A Three Source Theory for the Synoptic Problem

Ron Price

Introduction

The prevailing Two Source Theory (hereafter "2ST") posits that the author of Matthew's gospel (hereafter "Au_Matt") and the author of Luke's gospel (hereafter "Au_Luke") independently used the same two written sources, namely Mark and the sayings source Q. The priority of Mark is well established.^[1] However ever since the publication of A.M.Farrer's "On Dispensing with Q",^[2] there has been a growing unease with the 2ST. The most persistent modern supporter of Farrer's view that Au_Luke used Mark and Matthew has been Michael D. Goulder, who wrote a two volume commentary on Luke in order to demonstrate the feasibility of Au_Luke's dependence on these two written sources only.^[3] Hence the label "Farrer-Goulder Theory" (hereafter "FGT"). It is frequently assumed that the 2ST and the FGT are the only reasonable alternatives given Markan priority. This assumption is fundamentally flawed, as is clear from the following quotation of C.M.Tuckett:

It might be that Luke used Q for most of the 'double tradition', but that he also used Matthew's gospel and used it occasionally. Such a theory…is not logically impossible and has had a number of distinguished supporters, past and present.^[4]

Yet none of these supporters has set out a systematic analysis of the basis and implications of such a Three Source Theory (hereafter "3ST"). Before we attempt to do this, we must deal with a common objection: if Au_Luke knew Matthew, surely this eliminates the need for a sayings source. This is not the case. For the arguments from doublets and from the relative primitivity of some of the Lukan pericopae (e.g. 6:20b; 11:2b-4), do not simply vanish if it is decided on other grounds that Au_Luke knew Matthew. Furthermore, it is doubtful whether mere oral tradition could explain how Matthew, written 50 years after the crucifixion, could contain so many authentic-looking sayings which form the majority within the group normally labeled 'Q'.^[5]

Luke's Dependence on Matthew

The main evidence that Au_Luke knew Matthew is as follows: Au_Luke refers to "many writers" who have "undertaken to draw up an account of the events [relating to Jesus]" (Lk 1:1). Q does not fit this description, and under no stretch of the imagination could one gospel (Mark) be referred to as "many". So allowing for a slight exaggeration, Au_Luke was probably referring to Mark and Matthew.^[6] This is the common sense explanation of the verse. Knowing about Matthew and being a good scholar, Au_Luke would undoubtedly have made sure that he obtained a copy of it. He decided to write his own account (Lk 1:3), which can only mean that he was not satisfied either with Mark or with Matthew. Clearly Au_Luke thought he could do better, so we should not be too surprised to find him correcting, developing and even entirely rewriting passages which he found in the earlier gospels.

Luke's Sermon on the Plain is most naturally understood as being named (in effect) and framed by someone who knew Matthew's Sermon on the Mount. Surely it is not a coincidence that Au_Luke had Jesus come down a hill and stop on level ground (Lk 6:17), which looks like a deliberate contrast with Au_Matt's "he went up a mountain". Nor is it a coincidence that Au_Luke created a sermon scenario framed by "he began to speak" (6:20) and "when he had finished addressing the people" (7:1) which closely resembles the Matthean frame "he began to address them" (5:2) and "when Jesus had finished this discourse the people ..." (7:28). For the sermons' closing sentences in both Matthew and Luke are editorial ^[7] and so cannot have been derived from the sayings source.

According to U.Schnelle there are about 700 'minor agreements' between Matthew and Luke against Mark.^[8] It is very unlikely there would have been so many if Au_Luke had never seen a copy of Matthew's gospel. Schnelle tries to solve the problem of the minor agreements by proposing that Au_Matt and Au_Luke used a revised edition of Mark which he labels 'Deuteromark'. But he can see no theological differentiation between the canonical Mark and Deuteromark, nor any rationale for the production of Deuteromark, calling it merely a "stratum of editorial revision".^[9] This is quite unsatisfactory. Moreover no explanation is offered as to why the supposed first edition (Mark) is still extant whereas the supposed later edition (Deuteromark) has been lost.

