Any serious discussion of the Synoptic Gospels must, sooner or later, involve a discussion of the literary interrelationships among Matthew, Mark, and Luke. This is essential in order to see how an author used his sources (both for reliability’s sake as well as for redactional criticism), as well as when he wrote.
Robert H. Stein’s The Synoptic Problem: An Introduction1 summarizes well the issues involved in the synoptic problem—as well as its probable solution. For the most part, our discussion will follow his outline.2
It is quite impossible to hold that the three synoptic gospels were completely independent from each other. In the least, they had to have shared a common oral tradition. But the vast bulk of NT scholars today would argue for much more than that.3 There are four crucial arguments which virtually prove literary interdependence.
The remarkable verbal agreement between the gospels suggests some kind of interdependence. It is popular today among laymen to think in terms of independence—and to suggest either that the writers simply recorded what happened and therefore agree, or that they were guided by the Holy Spirit into writing the same things. This explanation falls short on several fronts.
This approach is historically naive for the following reasons.
First, it cannot explain the differences among the writers—unless it is assumed that verbal differences indicate different events. In that case, one would have to say that Jesus was tempted by the devil twice, that the Lord’s Supper was offered twice, and that Peter denied the Lord six to nine times! In fact, one might have to say that Christ was raised from the dead more than once if this were pressed!
Second, if Jesus spoke and taught in Aramaic (at least sometimes, if not usually), then why are these verbal agreements preserved for us in Greek? It is doubtful that each writer would have translated Jesus’ sayings in exactly the same way so often.
Third, even if Jesus spoke in Greek exclusively, how is it that not only his words but his deeds are recorded in verbal identity? There is a material difference between remembering the verbiage of what one heard and recording what one saw in identical verbiage.
Fourth, when one compares the synoptic materials with John’s Gospel, why are there so few verbal similarities? On an independent hypothesis, either John or the synoptics are wrong, or else John does not record the same events at all in the life of Jesus.
This approach is also naive regarding the role of the Spirit in inspiring the authors of the gospels.
First, if identical verbiage is to be attributed to Spirit-inspiration, to what should verbal dissonance be attributed?
Second, since John’s Gospel is so dissimilar (92% unique), does this imply that he was not inspired by the Spirit in the writing of his gospel?
In sum, it is quite impossible—and ultimately destructive of the faith—to maintain that there is total independence among the gospel writers.
Although there is a great deal of disagreement in the order of the pericopae among the synoptic gospels, there is an even greater amount of agreement. If one argues that the order is strictly chronological, there are four pieces of data which overrule this. First, there is occasional disagreement in the order. For example, many of Matthew’s parables in chapter 13 are found in Luke 8 or Luke 13. The scribe who approached Jesus about the great commandment is placed in the Passion Week in Matthew and Mark, and vaguely arranged elsewhere in Luke. Second, it is evident that quite a bit of material is grouped topically in the gospels—e.g., after the Sermon on the Mount in Matthew come several miracles by Jesus. Indeed, “Matthew has furthermore arranged his entire Gospel so that collections of narratives alternate with collections of sayings.”4 Third, the early patristic writers (e.g., Papias) recognized that the gospel writers did not follow a strict chronological arrangement. Fourth, there is a studied reserve in the gospels from pinpointing the dates of the various incidents. Introductory comments such as, “immediately,” “after this,” “on another occasion,” “one day,” etc. are the norm. In other words, there seems to be no intent on the part of the evangelists to present a strict chronological sequence of events.
“One of the most persuasive arguments for the literary interdependence of the synoptic Gospels is the presence of identical parenthetical material, for it is highly unlikely that two or three writers would by coincidence insert into their accounts exactly the same editorial comment at exactly the same place.”5 One of the most striking of these demonstrates, beyond the shadow of a doubt, the use of written documents: “When you see the desolating sacrilege . . . (let the reader understand) . . . ” (Matt 24:15/Mark 13:14). It is obvious that this editorial comment could not be due to a common oral heritage, for it does not say, “let the hearer understand.” Cf. also Matt 9:6/Mark 2:10/Luke 5:24; Matt 27:18/Mark 15:10.
