourse into the narrative, Luke more subtly interweaves deeds and sayings". [34]
It is straightforward to see this in practice. Where in Matthew the double tradition pericope "Care and Anxiety" (Matt. 6.25-34 // Luke 12.22-34) occurs in the middle of the longest and most famous monologue in his Gospel, the Sermon on the Mount, in Luke it occurs in an excellent and appropriate literary context following on from his unique parable of the Rich Fool (Luke 12.15-21). The parable warns those members of the crowd (who still have possessions, 12.13-14) that life does not consist of the abundance of such possessions - death may strike at any time. The teaching builds on this but develops it, utilizing similar imagery but now changing the audience, exhorting "the disciples" (12.22) not to be anxious about their lack of possessions, something that is a prerequisite for discipleship in Luke (e.g. 5.11, 5.28, 14.33; cf. 6.20). This kind of sensitive narrative arrangement, so typical of Luke, gives some indication of how overstated it is to speak of Luke "demolishing" Matthew's Sermon on the Mount and "scattering the ruins to the four winds". We might more appropriately speak of Luke critically, sensitively and creatively re-working Matthean materials in accordance with his own special narrative agenda. [35]
We should not be surprised about such elements in Luke's order, though, if we decide to take his Preface seriously. After all, Luke appears to be critical of predecessors' attempts to write narratives of the Jesus story (Luke 1.1) and in the same breath stresses not only that he has investigated everything carefully (1.2) but also that he is going to write to Theophilus accurately and in order (1.3). On the Q theory, there is little reason for this overt stress on order, since Luke's order is usually taken to be similar to that of Mark and Q. But on the Farrer theory, the stress is understandable: Luke is making clear that he is critical of his predecessors' work and that his radical re-ordering of Matthew is in the interest of providing Theophilus with the truth of those things in which he has been catechized.
We have allowed ourselves to be guided in this matter by the Q hypothesis for too long. It is no longer acceptable to say that Luke's ordering of double tradition material is "artistically inferior" to Matthew's. In doing so, we only succeed in basing our synoptic theory on a faulty premise compounded by a dubious value judgement informed by assumptions that need no longer (arguably should no longer) be ours.
(4) "The Minor Agreements between Matthew and Luke against Mark are too minor to make the case against Q"
For some time now, Q sceptics have been drawing attention to the Minor Agreements as the Achilles Heel of the Two-Source Theory. [36]
If Matthew and Luke redacted Mark independently of one another, the premise behind the Q theory, then we should not expect to see the number and quality of Minor Agreements that in fact we do see. [37]
In an attempt to counter the force of the Minor Agreements, Q theorists have recently gone on the offensive, claiming that the Minor Agreements are problematic not only for themselves but also for Q sceptics. Tuckett makes the point with his characteristic clarity and force:
"The fact that the Minor Agreements are so minor makes it very hard to believe that Luke has been both influenced positively by Matthew's text in such (substantively) trivial ways, but also totally uninfluenced by any of Matthew's substantive additions to Mark. Undoubtedly the Minor Agreements constitute a problem for the 2ST, but precisely their minor nature constitutes a problem for Goulder's theory as well." [38]
Tuckett's statement illustrates a real difficulty arising from the degree of emphasis that scholars like Michael Goulder have placed on the case from the Minor Agreements. [39]
In the attempt repeatedly to point to a concrete difficulty for the Two-Source Theory, Q sceptics have inadvertently given the impression that the best evidence in their favour is only "minor". But this perspective is, I think, just a trick of the light, and on closer examination it turns out to be based a fallacy. [40]
Part of the fallacy we have seen already in relation to the question of Luke's alleged lack of Matthew's additions to Mark in the triple tradition. On the Farrer theory it is simply not the case that Luke is "totally uninfluenced by any of Matthew's substantive additions to Mark". On the contrary, Luke regularly includes Matthew's substantive additions to Mark, but these tend to get placed into a special category of their own labelled "Mark-Q overlap". The Minor Agreements are thus only part of a broader spectrum and we might properly speak of a sliding scale of Matthean influence on Luke, from pure triple tradition passages which feature Minor Agreements, to Mark-Q overlap passages which feature major agreements between Matthew and Luke against Mark, to double tradition passages where Luke is dependent solely on Matthew. It is the attempt to categorize all of the data in accordance with the demands of the 2ST, Minor Agreements into one class, Mark-Q overlaps into another, double tradition into another, that causes us to miss this.