As we shall see below, several of the double tradition pericopae seem not to have been in the sayings source, either because of inconsistency, or because they appear to have been composed by Au_Matt. The most likely alternative is that Au_Luke took them from Matthew. Many scholars date Luke five years or more later than Matthew. If this relative dating is correct, then notwithstanding the limitations of first century communications it is hardly conceivable that Au_Luke could have been ignorant of what was to become the most popular synoptic gospel in the early centuries of the Christian Church.

The Incongruity of Q

Q as normally understood is quite incongruous as a stand-alone document. The mixture of many sayings with a few narratives in Q makes the identification of its genre extremely problematic.^[10] Moreover the narratives in Q are concentrated near the beginning. Thus Q appears to make a start on the story of Jesus, then abandon it, escaping into the realm of monologue. In an attempt to justify this, Kloppenborg has found parallels in Near Eastern literature which include examples of an "account which demonstrates in narrative fashion that the sage is indeed worthy of the hearer's attention".^[11] But in the more relevant arena of early Jewish or Christian documents, there is no parallel with such a demonstration. The miracles in Q do not preface the sayings, but they are all in the first half, making their distribution appear simply lopsided. Even the monologue lacks consistency, for some of the sayings are attributed to John the Baptist. In addition, Q is so incoherent that its leading advocates have thought it necessary to divide it up, often into three layers or editions. The 2ST would have been more convincing if Q had not lent itself to this surgery. It is impossible either to date or to locate the original composition of Q.^[12] It has no identifiable Sitz im Leben. This is very strange considering how much we know about the early followers of Jesus from Acts and the letters of Paul. It is further testimony against the existence of a document too closely tied to the double tradition material. There is no independent witness to the existence of Q outside the synoptic gospels.

Some Pericopae Are Wrongly Assigned to Q

Terminology and procedure

Some of the pericopae normally assigned to Q include stylistic and other features which make it likely that they were never in Q but were adapted by Au_Luke from a copy of Matthew. For convenience, the sayings thus extracted from Q will be designated "xQ", for 'ex-Q', and the remaining Q sayings will be designated "sQ", for '(pure) sayings-Q'. We examine in turn four reasons for classifying a pericope as originating in Matthew rather than Q. At least two of these reasons apply to each of the pericopae which will be assigned to xQ.

Narratives

There is at least a prima facie case by analogy with the extant Gospel of Thomas that Q contained only sayings, indeed only sayings attributed to Jesus. This would eliminate the lopsidedness mentioned above. The following Matthean pericopae contain narrative, with distinct sentences having non-trivial story content (ignoring asked/said/replied & came/went) indicated in parentheses: 3:7-12 John the Baptist's Testimony (3:7); 4:1-11 the Temptation (4:1,2,5,8,11); 8:5-13 the Centurion's Servant (8:5,10,13); 11:2-19 John the Baptist's Inquiry (11:2); and 12:22-32 the Beelzebul Controversy (12:22,25).

Matthean Style

Phrases

The following phrases look distinctly Matthean:

1.         "You brood of vipers" (Mt 3:7 // Lk 3:7, plus Mt 2; Mk 0; Lk 0).

2.         "Every tree that does not bear good fruit is cut down and thrown into the fire" (Mt 3:10 // Lk 3:9, plus Mt 1; Mk 0; Lk 0).

3.         "gather…wheat…into…barn" (Mt 3:12 // Lk 3:17, plus Mt 1; Mk 0; Lk 0).

4.         "If you are the Son of God" (Temptation (twice), plus Mt 1; Mk 0; Lk 0).

5.         "there will be weeping and gnashing of teeth" (Mt 8:12 // Lk 13:28, plus Mt 5; Mk 0; Lk 0).

6.         "man" (a;nqrwpoj) + a noun (Mt 11:19 // Lk 7:34, plus Mt 5; Mk 0; Lk 0).

7.         "not/no one…in Israel" (Mt 8:10 // Lk 7:9, plus Mt 1; Mk 0; Lk 0).

