Essay: The Synoptic Problem

The obvious similarity between the three synoptic gospels demands explanation. It is verbally so close over long stretches that the explanation can be only at the level of a written text, written either physically or at any rate verbatim in the memory. The principal solutions proposed are three, the Two-Source Theory, the Two-Gospel Hypothesis or the Neo-Griesbachian solution, and the Multiple Attestation theory. The Goulder theory also deserves discussion, namely that Mark is the first gospel and that Matthew is formed simply by midrash from Mk, without any further source.

External Evidence

Before a discussion of the real evidence, it is important to exclude a large body of evidence sometimes quoted but of little or no probative value, namely the external evidence. The only real external evidence is that of Papias, and the quotations in Clement and the Didache. None of these provide worthwhile indications for our problem.


Papias was bishop of Hierapolis 125-150. He is quoted by Eusebius:

And the Elder used to say this, 'Mark became Peter's interpreter and wrote accurately all that he remembered, not, indeed, in order, of the things said or done by the Lord. For he had not heard the Lord, nor had he followed him, but later on, as I said, followed Peter, who used to give teaching as necessity demanded, but who did not make, as it were, an arrangement of the Lord's oracles, so that Mark did nothing wrong in thus writing down single points as he remembered them. For to one thing he gave attention, to leave out nothing of what he had heard and to make no false statements in them....Matthew collected the sayings in the Hebrew dialect and each interpreted them as best he could.

Here we have the link between Mark and Peter. Apart from a possible reference in Justin, c.150 (depending on whether 'his' refers to Jesus or Peter, and the former seems to me far more likely), this is otherwise unattested. It could well depend upon 1 Pt 5.13, where the author (is it really Peter?) sends greetings from 'my son Mark' to the churches.

The statement on Matthew has been endlessly disputed, the following points being the most contentious:

1. What does logia mean? It could be 'sayings'; it could be wider, to include the actions as well, stories about.

2. Dialekto could mean 'language', but in this case is it Hebrew or Aramaic. In either case, our Matthew as it stands shows no sign of being translated out of either of these languages. Perhaps it means 'idiom', and Matthew is indeed the most semitic of the gospels. It is difficult to conceive, however, that Mt collected them already in semitic form; this semitic colouring must come from the author of the gospel.

3. Ekastos could mean 'both', since ekateros is rarely used at this date. Does Papias mean that Mk and Lk used Mt, that is, that they are derived from him, or that 'everyone' did?

4. Ermeneusen should mean 'translated', especially after mention of dialektos. It could also mean 'explained', but then the pessimistic note of 'as best ,.he could' would have little sense.

This evidence is therefore useless, and can be used only to confirm and fit in with a theory held on other grounds, rather than serving as grounds for any particular theory. The most striking point is that Papias clearly thinks that Mt and Mk worked independently, Mk depending on Peter, and Mt making his own collection of logia. This is clearly not the case: one of these may be correct, but not both. In view of these difficulties and unlikelihoods, it is perhaps best to dismiss Papias' evidence altogether. Eusebius himself does not quote Papias with much confidence.

But the most important point is made by Goulder (Luke, p. 33): Papias is merely attempting to defend the gospels against charges of inconsistency,

1. in order: because Mark, reporting Peter's sermons, had no concern for order, and

2, in wording: because Matthew as we have it is a casual translation from Aramaic.

Seen in this perspective, Papias seems obviously to be clumsy special pleading.

1 Thessalonians, Clement, Didache It is remarkable that all these documents use material which is close to Matthew. It has frequently been argued that they are dependent on Mt, and that Mt must therefore have an early date. However the recent realisation that there are not only possibly previous partial written texts but certainly also a long oral tradition behind the gospels has made it necessary to demand much more exactitude before accepting this claim. It is certainly probable that the final author of Mt drew on the same traditions as the authors of these three works; but subtle variations in detail make it unnecessary or impossible to claim that these works are directly dependent on the gospel. For the Didache a recent examination by Willi Rordorf, 'Does the Didache contain Jesus Tradition independently of the Synoptic Gospels?' (Oral Tradition and the Gospels, 1991), makes clear that the Didache depends on the same traditions as Mt rather than directly on Mt's gospel. These writings therefore provide no argument for an early date for Mt, and therefore no argument for the priority of Mt.

If the external evidence down to the middle of the second century provides no clue about the priority or interrelationship of the gospels, it can hardly be that a reliable tradition was kept secretly. Much of the subsequent tradition is clearly derived from Papias, much of it is simply the result of guesswork and horror vacui. We may therefore disregard it and turn to the arguments from internal evidence.

Internal Evidence

1. Two-Source Theory

a. Theory

Sources of Mt and Lk are Mk + Q.

In triple-tradition passages, source of Mt and Lk is Mk.

a. Order: Mk has 99 pericopes of which 95 are incorporated at least partially into Mt, 76 into Lk. Their order follows Mk except in cases which can be explained on various grounds:

Mt alters position of 13, each time advancing in order to deal exhaustively with a topic on its first appearance, which is typical of his orderly teaching,4.25, 5.13, 7.1-5, 8.1-4, 8.23-27, 9.18-26, 1.1-4, 10.17-25, 13.12, 21.12-17, 25.14-30.