Tuckett, for example, deals with the different elements of Luke's agreement with Matthew against Mark under these different headings. When writing on the Minor Agreements he asks, "Why should Luke have allowed Matthew's text to influence him in such a minor way as to create these small agreements, but rarely in any major way?" [41]
When talking about the Mark-Q overlaps, however, he rightly describes them as "major agreements" between Matthew and Luke against Mark:
"Indeed the agreements between Matthew and Luke are so extensive (and it is only because they are so extensive that one postulates a parallel, non-Markan version) that these texts are often called 'major agreements'. They really constitute a separate category and cause no real difficulty for the Two-Source theory provided one accepts the possibility in principle of overlapping sources." [42]
Thus, where Luke agrees with Matthew against Mark in minor ways, we tend to call this "Minor Agreement". Where Luke agrees with Matthew against Mark in major ways, we call this "Mark-Q overlap". The pervasive use of this nomenclature, which is derived from the Two-Source Theory, is causing us to miss one of the most interesting features among all the Synoptic data, the fact that there is a continuum, from pure triple tradition to pure double tradition, with varying degrees of agreement along the way, from relatively minor to quite major agreement between Matthew and Luke against Mark. The point, then, is this - the Minor Agreements are indeed troubling to the Two Source Theory but they are only troubling in so far as they constitute one (albeit key) element in a broader spectrum of evidence, all of which is conducive to the Farrer Theory.
(5) "Luke's occasional greater primitivity over against Matthew necessitates the existence of Q"
The argument from "alternating primitivity" is a key element in the standard case for the existence of Q. For if sometimes Matthew and sometimes Luke has the more primitive wording in the double tradition, this might well seem to be a sign that both were dependent on a prior document. As Davies and Allison, for example, make clear:
'Sometimes it is the First Evangelist who seems to preserve the more original form of saying appearing in the double tradition, at other times it is Luke. This is inexplicable if one evangelist is following the other.' [43]
However, there are major difficulties with this apparently straightforward argument. The difficulties can be divided into two categories: (a) problems with the means by which scholars arrive at the conclusion that Luke's material is more original than the Matthean parallels and (b) problems with the assumption that greater primitivity in Luke would necessitate the existence of a Q document.
(a) The conclusion that Luke often witnesses to the more primitive Q wording is based in part on fallacies exposed by Michael Goulder. These fallacies are still not fully appreciated in much literature on the Gospels, so I will attempt to summarise them here:
(i) Self-Contradiction in the Reconstruction of Q: One of the principles in the reconstruction of Q is that language characteristic of Matthew is unlikely to have belonged to the original wording of Q. Arguments of the form "this expression is characteristic of Matthew's redaction, so it is unlikely that it stood in Q" occur regularly in the literature. So where Matthew and Luke have different wording in the same double tradition passage, and where Matthew's wording is particularly characteristic of his writing, it is often concluded that Matthew and not Q is responsible for those Matthean elements. The knock-on result is that the Lucan version, on such occasions, will appear to be more primitive than the Matthean version.
Michael Goulder has pointed to a major problem for this principle by drawing attention to double tradition material in which characteristic Matthean expressions are present in both the Matthean and the Lucan versions, including famously Matthean expressions like o)ligo/pistoi (Matt. 8.26 R; 16.8 R, 14.31 M; Matt. 6.30 // Luke 12.28 QC [44]
); e0kei= e1stai o9 klauqmo\j kai\ o9 brugmo\j tw~n o0do/ntwn (Matt.13.42, 50; 22.13; 24.51; 25.30; Matt. 8.12 // Luke 13.28 QC) and "And when Jesus had finished [these sayings] . . ." (Matt. 11.1; 13.53; 19.1; 26.1; Matt. 7.28 // Luke 7.1). Since such distinctively Matthean expressions occur in this bedrock "QC" material, a major question mark is placed over one of the standard principles used in the reconstruction of Q. Since Q's style is clearly Matthean on these occasions, reconstructing Q elsewhere by eliminating the Matthean elements - the standard practice - becomes problematic. We cannot assume that the Matthean expressions occurring in his versions of double tradition material, but absent in the Lucan parallels, are due to Matthean redaction rather than to Q. In other words, the notion that Luke often features the more original Q wording is based in part on a fallacy in the way that this "original wording" is calculated. If we accept the existence of Q, we know that its style was (at least) sometimes Matthean in nature. This deprives us of one of the main grounds for concluding that Matthean language in Q passages where Matthew and Luke differ is secondary. [45]