8.         "take offence at" (skandalisqh/| evn) (Mt 11:6 // Lk 7:23, plus Mt 3; Mk 0; Lk 0).

9.         "prophets…law"/"law…prophets" (Mt 11:13 // Lk 16:16, plus Mt 3; Mk 0; Lk 0).

10.       "my Father" (Mt 11:27 // Lk 10:22, plus Mt 16; Mk 0; Lk 3).

11.       "last…worse than first" (Mt 12:45 // Lk 11:26, plus Mt 1; Mk 0; Lk 0).

12.       "would not" (Mt 23:37) is comparable to the refusals in 22:3 (also xQ) and 21:29.

Goulder ^[13] recognizes phrases as being better than individual words as indicators of Matthean authorship. The idea is that the longer the repeated phrase, the more likely that it is a characteristic of the author. ^[14] The first six phrases on this list are all in Goulder's list. He does claim some other phrases to be Matthean, but they are in pericopae which otherwise show no signs of a Matthean origin.

Individual words

The Lament for Jerusalem contains a whole set of words which are arguably Matthean: "stoning" (liqobolou/sa, Mt 23:37 // Lk 13:34, plus Mt 1; Mk 0; Lk 0); "how often" (posa,kij, Mt 23:37 // Lk 13:34, plus Mt 1; Mk 0; Lk 0 ); "brood", though translating different Greek words, used only here and in the phrase "brood of vipers" (see above). Other individual Matthean words worth noting are "burn" (katakau,sei, Mt 3:12 // Lk 3:17 plus Mt 2; Mk 0; Lk 0); "[play the] flute" (Matt 11:17 // Lk 7:32 plus Mt 1; Mk 0; Lk 0); "babies" (nhpi,oij, Mt 11:25 // Lk 10:21, plus Mt 1; Mk 0; Lk 0); "throne" (qro,nou, Mt 19:28 // Lk 22:30, plus Mt 4; Mk 0; Lk 2); "faithful" (pisto.j) applied to a servant (Mt 24:45 // Lk 12:42, plus Mt 2; Mk 0; Lk 0).

Explicit Old Testament Quotations

Au_Matt was very keen on quoting the Old Testament in support of his presentation of the gospel. The following pericopae contain explicit OT quotations using the standard formula: "it is written": 4:1-11 the Temptation (multiple references) and 11:2-19 John the Baptist's Inquiry (composite reference in 11:10).

Illustrations of aphorisms

Finally, only Matthew has parables which look as if they were specifically written in order to illustrate an aphorism. These are the Workers in the Vineyard (20:1-10, M) illustrating Mt 20:16; the Wedding Banquet (22:1-10, xQ) illustrating Mt 22:14; the Ten Maidens (25:1-12, M) illustrating 25:13; the Talents (25:14-30, xQ) illustrating Mt 25:29. Thus the Wedding Banquet and the Talents are characteristically Matthean. (The former is "thoroughly Matthean" and is tentatively assigned to M by Davies & Allison.^[15] With regard to the latter, the "straightforward explanation" of the muddle in the Lukan version is that "Luke has attempted to reframe the parable that he found in Matthew but his ambition, on this occasion, exceeds his capability".) ^[16]

Matthean Context

Several pericopae look as if either they were written specifically for their Matthean context, or they were dependent on information supplied earlier in the gospel, and if so they cannot have been in Q. The Temptation story (Mt 4:1-11) is remarkably suitable for its context, in between the baptism (c.f. the crossing of the Red Sea) and the Sermon on the Mount (c.f. the giving of the law on Sinai).^[17] In particular, our familiarity with the NT makes it difficult for us to see the incongruity of the question: "If you are the Son of God…" in the context of Q. It makes so much more sense in Matthew coming three verses after the declaration: "This is my beloved Son".