Lk is less uniform, but changes 3.19-20, 4.16-30 for theological reasons, 5.1-11 to get evidence for apostles to follow, 8.19-21 after omitting Mk's framework, 12.1 to improve comprehensibility of his great interpolation,22.24-30, 22.21-23 to form his Abschiedsrede, 22.55-62, 22.63-68 to clarify his passion narrative.

b. Details within pericopes: detailed changes can be explained on theoogical or literary grounds as editing of Mk by Mt and Lk, e.g. both have increased reverence for Jesus and for the apostles, both improve Mk's rough Greek, Mt is more semitic and Lk more polished and historically-minded.

In double-tradition passages, almost entirely sayings of the Lord, source of Mt and Lk is Q, a collection of sayings. There is so much similarity between the two that some written document is indicated. But neither knows the other, and the passages occur in different order in each. It is often difficult to say which has the form more approximate to the original Q.

b. Difficulties

a. Minor Agreements. At least 200 minor agreements, many of which can be explained away as the result of similar editorial processes, e.g. improvement of Mk's rough Gk, removal of historic present, silly diminutives (klinidion), replacement of monotonous kai by de (NB in many cases reverse is possible: an author might prefer roughness). Authorities differ about how many and which are significant, e.g. Hawkins has 21 of which Lagrange leaves 11 aside, though he adds 11 more. But the total number of them is a difficulty, especially the agreeing omissions. This difficulty seems to me tolerable.

Proto-Mk or Deutero-Mk. To make minor adjustments it is possible to postulate an earlier form, where Mk did not have 2.27, which was then added after Mt and Lk had made their copy, or a later form which already contained improvements made by both Mt and Lk. This deals with many minor agreements. But hard to see that some Lukan passages could have developed without knowledge of Mt, e.g. Lk 9.10-11; 10.25-28; 17.1-2.

For some passages, e.g. on divorce, it must be admitted that both Mt and Mk have unsurprising personal additions (Mt refers to the semitic controversy about the grounds of divorce; Mk refers to the possibility of women initiating divorce (not possible in the Jewish world); either the other removed these expansions as being of no interest, or - and this is usually quietly accepted - the later evangelist added them, having copied off an earlier version. But this argument is often complicated by textual criticism.

b. M-Q overlap. All accept that there are seven passages where Mk and Q overlap; these are all passages where the sayings-material needs to be embedded in narrative, e.g. Temptations: Mt 3.11, 4.1-2, 5.13, 12.22-32 (one of the clearest, where a mere glance shows that some bits are pure Mk-Mt, some bits triple, some Mt-Lk), 13.31-32, 18.6-7, 22.34-40. But in fact these passages have to go on being multiplied. The chief difficulty is that verbatim agreement is needed between Mk and Q, so that there must be literary dependence, which merely pushes the difficulty further back. What would then have been the principle of selection used by each, and why did they leave out so much good material? In fact BH Streeter veers between his 1911 opinion that Nk used Q and his 1924 view that they are independent.

2. Two-Gospel Theory or Neo-Griesbach

a. Theory proposed by Griesbach, first editor of a scientific synopsis, in 1776, revived by Orchard and Farmer. Matthew copied by Luke, and Mark synopsis of both, zigzagging. Zigzags within expressions, e.g. Mk's 'when evening came (taken by Mt) when the sun was setting' (taken by Lk). In fact Tuckett replies that there are 213 Markan double-expressions of which only 17 cases are split by the Mt and Lk. But there are so many zigzags that the most economical explanation is such copying rather than independent editing for other reasons in each case. Similarly within pericopes, e.g. Mk 1.32-34, 10.13-16. Finally, in the order of pericopes, e.g. in Mk 4.10-6.52, each time one of Seitenreferenten stops supplying/supporting Mk, the other steps in. As Tuckett, insists this is merely one explanation of the order, no better and no worse than the order-argument of the Two-Source Theory above; the decision must rest on other grounds.

b. Objection general rather than technical: what would the point of Mk be? He adds only vivid detail and a couple of not very special pericopes and duplicate-expressions. But he leaves out so much of the teaching that is important, infancy stories,parables, Lord's Prayer, post-resurrection appearances. This would not be merely for simplicity, to present a simple portrait of Jesus for dramatic effect or for a simple audience.

In detail, why should Mk often follow only Mt, leaving out those parts of Mt already used by Lk? This is an illogical and complicated procedure which would have to have been adopted by Mk in some passages where there are plenty of parallels Mk-Mt and Lk-Mt but none Mk-Lk.

3. Multiple Attestation

Detailed and consistent work on each of the pericopes suggests to Boismard that each gospel was dependent on an earlier version of at least two of the gospels. A is a Palestinian proto-gospel, B a gentile-Christian revision of it, C is in fact the wild-card of no particular character. Mc-inter provides narrative framework of Mk.

Much use of patristic texts, e.g. (final absurdity) on basis of Epiphanius using Korban and s'approchant de Boismard claims that Epiphaius has kept the Mt-inter text of the story, which was copied by Lk 21.1-4 but later dropped by Mt-final.