(ii) The Lucan Priority Fallacy: There is a related problem. The calculation that Lucan forms of Q sayings are sometimes more original than their Matthean counterparts is also based on a feature of Luke's style. Luke is a subtle and versatile writer with a large vocabulary and a tendency to vary his synonyms. Matthew, on the other hand, has a more pronounced, easily recognisable style, and he does not have so rich a vocabulary. It is consequently much less straightforward to judge Lucan redactional activity than it is to pick out where Matthew has edited sources, and it is correspondingly easy to jump to the conclusion that an apparently 'un-Lucan' form is a 'pre-Lucan', Q form. Again and again one sees in the literature claims that a given word is "un-Lucan and therefore pre-Lucan", claims that artificially reinforce the notion that Luke's version is more primitive than Matthew's. [46]
These fallacies in the calculation of the original wording of Q are widespread and it is arguable that they account, in part at least, for many of the decisions that Q scholars make in favour of the priority of Lucan versions of Q. They do not, of course, account for all such decisions. Sometimes it is simply the case that good arguments for the secondary nature of the Lucan version have been ignored, one of the best examples of which is Luke's "Blessed are the poor" (6.20) over against Matthew's "Blessed are the poor in spirit" (Matt. 5.3). [47]
At other times it may be because the Lucan version indeed exhibits, when all things have been considered, signs of greater primitivity. I do not think that the number of such passages is large, but it is worth looking at what the implications of the presence of such passages would be. Do they necessitate the existence of Q?
(b) Let us say that we were to find a Lucan form of a Q saying that appeared on solid grounds to be more original than its Matthean counterpart. Would this witness to the existence of Q? It is universally assumed that the answer to this question is affirmative, but we need to observe that this too is based on a fallacy. Occasional signs of greater primitivity are only a difficulty for Luke's literary knowledge of Matthew if we are prepared to deny the role of oral tradition in Gospel relationships, [48] if we assume oral traditions of the Jesus story died out as soon as each evangelist committed them to papyrus, or if we think that oral traditions each travelled down their own individual, exclusive pipelines, one for Mark, one for Q, one for L and so on.
The situation, in other words, is like this. If we grant Luke's literary dependence on both Matthew and Mark, it is inherently plausible to imagine this literary dependence interacting with Luke's knowledge of oral traditions of some of the same material. This makes it inevitable that, on occasion, Luke will show knowledge of some more primitive traditions. Take, for example, the Lord's Prayer (Matt. 6.9-13 // Luke 11.2-4), one of the key examples of a supposedly more primitive Lucan form, and, as a liturgical text, the kind of passage that we will expect to have been particularly prone to local variation in oral tradition. Even today, where the Lord's Prayer is often known primarily orally and not in dependence on a written text, one finds local variation. The same kind of thing seems highly likely to have been the case when Luke comes to write his version of the prayer in 11.2-4. He looks at the Matthean version but re-writes it in line with the version more familiar to him from frequent recitation in his own tradition. Just as many Catholics today end the prayer where Matthew ends it, at "Deliver us from evil", without adding "Thine be the kingdom, the glory and the power, for ever and ever Amen", so Luke ends his prayer with "Lead us not into temptation" and not with "But deliver us from evil", in spite of the fact that the latter is present in his text of Matthew. Just as Catholics today know of the existence of the "Thine is the kingdom" clause, but choose not to use it because of familiarity and loyalty to tradition, so too it is hardly difficult to think of Luke knowing the clause "But deliver us from evil" but not using it for the same kinds of reason.
We can see the same principle at work in Luke's use of Mark. We do not always automatically accord priority to the Marcan version of traditions they share. In the case of the Eucharistic tradition, for example (Matt. 26.26-29 // Mark 14.22-25 // Luke 22.15-20), Luke appears to have different, arguably "more original" elements over against Mark - indeed we are lucky enough to have an independent witness to this in Paul's first epistle to the Corinthians (1 Cor. 11.23-26). Given that most of us happy to grant that there are, on occasions, some more primitive elements in Luke's triple tradition passages, not withstanding Luke's agreed literary dependence on Mark, we should not be surprised to see the same thing from time to time in the double tradition material that Luke shares with Matthew.