John the Baptist's Inquiry (Mt 11:2-19) contains what looks distinctly like a summary (11:5) of the miracles in Matthew chapters 8-9. Also the mention of the wilderness in Mt 11:7 assumes that the reader knows that John the Baptist preached in the wilderness. But this prior information came in Mt 3:1. It is not in Q. (For this point about the wilderness I am indebted to Mark Goodacre.) The "Woe to Chorazin" is placed appropriately in a sub-section consisting of chapters 11-12 on the theme of "The unbelief of Israel".^[18] Au_Luke's location of this pericope (10:13-15) was probably influenced by its similarity to the Sodom saying (10:12). For him it may have been an indication that the Gentile mission (10:1-12) was to have greater success than the missions to Jewish towns. But in Q as normally reconstructed there was no separate mission to Gentiles. Without the context of a Jew/Gentile contrast, the mention of failures is anomalous coming immediately after the mission's carefully prepared inauguration.

The parable of the unclean spirit (Mt 12:43-45) is moderately intelligible following not long after the Beelzebul controversy. But the latter will be designated xQ, and as the only reference to exorcism the parable would be anomalous. The saying about thrones (Mt 19.28d) makes more sense in the Matthean version where there are 12 thrones, however, this implies a familiarity with the concept of 12 special disciples (e.g. Mt 10:1) which is absent from Q. It also makes more sense when preceded by a reference to Jesus (the Son of man) sitting on his throne as in the first part of the verse. But this first part of the verse contains the lengthy Matthean phrase: "when the Son of man [is seated/will sit] on his glorious throne" (compare 19:28 with 25:31). Therefore, it is almost certain that the whole verse is Matthean and the saying was not in Q. The Wedding Banquet (Matt 22:1-10) "is an allegory very much influenced by 21.33ff."^[19] The Lament for Jerusalem (Mt 23:37-39) looks as if it was written for its Matthean context, for the verses "round off and summarize what has gone before."^[20]

Incompatibility with core sQ sayings

One of the main arguments for the sayings source is that from doublets. We define the core sQ sayings as those which are doublets in either Mathew or Luke, also occurring in Mark, and can reasonably be explained by the use of two written sources. There are around 20 to 25 of them (depending on how they are delineated), and they can be used as the yardstick by which to measure the stylistic and theological compatibility of other candidate sayings.

Stylistically Jesus' Thanksgiving is closer to John than to the Synoptics and is totally incompatible with the core sQ sayings. Miracles are mentioned in the Centurion's Servant, John the Baptist's Inquiry, Woe to Chorazin and the Beelzebul story, however Mt 12:39 // 16:4 // Mk 8:12 states clearly that no sign (shmei/on, a term used frequently in John's gospel to mean "miracle") will be given to this generation. Instead, sQ implies by analogy with Solomon and Jonah that people should repent at the wisdom (sofi,an) and proclamation (kh,rugma) of Jesus. In Mt 11:13 // Lk 16:16 the Mosaic Law is seen to have been superseded already, whereas in Mt 5:18 // Lk 16:17 (sQ), it has yet to be fulfilled. (Compare Mk 13:31 and parallels, which probably arose from Markan adaptation of the sQ saying.) In the Beelzebul story, the kingdom of God is said to have come already, whereas in Mt 10:7 // 4:17 // Mk 1:15 (sQ) it has not yet come, c.f. the plea "May your kingdom come" in the Lord's Prayer (also sQ). In the parable of the Servant set over the household, the master seems to have been delayed (24:48), and in the parable of the Talents a long period of time elapses (Mt 25:19). These are inconsistent with the imminent expectation in Mt 10:7 // 4:17 // Mark 1:15 (sQ).

Summary of the evidence

Key:

Thus, up to four independent criteria show that the thirteen pericopae in the table above were never in Q. These pericopae are therefore labeled xQ. They were probably mostly composed by Au_Matt.^[22]

Conclusion

As can be seen from the source relationship diagrams below, the 3ST provides a natural compromise between the predominant 2ST and the FGT. Au_Luke's use of sources parallels the way his contemporary Josephus used some of his sources. For a similar source diagram represents the use in "Antiquities" (c.f. Luke) of Genesis (c.f. sQ), Kings (c.f. Mark) and Chronicles (c.f. Matthew).