Luke begins his gospel in a manner similar to ancient historians: “Inasmuch as many have undertaken to compile a narrative . . . it seemed good to me also . . . to write an orderly account for you . . . .” In the least this implies two things: (1) Luke was aware of written (and oral) sources based on eyewitness accounts; (2) Luke used some of these sources in the composition of his gospel.
Stein has summarized ably what one should conclude from these four areas of investigation:
We shall see later that before the Gospels were written there did exist a period in which the gospel materials were passed on orally, and it is clear that this oral tradition influenced not only the first of our synoptic Gospels but the subsequent ones as well. As an explanation for the general agreement between Matthew-Mark-Luke, however, such an explanation is quite inadequate. There are several reasons for this. For one the exactness of the wording between the synoptic Gospels is better explained by the use of written sources than oral ones. Second, the parenthetical comments that these Gospels have in common are hardly explainable by means of oral tradition. This is especially true of Matthew 24:15 and Mark 13:14, which addresses the readers of these works! Third and most important, the extensive agreement in the memorization of the gospel traditions by both missionary preachers and laypeople is conceded by all, it is most doubtful that this involved the memorization of a whole gospel account in a specific order. Memorizing individual pericopes, parables, and sayings, and even small collections of such material, is one thing, but memorizing a whole Gospel of such material is something else. The large extensive agreement in order between the synoptic Gospels is best explained by the use of a common literary source. Finally, as has already been pointed out, whereas Luke 1:2 does refer to an oral period in which the gospel materials were transmitted, Luke explicitly mentions his own investigation of written sources.6
There are three types of theories which have arisen to explain the literary relationships among the synoptic gospels. First, Schleiermacher in 1817 held that the apostles had written down brief memorabilia which were later collected and arranged according to their particular type of genre. The problem with this view is that it fails to explain the overall arrangement of the synoptic gospels.
Second, G. E. Lessing (1776) and J. G. Eichhorn (1796) argued for an Ur-Gospel, written in Aramaic, which ultimately stood behind the synoptic gospels. The various synoptic writers then used different revisions/ translations of this Ur-Gospel. The main problem with this theory is that it looks no different than an Ur-Mark which, in turn, looks no different than Mark. Thus, rather than postulating any kind of Ur-Gospel, a simpler theory which accounted for the data just as well was that Mark stood behind Luke and Matthew.
Third, the theory of interdependence (sometimes known as utilization) has been suggested. In other words, one or more synoptic gospel used one or more synoptic gospel. Altogether there are eighteen possible permutations of this theory,7 though three have presented themselves as the most plausible: (1) the Augustinian hypothesis: Matthew wrote first and was utilized by Mark whose gospel was used by Luke; (2) the Griesbach hypothesis (suggested by J. J. Griesbach in 1776): Matthew wrote first and was used by Luke, both of whom were used by Mark; and (3) the Holtzmann/Streeter hypothesis (suggested by H. J. Holtzmann in 1863, and refined [and complicated!] by B. H. Streeter in 1924): Mark wrote first and was used independently by Matthew and Luke.8
The majority of NT scholars hold to Markan priority (either the two-source hypothesis of Holtzmann or the four-source hypothesis of Streeter). This is the view adopted in this paper as well.9 Stein puts forth eight categories of reasons why Mark ought to be considered the first gospel. Though not all of his arguments are of equal weight, both the cumulative evidence and several specific arguments are quite persuasive.
Mark’s brevity can be measured in terms of verses or words:
When one compares the synoptic parallels, some startling results are noticed. Of Mark’s 11,025 words, only 132 have no parallel in either Matthew or Luke. Percentage-wise, 97% of Mark’s Gospel is duplicated in Matthew; and 88% is found in Luke. On the other hand, less than 60% of Matthew is duplicated in Mark, and only 47% of Luke is found in Mark.10
What is to account for the almost total absorption of Mark into Matthew and Luke? The Griesbach hypothesis11 suggests that Mark was the last gospel written and that the author used Matthew and Luke. But if so, why did he omit so much material? What Mark omits from his gospel cannot be considered insignificant: the birth of Jesus, the birth of John the Baptist, the Sermon on the Mount, the Lord’s Prayer, the resurrection appearances by Jesus,12 much teaching material, etc. Further, he has abbreviated accounts of the Lord’s temptation and baptism. There are two reasons13 usually given as to why Mark would omit so much material: (1) Mark wanted to provide an abridged gospel for use in the churches; (2) Mark only wanted to record material that was found in both Matthew and Luke, perhaps on the analogy of Deut 17:6-7/19:15 (the voice of at least two witnesses confirmed a truth). Both of these reasons seem inadequate however, for the following reasons.