The difficulty is that scholars have routinely confused issues of literary priority with issues over the relative age of traditions. The theory of Luke's literary dependence on Mark and Matthew does not necessitate the assumption that his material is always and inevitably secondary to Matthew's and Mark's. Just as most of us do not deny the likelihood that Luke creatively and critically interacted with the living stream of oral tradition when he was working with Mark, so too we should not think it odd that he might have interacted with Matthew in the light of his knowledge of similar material in oral tradition. [49]
As usual, we will do well to pay attention to what Luke tells us he is doing in 1.1-4. Does he not imply that he has been engaging carefully not only with those who have undertaken to write narratives (1.1), but also with the things that have been "handed down by those who from the first were eyewitnesses and servants of the word" (1.2)? If we take this seriously, we have a picture of an evangelist who is critically and creatively interacting with both traditions and writings of "the events that have been fulfilled among us". The time has come to distinguish properly between direct literary use of a prior text and knowledge of oral traditions, both of which are key in the composition of Luke's Gospel. Since practically all scholars accept that Luke was familiar with oral traditions of Jesus material, it is quite reasonable to assume that some of these traditions will have overlapped and interacted with his direct knowledge of his written sources. It is quite reasonable, in other words, to dispense with Q.
Conclusion
In spite of the steady increase in uncertainty on the Synoptic Problem in some circles, Q has remained Marcan Priority's favourite companion. But we should not allow Q a monopoly on the affections of Marcan Priority. For while the Priority of Mark has stood the test of time, Q looks less secure. Its essential premise, Matthew's and Luke's independent use of Mark, is based in large part on arguments which at first hearing sound plausible but which on closer inspection turn out to be flawed. It is not acceptable to speak of Luke lacking Matthew's additions to Mark in triple tradition (fallacy 2), nor will it do simply to repeat the dubious value judgement on Luke's order (fallacy 3). The Minor Agreements are only "too minor" to be significant if they are viewed outside of the larger continuum of agreement between Matthew and Luke against Mark (fallacy 4) and the argument from "alternating primitivity" requires both dubious assumptions about the reconstruction of Q and the failure to recognise a distinction between literary priority and the relative age of traditions (fallacy 5). Least acceptable of all is the simple equation of Marcan Priority with the Two-Source Theory, thereby eliminating the challenge from the Farrer Theory (fallacy 1).
Fitzmyer once said that the great advantage of the Q hypothesis was its usefulness, its Brauchbarkeit, and in some ways this is right. [50]
It is an hypothesis that as part of the Two-Source Theory has allowed us the means to get familiar with Marcan Priority, to work with it, to test it and to re-imagine Christian origins with it. In an era of scholarship less sensitive than ours to the literary creativity of Luke, Marcan Priority needed a partner like Q to help establish its plausibility. But its usefulness was only for a season. The more we work with Q in the interpretation of Matthew and Luke, the more we will see its limitations for doing what we now want to do in Gospel studies. We are grateful to Q - it has served us well - but now it is time for it to bow out and to give way to a worthy successor, a theory that allows us scope to appreciate the evangelists' literary achievements while strongly affirming, indeed building on the Priority of Mark.
Notes
1. Mark Goodacre (ed.), Mark Without Q: A Synoptic Problem Web Site, http://www.bham.ac.uk/theology/q (1997-2000). 
2. Austin Farrer, "On Dispensing with Q", D. E. Nineham (ed.), Studies in the Gospels: Essays in Memory of R. H. Lightfoot (Oxford: Blackwell, 1955), pp. 55-88; reproduced at Mark Goodacre (ed.), Mark Without Q, http://www.bham.ac.uk/theology/q/farrer.htm. The so-called "Farrer theory" effectively originated a little earlier - see James Hardy Ropes, The Synoptic Gospels (First Published 1934; Second Impression with New Preface, London: Oxford University Press, 1960), pp. 37, 66-8 and 93-4. For the Oxford connection, see, among others, Michael Goulder, Midrash and Lection in Matthew (London: SPCK, 1974), the Speakers' Lectures in Oxford in 1969-71; John Drury, Tradition and Design in Luke's Gospel (London: Darton, Longman and Todd, 1976), written while Drury was chaplain of Exeter College, Oxford; E. P. Sanders and Margaret Davies, Studying the Synoptic Gospels (Philadelphia: Trinity Press International; London: SCM, 1989), written while Sanders was Dean Ireland Professor at Oxford; and Eric Franklin, Luke: Interpreter of Paul, Critic of Matthew (JSNTSup, 92; Sheffield: Sheffield Academic Press, 1994).