Boismard claims such minute details as influence of Lk on Mk, e.g. a strikingly Lukan expression, 'baptism of repentance', uyistou Mk 5.7, plhn Mk 12.32.

This leaves room for every sort of linkage, and simply dissolves the problem. But are these sources definite enough and complete enough to be literary ones? Have they any definite shape? They are characterised by Boismard by tendencies and interests, but what makes them hang together? Is not such eclecticism characteristic rather of oral transmission than of literary contact? Many of Boismard's explanations are possible but not compelling, and they involve just too much conjecture and guesswork.

4. Markan Priority without Q

Michael Goulder revives the Austin Farrer theory ('On Dispensing with Q'). This is that Matthew drew on Mk and Lk on both. Goulder also holds that the consistency of the language and imagery throughout the editing of Mt show that it is the work of one mind. Similarly the style and approach throughout Lk is so consistent that the so-called 'Q-passages' are best explained as coming from the same mind as those normally allotted to the Lukan Sondergut. There is thus no ned for Q.

a. Lk depends on Mt

There are some striking Matthaean expressions in Lk, e.g. 13.28 'weeping and grinding of teeth', which occurs only here in Lk but several times and characteristically in Mt (and in Mt backed up by several other typical Matthean expressions). Another element there culled from Mt is the ekballomenouj exw, which makes sense in Mt but not in Lk, since in Lk they are already outside. Further, in Lk 7.27 emprosqen, a favourite Matthean word, is inserted by both Mt and Lk, occurring neither in the original OT quotation nor in Mk; presumably it is culled from Mt. It would also be typical of Mt to correct Mk's attribution of this Malachi phrase to Isaiah, and to remove it from the Baptist pericope but use it elsewhere; in this he is followed by Luke. Also Lk 22.64 has typically Matthean phrase tij estin and Matthean ho + participle.

There are numerous other 'corrections' of Mt by Lk, e.g. Lk 7.18-23, where Lk historicises, introducing some instant miracles (and changing the tenses) on the grounds that in Mt's version the Baptist's messengers had not seen the miracles to which Jesus refers.

b. The Matthean material additional to Mk is consistent throughout, and so is marked by one mind.

i. Parables: Mk's parables are nature-parables, Mt's (like those of the rabbis) are concerned with human relationships. All Mt's 13 long parables are contrasts. Typically of rabbinic stories they move on a grand scale (a king with a private army, a huge debt, talents entrusted to servants). Lk's parables are again different.

ii. Imagery: Mt develops the Markan imagery (22 different animals, instead of Mk's camel, sheep, puppies, birds), moving into a different world from the Galilean village (e.g. 13 terms for measure and money, instead of Mk's simple metron, modion, denarion; Mk has 9 uses of 8 economic images, Mt adds 82 uses of 21 images).

iii. Argumentation: this is shown by Goulder to be the same, rabbinic-style argument throughout, with detailed classification. It is particularly important to show that the same techniques are applied in Mt's editing of Markan material as in the non-Markan or Q material, to avoid the accusation that these marks come from Mt's source rather than his own head. In fact this can be done:

e.g. Mt 10.26, 39 sharpens 'pardics' (a four-point antithetical paradox, so called from 'Can the Ethiopian change his skin or the leopard his spots?'),found in Mk, also (very neatly) Mt 26.33. Another rhythmical structure typical of Mt is the 'poteric' (a parallelism with the same root three times, so called from 'Can you drink the draught that I am drinking?'); Mt 7.2 both improves and doubles the one at Mk 4.24. At Mt 10.41 he triples the figure in Mk 9.41.

c. Lk is dependent on no other document than Mt and Mk

The same style and approach runs throughout Lk, in the editing of Mk, the 'Q-passages' and the Lukan Sondergut. There is therefore no need to postulate any document other than those two, since there is no particular character which needs to be derived from it. The most convincing part of the demonstration is not the vocabulary, which has been widely used in many different ways, but the anaylsis of the style. Goulder shows this by a large range of elements (use of introductions and endings, development of characters, soliloquies, conversations, fascination with work and enjoyment of parties, colourful details over Mt, a middle-class world, a certain logical weakness and prosaic heaviness, double and quadruple rhythms). But he stresses that Lk does not create his material ex nihilo, 'there is always a kernel of gospel tradition behind everything Lk writes' (p. 123). Lk transforms his material in the same way as other contemporary works, e.g. as Targ. Jonathan or Test. Reub. transform Genesis. In just this way Lk transforms into the parable of the Prodigal Son Mt's story of the Father and Two Sons, using all his well-established techniques of characterisation, soliloquy, conversation, his fascination with work and his penchant for words like anastas.


New Jerome Biblical Commentary - art 40, a spirited defence of two-source theory by Frans Neirynck

EP Sanders & M Davies, Studying the Synoptic Gospels, 1989, Pt II.

MD Goulder, Midrash and Lection in Matthew (1974)

id., Luke - A New Paradigm (1989)

Christopher Tuckett, The Revival of the Griesbach Hypothesis (1983)