(1) Mark’s Gospel is not really an abridgment: “whereas Mark is considerably shorter in total length than Matthew and Luke, when we compare the individual pericopes that they have in common, time and time again we find that Mark is the longest!”14 In other words, Mark’s Gospel, where it has parallels with Matthew and Luke, is not an abridgment, but an expansion. Not only this, but the very material he omits would have served a good purpose in his gospel. For example, Mark attempts to emphasize Jesus’ role as teacher (cf. 2:13; 4:1-2; 6:2; 8:31; 12:35, 38, etc.), yet he omits much of what he actually taught. The best explanation of this would seem to be that he was unacquainted with some of these sayings of Jesus, rather than that he intentionally omitted so much—in particular, the Sermon on the Mount. “An abridged work becomes shorter by both eliminating various materials and abbreviating the accounts retained.”15 But the material which Mark eliminates is quite inexplicable on the assumption of Markan posteriority; and the accounts which he retains are almost always longer than either Luke’s or Matthew’s.
(2) It is fallacious to argue that Mark only wanted to record material found in both Matthew and Luke. Yet, W. R. Farmer comes close to this view when he writes that Mark’s Gospel was created as:
a new Gospel out of existing Gospels on an “exclusive” principle. . . . [It was written for liturgical purposes as] a new Gospel [composed] largely out of existing Gospels concentrating on those materials where their texts bore concurrent testimony to the same Gospel tradition. The Gospel of Mark to a considerable extent could be understood as just such a work . . . 16
There is a threefold problem with this. First, it is rather doubtful that Mark intended to write his gospel by way of confirming what was found in both Matthew and Luke. There is little evidence in his gospel that this was an important motif. Rather, if any gospel writer employed this motif, it was Matthew not Mark.17
Second, there is much material—and very rich material—found in both Matthew and Luke that is absent in Mark. In particular, the birth narrative, Sermon on the Mount, Lord’s Prayer, and resurrection appearances. If Mark only produced material found in both Matthew and Luke, why did he omit such important passages which are attested by these other two gospels?
Third, it is quite an overstatement to say that Mark only produced material found in the other two: much of his gospel includes pericopes which are found in only one other gospel.
For examples of exclusively Mark-Matthew parallels, note the following: the offending eye/hand (Matt. 5:29-30 and 18:8-9/Mark 9:43-47); the details about the death of John the Baptist (Matt. 14:3-12/Mark 6:17-29); Jesus walking on the water (Matt 14:22-33/Mark 6:45-52); Isaiah’s prophecy about a hypocritical people and Jesus’ application (Matt 15:1-20/Mark 7:1-23); the Syrophoenicean woman pericope (Matt 15:21-28/Mark 7:24-30); the healing of the deaf-mute (Matt 15:29-31/Mark 7:31-37); the feeding of the four thousand (Matt 15:32-39/Mark 8:1-10); Elijah’s coming (Matt 17:10-13/Mark 9:11-13); the withering of the fig tree (Matt 21:20-22/Mark 11:20-26); the soldiers’ mockery of Jesus before Pilate (Matt 27:28-31/Mark 15:17-20).
What these double-gospel parallels reveal is two things: (1) Mark did not follow the principle of exclusivity, for he includes quite a bit of material which is found only in one other gospel; (2) Mark parallels Matthew far more often than he does Luke (only two pericopes in Mark-Luke vs. ten in Mark-Matthew), negating Farmer’s claim that where Mark only followed one gospel he did so in a balanced way, preferring neither Matthew nor Luke.18
Against a theory of Matthean priority stands the supposition that Luke and Matthew used additional source(s). If so, then the reason they shortened the pericopes they shared with Mark was so that they might include other materials within the length of their scrolls.19
In sum, we could add the now famous statement of G. M. Styler: “given Mk, it is easy to see why Matt. was written; given Matt., it is hard to see why Mk was needed.”20
Stein lists three broad categories of Mark’s poorer stylistic abilities: (1) colloquialisms and incorrect grammar, (2) Aramaic expressions, and (3) redundancies. The first and second arguments are significant for pericopes which Mark shares with either Matthew or Luke; the third is valuable for considering material omitted in Mark.