In addition to the priority of Mark on which both the 2ST and the FGT are based, the 3ST retains the best special features of the 2ST: explaining most of the doublets; the greater primitivity of some Lukan pericopae in the double tradition; the presence of authentic-looking sayings in Matthew. It also retains the best special features of the FGT: the acknowledgement that Au_Luke knew Matthew thus providing a simple explanation for most 'minor agreements', and the absence of an incongruous sayings source apparently unknown to early Christians. True, the 3ST does postulate the existence of a sayings source. But it is coherent, well structured, and has a precise Sitz im Leben as well as a probable historical reference.

Notes

1. Markan priority remains a cornerstone for the great majority of synoptic scholars. It has received further support recently from Mark Goodacre's "Fatigue in the Synoptics" in New Testament Studies 44 (1998) 45-54

2. In D.E.Nineham (Ed.), Studies in the Gospels: Essays in Memory of R.H.Lightfoot (Oxford: Blackwell, 1955) 55-88

3. M.D.Goulder, Luke: A New Paradigm (Sheffield: JSOT Press, 1989)

4. C.M.Tuckett, "On the Relationship between Matthew and Luke", New Testament Studies (30) 1984, 130

5. The role of oral tradition is much disputed. We regard it as quite unable to explain the authentic-looking sayings in Matthew for two main reasons. Firstly even in Paul's time oral tradition was not stable, as we can see from his complaint in Gal 1:6 about the Galatians' readiness to replace one "gospel" by another. (For this point I am indebted to T.J.Weeden). Secondly, sayings could be distorted even where a written source was being used, e.g. "You are the Christ" in Mk 8:29 became "You are the Christ, the Son of the living God" in Mt 16:16. After 50 years in oral tradition, few Matthean sayings would have retained signs of authenticity.

6. Goulder refers to "... the many guests at the great dinner in ch. 14, who turn out to be only three" (Ibid., 681). This gives some support to the proposal that Au_Luke referred to the author of Mark plus the author of Matthew as "many".

7. W.D.Davies & D.C.Allison, The Gospel According to Saint Matthew (Edinburgh: T & T Clark, 1988-1997) I, 724

8. U.Schnelle, The History and Theology of the New Testament Writings (ET, London: SCM, 1998) 170

9. Ibid., 171-72

10. Ibid., 191f.

11. J.S.Kloppenborg, The Formation of Q: Trajectories in Ancient Wisdom Collections (Philadelphia: Fortress, 1987) 326

12. Davies & Allison, ibid., I,121

13. M.D.Goulder,  "Self-Contradiction in the IQP" in  J ournal of Biblical Literature 118 (1999) 506-17

14. Goulder's arguments are countered by R.A.Derrenbacker jr. & J.S.Kloppenborg Verbin: "Self-Contradiction in the IQP? A Reply to Michael Goulder" in JBL 120 (2001) 57-76. They consider Goulder's criteria for identifying Mattheanisms to be too lax. But what Goulder's critics underestimate is the quirkiness of several of these phrases. Thus in the list above, "brood of vipers", "weeping and gnashing of teeth" and ANQRWPOS + noun are especially quirky. It is one thing to copy a quirky phrase through reverence for a written source, quite another to use it where there is no such constraint. On the 2ST, these phrases must have been used with no apparent constraint by two different authors. The simpler hypothesis is that all the unconstrained occurrences originated from a single author with an unusual fondness for quirky phrases.

15. Davies & Allison, ibid., III,194

16. Goodacre, ibid., 56

17. J.C.Fenton,  Saint Matthew (Harmondsworth, Middlesex: Penguin Books, 1963) 62

18. Fenton, ibid., 171

19. Davies & Allison, ibid., III,197

20. Fenton, ibid., 378

21. See e.g. B.L.Mack, The Lost Gospel (Shaftesbury, Dorset: Element, 1993) 8

22. According to Goulder, "... the Q material was to a large extent Matthew's own elaboration of Mark" (Luke, ibid., 52). We agree only with regard to the xQ material.