3. Beverley Roberts Gaventa, Mary: Glimpses of the Mother of Jesus (Personalities of the New Testament; Edinburgh: T & T Clark, 1999), p. 24, n. 4. I am aware, of course, that this comment (and many others like it) are simply little statements in passing in footnotes. But that reinforces the point -- what scholars give away in passing can be a reliable guide to common assumptions.
4. Though even within the Two-Source Theory there are exceptions to this, e.g. Robert Gundry, who maintains the Two-Source Theory but thinks that Luke also knew Matthew. See, for example, "Matthean Foreign Bodies in Agreements of Luke with Matthew against Mark. Evidence that Luke used Matthew" in F. Van Segbroeck et al (eds.), The Four Gospels. Fs. F. Neirynck (BETL 100; Leuven: Leuven University Press, 1992), Vol. 2, pp. 1467-95.
5. Defenders of the Farrer theory, most notably Michael Goulder, John Drury and Eric Franklin, tend to take Marcan Priority for granted in their Q scepticism. See, however, my "Fatigue in the Synoptics", New Testament Studies 44 (1998), pp. 45-58, reproduced at Mark Goodacre (ed.), Mark Without Q: A Synoptic Problem Web Site, http://www.bham.ac.uk/theology/q/fatigue.htm, which attempts to combine a fresh argument for the Priority of Mark with a related argument for Luke's use of Matthew. See also my The Synoptic Problem: A Way Through the Maze (Sheffield Academic Press, forthcoming).
6. The authors own preferred name for their hypothesis is "The Two-Gospel Hypothesis", Matthew and Luke "the two Gospels" acting as sources for Luke. However, Michael Goulder has argued that the term is essentially confusing because the Farrer Theory is also, in a way, a "two Gospel hypothesis", with Mark and Matthew the sources for Luke. See further Michael Goulder, "Luke's Knowledge of Matthew" in Georg Strecker (ed.), Minor Agreements: Symposium Göttingen 1991 (Göttingen: Vandenhoeck & Ruprecht, 1993), pp. 143-60 (p. 143, n. 1).
7. The first major publication of this group was Allan J. McNicol with David L. Dungan and David B. Peabody (eds.), Beyond the Q Impasse - Luke's Use of Matthew: A Demonstration by the Research Team of the International Institute for Gospel Studies (Valley Forge PA: Trinity Press International, 1996). For a detailed review and critique, see my "Beyond the Q Impasse or Down a Blind Alley", JSNT 76 (1999), pp. 33-52. See also the review by Christopher Tuckett, JBL 117 (1998), pp. 363-5; also in SBL Review of Biblical Literature (1998), http://www.bookreviews.org/Reviews/1563381842.html.
8. This is not meant pejoratively. It is arguable that the dominance of the Two-Source Theory necessitates careful thinking about how to organise and advertise the alternatives.
9. See, for example, J. Tyson, "The Synoptic Problem" in his The New Testament and Early Christianity (New York; London: Macmillan; Collier Macmillan, 1984), pp. 148-58.
10. This is the remarkable claim of Arland Jacobson, "The Griesbach Hypothesis has succeeded in establishing itself as the only real alternative to the Two Document Hypothesis." (The First Gospel: An Introduction to Q (Foundations & Facets; Sonoma, CA: Polebridge, 1992), pp. 5-6).
11. Stephen Patterson, for example, regards Goulder's views as "more obscure" than Griesbach: Review of C. M. Tuckett, Q and the History of Early Christianity, JBL 117 (1998), pp. 744-6 (p. 744); also at SBL Review of Biblical Literature (1998),
http://www.bookreviews.org/Reviews/156563246X.html.
12. "I consider the assumption that Matthew and Luke used a second source in addition to Mark to be assured, despite all criticism. Neither the priority of Matthew nor proto-Luke hypotheses can resolve the Synoptic Problem to any comparable extent." ("Q: Sayings of Jesus or Logia?" in R. Piper, ed., The Gospel Behind the Gospels: Current Studies on Q (NovTSup, 75; Leiden: E. J. Brill, 1994), pp. 97-116 (p. 97). I am grateful to Michael Goulder for this observation, "Is Q a Juggernaut?", JBL 115 (1996), pp. 667-81, reproduced at Mark Goodacre (ed.), Mark Without Q, http://www.bham.ac.uk/theology/q/goulder.htm (p. 668, n. 11).