For example, Mark uses kravbattonin 2:4, a slang word for “mattress” which was banned by such literary writers of the period as Phrynichus and Moeris. The parallels in Matthew and Luke change the word to some form from the root klin- (klivnh, klinivdio"), which was an acceptable literary term. This argument gains strength when it is seen that neither Matthew nor Luke ever uses kravbatton23 (though Mark on three occasions does use the correct word).
Secondly, it is characteristic of Mark to use fevrw in the sense of “lead,” while, strictly speaking, a[gw means “lead,” and fevrw means “bring, carry.” Cf. Mark 7:32 and 8:22.
Sir John C. Hawkins added numerous other grammatical anomalies in Mark including instances of anacoluthon and instances of asyndeton which were corrected or deleted in Matthew or Luke.24
Many have seen Aramaisms in Mark in the very warp and woof of his grammar; in addition to these are seven clear Aramaic expressions in Mark. For example, in Mark 3:17 James and John are called “Boanerges,” an expression not found in the parallels in either Matthew or Luke. Mark speaks of the “Corban” (Mark 7:11), an expression deleted in Matthew’s parallel. Cf. also Mark 7:34/Matt 15:30; etc. “In these seven illustrations the Aramaic expression is missing in all five parallel accounts in Luke and in at least five of the seven parallel accounts in Matthew. . . . for Mark to have added into his Gospel all these Aramaisms, which were not in his source(s), is unexplainable.”25
Mark has redundant expressions on several occasions where both Matthew and Luke omit the unnecessary phrases. For example, in Matt 27:35 we read that the soldiers “divided his garments among them by casting lots”; Luke 23:34 parallels this with “they cast lots to divide his garments”; Mark, on the other hand, adds material easily implied in the others: “they divided his garments among them, casting lots for them, to decide what each should take” (Mark 15:24). Cf. also Mark 2:18/Matt 9:14/Luke 5:33.
T. R. W. Longstaff has recently argued that Mark’s redundancies are merely conflations by Mark of what he found in Matthew and Luke. This is analogous to early Byzantine scribal tendencies to conflate material found in earlier witnesses (i.e., in Alexandrian and ‘Western’ MSS).26 However, this view is inadequate because of the 213 conflations detected in Mark, on only 17 occasions are there two “prongs”—one in Matthew and the other in Luke—which could form the basis for conflation in Mark.27 Thus in something quite a bit less than 10% of the instances could conflation be detected as the motive!
What further argues against the possibility of conflation is the motive:
It is difficult to think that Mark chose to eliminate such material as the Beatitudes, the Lord’s Prayer, and the birth narratives but chose in the examples above to enlarge his accounts by the use of redundant expressions. Such a use of Matthew and Luke by Mark is much more difficult to accept than to believe that Matthew and Luke tended to make such redundant expressions shorter. The redundancy of Mark is best explained on the basis of a Markan priority.28
There are several passages in Mark which paint a portrait of Jesus (or the disciples, etc.) that could be misunderstood. These passages have been altered in either Matthew or Luke or both on every occasion. It is the conviction of many NT scholars that this category is a very strong blow to the Griesbach hypothesis—and one which has not been handled adequately by Matthean prioritists.29 Among the several possible passages which scholars have noticed, the following are particularly impressive to me. Still, the cumulative effect is what makes the biggest impression.
(1) Mark 6:5-6/Matt 13:58—“he could not do any mighty work there except . . . ”/“he did not do many works there . . . because of their unbelief.” On this text Farmer comments: “the passage offers no clear indication that . . . Matthew has ‘toned down’ a phrase in Mark which ‘might cause offense or suggest difficulties’.”30 But this ignores the verbs used, for Mark suggests inability on Jesus’ part, while Matthew simply indicates unwillingness (oujk ejduvnato vs. oujk ejpoivhsen). Cf. also Mark 1:32-34/Matt 8:16/Luke 4:40 for a similar text.
(2) Mark 10:18/Matt 19:17/Luke 18:19—“Good teacher . . . Why do you call me good?” (in Mark and Luke) vs. “Teacher . . . Why do you ask me about what is good?” (Matthew). The text, as Mark has it, might imply that Jesus denies his own deity. It is apparent that Luke did not read it that way, but Matthew probably did. Indeed, in the Holtzmann/Streeter view, Matthew and Luke copied Mark independently of one another. Thus what might offend one would not necessarily offend the other.31
(3) Mark 3:5/Luke 6:10—“he looked around at them with anger/he looked around on them all.” Matthew omits the verse entirely, though he includes material both before and after it (12:12-13). That Luke would omit a statement regarding Jesus’ anger is perfectly understandable.
(4) Mark 1:12/Matt 4:1/Luke 4:1—“the Spirit drove him into the desert” (Mark)/ “Jesus was led into the desert by the Spirit” (Matthew and Luke). Mark uses the very harshejkbavllw, while Matthew and Luke use (ajn)avgw, a much gentler term, to describe the Spirit’s role in bringing Jesus to the desert for temptation.
(5) Mark 8:24-26—the different stages of a particular healing story, omitted in Matthew and Luke. The blind man is partially healed the first time by Jesus, then fully the second time. This is the only healing story in the synoptic gospels which required two stages. Perhaps this was the reason for its omission in Matthew/Luke, or perhaps it was the fact that saliva was used as the means of healing.32
(6) Mark 3:20-21—The statement that Jesus’ mother and brothers tried to seize him because they said that he was insane (ejxevsth). Neither Matthew nor Luke have this verse, apparently because it would cast aspersions on Jesus’ mother and brothers.
The Argument from Verbal Agreements
Stein points out that “Matthew-Luke agreements against Mark are considerably less frequent than any of the other forms of agreement”33 and that what best explains this phenomenon is Markan priority in which Matthew and Luke copied Mark independently of one another. In particular, Markan priority best answers three questions:
(1) Why at times Matthew and Mark agree against Luke—Luke diverges from his Markan source whereas Matthew does not.
(2) Why at times Mark and Luke agree against Matthew—Matthew diverges from his Markan source whereas Luke does not.
(3) Why Matthew and Luke seldom agree against Mark—this would require a coincidental change on the part of Matthew and Luke of their Markan source in exactly the same manner.34
What has indisputably been considered to be the strongest argument for Markan priority is the argument from order. Karl Lachmann was the first to articulate it clearly. The basic argument is both positive and negative: (1) positively: when all three gospels share pericopae, Matthew and Luke agree in the order of those pericopae a great deal; (2) negatively: when either Matthew or Luke departs from the order of Mark in the arrangement of pericopae, they never agree against Mark. To put this another way: in the narratives common to all three, Matthew and Luke agree in sequence only when they agree with Mark; when they both diverge from Mark, they both go in different directions. What best accounts for this? Most NT scholars have assumed that Markan priority does. Some have gone so far as to say that Lachmann proved Markan priority.
In recent decades, however, students of the Griesbach school have debated the argument from order. In particular, B. C. Butler in 1951 boldly called this “the Lachmann fallacy.” His argument was that “if Matthew, Mark, and Luke are directly related to one another rather than being indirectly related through some earlier source which all three have independently copied, then the phenomenon of order no more supports the priority of Mark than priority of Matthew or Luke.”35 This is so because if Mark is the last gospel, then this author could have arranged his material on the basis of common arrangement between Matthew and Luke, and would have followed one or the other whenever they disagreed. This has been quite a tour de force for Matthean prioritists.36
There are four problems with this tour de force. First, this view must presuppose that either Matthew used Luke or that Luke used Matthew. Once that is assumed, several problems surface that are not easily explained.
Second, on this presupposition, one has to wonder why the second gospel (i.e., Matthew or Luke using the other) diverges in its order from the first so frequently. If Luke used Matthew, for example, why did he break up the Sermon on the Mount, leaving out several pericopae? Further, why did he alter/replace the birth narrative with one less colorful—and indeed, one less well suited to his purposes?