13. Helmut Koester, Ancient Christian Gospels: Their History and Development (London: SCM; Philadelphia: Trinity Press International, 1990), p. 130. Cf. Burton L. Mack, The Lost Gospel: The Book of Q and Christian Origins (New York: HarperSanFrancisco, 1993), "Even today there are scholars who continue to resist the two-source theory and to favour Matthew as the earliest gospel" (p. 20). Cf. also Luke Johnson, The Real Jesus (San Francisco: Harper, 1995), p. 52.
14. Craig. L. Blomberg, Jesus and the Gospels: An Introduction and Survey (Leicester: Apollos, 1997), p. 90, n. 27. The same unfortunate mistake is made by one of the key scholars writing from the Griesbachian perspective, David L. Dungan, A History of the Synoptic Problem: The Canon, the Text, the Composition and the Interpretation of the Gospels (Anchor Bible Reference Library; New York: Doubleday, 1999), "the modified Augustinian hypothesis proposed by B. C. Butler and Austin Farrer" (p. 369); again on pp. 376, 378 and 384-5.
15. The point is that there is some widespread ignorance about the Farrer theory. It should go without saying that there are, of course, some honorable exceptions, e.g. John Kloppenborg (ed.), The Shape of Q: Signal Essays on the Sayings Gospel (Minneapolis: Fortress, 1994), Chapter 1 (pp. 2-3); Dale C. Allison, Jr., The Jesus Tradition in Q (Harrisburg, Pennsylvania: TPI, 1997), p. 1; John Kloppenborg Verbin, Excavating Q: The History and Setting of the Sayings Gospel (Minneapolis: Fortress, 2000). Unfortunately, the latter was not published before the writing of this paper.
16. W. D. Davies and Dale C. Allison, A Critical and Exegetical Commentary on the Gospel According to Saint Matthew: The International Critical Commentary on the Holy Scriptures of the Old and New Testaments (Edinburgh: T & T Clark, 1988), p. 116. Davies and Allison are, however, aware of the Farrer theory - see p. 114, n. 66.
17. The Gospel According to Luke I-IX (Anchor Bible 28, New York: Doubleday, 1981), pp. 73-4. Cf. Christopher Tuckett, Q and the History of Early Christianity: Studies on Q (Edinburgh: T & T Clark, 1996), pp. 7-8: "Luke never appears to know any of Matthew's additions to Mark in Markan material. Sometimes, in using Mark, Matthew makes substantial additions to Mark, cf. Matt. 12.5-7; 14.28-31; 16.16-19; 27.19, 24. If Luke knew Matthew, why does he never show any knowledge of Matthew's redaction of Mark? It seems easier to presume that Luke did not know any of these Matthean additions to Mark and hence that he did not know Matthew."
18. The same error is made by Tuckett - see previous note.
19. Austin Farrer, "On Dispensing", p. 57.
20. It is worth adding that if indeed the theory of Marcan Priority is correct, Luke may well have known Mark for quite some time longer than he has known Matthew. In such circumstances, his familiarity with Mark might have made the Matthean additions clear to him. Fitzmyer is particularly keen to stress that "the pattern in these passages is such in the Matthean Gospel that they are best regarded as additions" (Gospel, p. 74, his emphasis). Given that this pattern seems so clear to Fitzmyer and others, we should not rule out the possibility (I would say likelihood) that the pattern was also clear to Luke.
21. Note also that some of the most impressive examples of this feature come, as we have seen, in Luke 3-4, covering material like John the Baptist and the Temptations. This is precisely the material that tends to be allowed an honorary exception to the rule when one is discussing the related phenomenon of Luke's order - Luke never, it is said, places double tradition material in the same triple tradition context as Matthew, except in Chapters 3-4. Although this exception is not specifically invoked in the context of Luke's alleged lack of Matthean additions to Mark, it may be that the same exception is thought to apply here. The Q sceptic's concern is that this is just the kind of rather bulky exception that tests a rule and finds it wanting. Further, much of Luke's agreement with Matthew's additions to Mark actually occurs outside of Luke 3-4.
22. Reginald H. Fuller, The New Testament in Current Study (London: SCM, 1963), p. 87.
23. "Synoptic Problem" in Bruce M. Metzger and Michael D. Coogan (eds.), The Oxford Companion to the Bible (Oxford: Oxford University Press, 1993), pp. 724-7 (p. 726).