Third, this view does not easily explain the large amount of material common to Matthew and Luke, but absent in Mark. But “if we once accept Matthew’s and Luke’s use of a major common source other than Mark to explain this common material, there seems little reason to reject the theory of Markan priority.”37
Fourth, a careful examination of Mark 1:1–6:6 and the parallels in Matthew and Luke38 reveals that the reasons for Luke’s/Matthew’s departures from Mark’s order are well-suited to their various literary purposes, while the supposition that Mark rearranged the material does not fit any easily detected pattern in his gospel.39
In sum, although it would be too bold to say that Markan priority is completely demonstrated by the argument from order, it certainly looks like the most plausible view. Once it is kept in mind that historical reconstruction is concerned with probability vs. possibility, rather than absolute proof either for or against a position, Markan priority stands as quite secure.
“There exist in the synoptic Gospels a number of literary agreements that can best be explained on the basis of a Markan priority. These involve certain omissions and wordings that make much more sense on the basis of Matthew and/or Luke having changed their Markan source than vice versa.”40
“Probably the most weighty argument used today in favor of a Markan priority involves the comparison of the synoptic Gospels in order to note their respective theological emphases.”41 Most commentators assume Markan priority (the commentaries by Mann, Guelich, and Gundry are rare exceptions). “In general it would appear that a Matthean use of Mark provides a clear and consistent redactional emphasis. The same can also be said of Luke’s handling of Mark. On the other had, from the viewpoint of a Markan redaction criticism, a Markan use of Matthew (and/or Luke) seems most unlikely.”42 Several examples can be adduced to show this.
This phrase occurs eleven times in Matthew, four in Mark and Luke. Sheer numbers do not do this justice. Matthew begins his gospel with this phrase (1:1). Further, when a comparison is made, pericope by pericope, it can be seen that this is truly a Matthean emphasis. Cf., e.g., Matt 12:22-24/Mark 3:22/Luke 11:14-15. If Matthew were the first gospel, why would Mark and Luke omit this phrase seven times? That they have no aversion to it is seen from the four references. Further, the four references in Mark match the four in Luke, suggesting that Luke used Mark but was unaware of Matthew.
Matthew’s ten (or eleven) introductory formulae (“this was to fulfill...”) are not duplicated exactly in either Mark or Luke. Since both Mark and Luke use other introductory formulae (such as “it is written”), this shows that they too were interested in linking the life of Jesus to the OT. But would they omit all of Matthew’s formulae? It is easier to believe that Matthew added them to his copy of Mark, in order to show to Jewish Christians that Jesus truly was the Christ. “That the formula quotations are secondary additions to the text is evident in Matthew 1:22; 2:15, 17, 23; 4:14; 8:17; 12:17; 13:35; 21:4; and 27:9. These passages could all be simply excised from their context, and although we would be much poorer as a result, their omission would never be noticed.”44
The word “immediately” (eujquv") is distinctively Markan, occurring over 40 times. Every time Matthew has the word, there is a parallel in Mark. Further, the alternate spelling, eujqevw", is almost always paralleled in Mark by eujquv". “Of the 18,293 words found in Matthew, 10,901 have Markan parallels. In these 10,901 words, ‘immediately’ occurs seventeen times, but in the 7,392 words in Matthew that do not have a Markan parallel, it occurs only once.”45 On the Griesbach hypothesis, we would expect to see twelve instances of “immediately” in the material which finds no parallel with Mark. In other words, Mark’s usage is consistent throughout, while Matthew’s increases only in parallels with Mark. This strongly suggests that Matthew used Mark.