24. Graham N. Stanton, "Matthew, Gospel of" in DBI, ad loc.
25. J. A. Fitzmyer, Luke I-IX, p. 74. Cf. also Christopher M. Tuckett, "Synoptic Problem" in ABD, VI, pp. 263-70 (p. 268).
26. B. H. Streeter, The Four Gospels: A Study of Origins (London: Macmillan, 1924), p. 183.
27. H. J. Holtzmann, Die synoptischen Evangelien, Ihr Ursprung und geschichtlicher Charakter (Leipzig: Engelmann, 1863), p. 130.
28. See above and footnote 22.
29. Robert Stein, The Synoptic Problem: An Introduction (Grand Rapids: Baker Books, 1987), p. 95: "Matthew's handling of his source, however, was far more thoroughgoing, for he has arranged this material into five 'books'. The result is a masterful organization whose artistry is well-recognized." Cf. also, "Within Matthew this material is found in five (or six) sections, all of which end with something like 'And when Jesus finished these sayings' . . . These five sections (5-7; 10; 13; 18; 23-25 [or 23; 24-25]) are artistically arranged within Matthew's Gospel . . ." (ibid.).
30. Stein acknowledges that much Q material occurs in Matthew 23 but tends to treat it as part of the fifth discourse (see previous note).
31. One is able to see the distribution of the double tradition material in Matthew and Luke quickly and easily by using Allan Barr, A Diagram of Synoptic Relationships (Second Edition with a New Introduction by James Barr, Edinburgh: T & T Clark, 1995; first published 1938). Also useful is a new internet resource, Alfred Suhl, Schemata und Übersichten über die synoptischen Evangelien (created 1999), http://wwuuser.uni-muenster.de/cit/Evv/synoptiker/Hauptrahmen/SynoptikerHauptframe.htm, which provides colour breakdowns of Matthew and Luke according to the Two-Source Theory.
32. Robert Stein, The Synoptic Problem, p. 95. Cf. also p. 96: "The arrangement of the material in Matthew is extremely well done. The Sermon on the Mount (Matt. 5-7) ranks as one of the greatest works of literature ever written. Why would Luke, who was by no means an inept writer, choose to break up this masterpiece and scatter its material in a far less artistic fashion throughout his Gospel?"
33. Cf. Farrer, "On Dispensing", p. 56. Farrer's observation that "our conception of the way in which the Gospels were composed has gradually altered", and the consequences for the Q hypothesis, is all the more true in the year 2000 than it was in 1955.
34. Luke Johnson, "Luke-Acts, Book of" in Anchor Bible Dictionary IV, pp. 405- 6.
35. I have written at greater length on the potential for fruitful interaction between source- and narrative-criticism of Luke in "Luke, Narrative-Criticism and the Q Hypothesis", as yet unpublished paper given to the SBL Annual Meeting Synoptics Section November 1998.
36. The key recent publication on the Minor Agreements is G. Strecker (ed.), Minor Agreements: Symposium Göttingen 1991 (Göttingen: Vandenhoeck & Ruprecht, 1993). For a good introduction to the problem the Minor Agreements pose for the Two-Source Theory, see E. P. Sanders and M. Davies, Studying the Synoptic Gospels (Philadelphia: TPI; London: SCM, 1989), pp. 67-83. For the role they play in the Farrer theory and especially in Michael Goulder's work, see further my Goulder and the Gospels: An Examination of a New Paradigm (JSNTSup, 133; Sheffield: Sheffield Academic Press, 1996), Chapter 3.
37. It is difficult to judge the precise number, or to assess the significance of the number of Minor Agreements. F. Neirynck, The Minor Agreements of Matthew and Luke against Mark with a Cumulative List (BETL, 37; Leuven: Leuven University Press, 1974) lists over 750, but of course many of these, as Neirynck has maintained in several publications, are straightforwardly attributable to independent redaction by Matthew and Luke. The stronger MAs, like that at Mark 14.65 (ti/j e0stin o9 pai/saj se;), are those that are most troubling to the 2ST.
38. Christopher Tuckett, Q and the History of Early Christianity, p. 28. Tuckett makes a similar point in "The Minor Agreements and Textual Criticism" in G. Strecker (ed.), Minor Agreements, pp. 119-41 (p. 140). See further note 42 below.
39. For Michael Goulder's work on the Minor Agreements, see especially "On Putting Q to the Test", NTS 24 (1978), pp. 218-34; Luke: A New Paradigm (JSNTSup, 20; Sheffield: JSOT Press, 1989); and "Luke's Knowledge of Matthew" in G. Strecker (ed.), Minor Agreements, pp. 143-60. For analysis and comment see my Goulder and the Gospels, Chapter 3.
40. This point develops a comment I made in Goulder and the Gospels, p. 126.
41. ABD VI, p. 267.
42. ibid. These observations should be qualified by drawing attention also to Tuckett's treatment of MAs and Mark-Q Overlaps in The Revival of the Griesbach Hypothesis: An Analysis and Appraisal (SNTSMS 44; Cambridge: Cambridge University Press, 1983), Chapters 7-8. He suggests that one of the conditions for seeing Mark-Q overlap is evidence of another source underlying the Matthean and Lukan version, i.e. the Matthean version will not be explicable solely as Matthean redaction of Mark. The difficulty with this perspective is that one of the key pieces of evidence in favour of another underlying source is the substantial agreement between Matthew and Luke against Mark - one can see this clearly in ABD quotation above. Further, Matthew's use of Mark need not imply that Matthew had no other sources for this material. The point is that Luke is here showing literary agreement with the Matthean version of the triple tradition, regardless of Matthew's own sources in reworking Mark.
43. W. D. Davies and D. C. Allison, Matthew, vol. 1, p. 116. The point is repeated again and again in the literature, e.g. Stein, Synoptic Problem, pp. 96-101; it is perhaps the most common argument for the existence of Q.
44. R = Redactional addition by Matthew on the assumption of Marcan Priority; M = only in Matthew; QC = common to Matthew and Luke.
45. For the original argument on the "Matthean Vocabulary Fallacy", see Michael Goulder, Luke: A New Paradigm, particularly pp. 11-15. For a critique, see my Goulder and the Gospels, Chapter 2, especially pp. 83-5. Goulder modified the argument, partly in response to my critique, in "Self-Contradiction in the IQP", JBL 118 (1999), pp. 506-17. This statement is stronger than the earlier version of the argument, partly because it focuses primarily on telling Matthean expressions and partly because it focuses specifically on the issue of the reconstruction of Q and not more generally (as before) on the supposed prevalence of Matthean vocabulary in Q.
46. Goulder, Luke, pp. 15-17.
47. A strong case can be made for Luke's secondariness here, in one of the most commonly given examples of alleged Lucan primitivity because of: (a) Luke's well-known preference for "the poor"; (b) Luke's parallel opening of his previous major discourse with good news to "the poor" (4.18); (c) the characteristic Lucan eschatological reversal involving "poor" and "rich" (6.20 and 23; cf. 1.52-3 etc.), later narrativized in the parable of the Rich Man and Lazarus (16.19-31); (d) the oft-ignored narrative-critical observation that Luke's Sermon is addressed to "the disciples", those who have "left everything" to follow Jesus (5.11, 5.28; cf. 14.33); (e) the non-Matthean nature of the term "poor in Spirit". See further my (as yet unpublished) paper, "How Blessed are the Poor?: Source-Critical Reflections on the First Beatitude in Matthew, Luke, Thomas and Q", read at the Synoptics Section, British New Testament Society conference, Bristol, September 1999.
48. This is effectively Goulder's position and is arguably one of the most problematic elements in his thesis. My Goulder and the Gospels, Part 2, accepted Goulder's thesis that Luke was a highly creative writer without the less plausible theory that Luke had virtually no knowledge of oral traditions. See especially pp. 284-7. Although John Drury also follows this path, it is not necessary in the acceptance of the Farrer theory - cf. Austin Farrer, "On Dispensing", p. 85 and Eric Franklin, Luke, throughout.
49. I am grateful to my former supervisor John Muddiman for setting my thoughts off in this direction. See his review of Michael Goulder, Luke in JTS 43 (1992), pp. 176-80.
50. J. Fitzmyer, Luke I-IX, p. 65, "the modified Two-Source Theory has at least led to all sorts of advances in gospel interpretation and has pragmatically established its utility'; cf. Arland Jacobson, The First Gospel, p. 5; W. D. Davies and D. C. Allison, Matthew, p. 98.
Order The Case Against Q (US) / UK