Mark uses an explanatory gavr in an editorial comment 34 times (of his 66 uses of this conjunction). Matthew, on the other hand, uses gavr 11 times in editorial comments (out of his 123 total uses), ten of which parallel Mark’s usage. “Statistically [assuming Matthean priority], one would expect approximately seven such clauses [in Matthew’s non-parallel material]. On the other hand, on the basis of Markan priority, one would expect a greater occurrence of the Markan stylistic feature in the sections of Matthew that have parallels to Mark than in the other sections, and this is exactly what we find.”46
Mark has 151 historical presents, compared to Matthew’s 78 and Luke’s nine. There was an aversion to the historical present by the most literary authors, which well explains Luke’s usage (five of his historical presents are, in fact, found in the parables of Jesus and do not belong to his own narrative style). This consistent use of the historical present by one author vs. the inconsistent use by the other two argues not only that Mark was the first gospel but also that Luke, at least, felt some aversion to the use of the historical present, and consequently chose to alter it to a more literary tense.47
In sum, the redactional argument gains weight on a cumulative basis. When the same redactional, grammatical, and stylistic patterns emerge in one gospel but are inconsistent in another gospel, one has to ask why. If the pattern is insignificant and merely stylistic (such as the use of conjunctions), then presumably the first gospel would be the more consistent one. On the other hand, if the pattern has meaning (e.g., “Son of David”) then the omission/addition of such a rich phrase by one writer would have to be intentional. On this score, it is much easier to see why an author would add such an expression than omit it. On both fronts, then (the significant and insignificant patterns), Mark looks like the source Matthew used, rather than vice versa.
There are many lines which one could draw to illustrate Mark’s more primitive theology. One particular piece of evidence is the use of “Lord” (kuvrio") in the Synoptic Gospels. Mark uses it of Jesus only six times in the triple tradition; Matthew, on the other hand, has it fifteen times in the triple tradition. “It seems reasonable, simply on the basis of numbers, to understand the greater number of instances in which Jesus is called kyrios in Matthew as a secondary development in which this favorite title of the early church is read more and more into the gospel accounts.”48 When one compares all three gospels in their triple tradition, it is evident that nowhere does Mark have “Lord” when either Matthew or Luke has a more primitive term (such as “Rabbi,” or “Teacher”), but on several occasions either Matthew or Luke changes Mark’s less colorful term to “Lord.”
To sum up reasons for Markan priority, the following eight arguments have been given.
(1) The argument from length. Although Mark’s Gospel is shorter, it is not an abridgment, nor a gospel built exclusively on Matthew-Luke agreement. In fact, where its pericopae parallel Matthew and/or Luke, Mark’s story is usually the longest. The rich material left out of his gospel is inexplicable on the Griesbach hypothesis.
(2) The argument from grammar. Matthew and especially Luke use better grammar and literary style than Mark, suggesting that they used Mark, but improved on it.
(3) The argument from harder readings. On the analogy of early scribal habits, Luke and Matthew apparently removed difficulties from Mark’s Gospel in making their own. If Matthean priority is assumed, then what is inexplicable is why Mark would have introduced such difficulties.
(4) The argument from verbal agreement. There are fewer Matthew-Luke verbal agreements than any other two-gospel verbal agreements. This is difficult to explain on the Griesbach hypothesis, much easier on the Lachmann/Streeter hypothesis.
(5) The argument from agreement in order. Not only do Luke and Matthew never agree with each other when they depart from Mark’s order, but the reasons for this on the assumption of Markan priority are readily available while on Matthean priority they are not.
(6) The argument from literary agreements. Very close to the redactional argument, this point stresses that on literary analysis, it is easier to see Matthew’s use of Mark than vice versa.
(7) The argument from redaction. The redactional emphases in Mark, especially in his stylistic minutiae, are only inconsistently found in Matthew and Luke, while the opposite is not true. In other words, Mark’s style is quite consistent, while Luke and Matthew are inconsistent—when they parallel Mark, there is consistency; when they diverge, they depart from such. This suggests that Mark was the source for both Matthew and Luke.
(8) The argument from Mark’s more primitive theology. On many fronts Mark seems to display a more primitive theology than either Luke or Matthew. This suggests that Matthew and Luke used Mark, altering the text to suit their purposes.
Of these eight arguments, the ones that have been most convincing to me are (in order): the argument from order, the argument from Mark’s harder readings (including his more primitive theology),49 the argument from length, and the argument from redaction. On the other hand, what those of the Griesbach school have failed at is to give a convincing reason as to why Mark was ever written. And once written, why would it ever be preserved?50
There are still two questions which must be resolved if Markan priority is to be established as the most probable hypothesis. First, there are numerous places where Matthew and Luke have common material that is absent from Mark. This raises the question as to whether they both used a common source or whether one borrowed from the other. Markan prioritists would say that they both used a common source—given the title “Q”51 (whose nature and existence are disputed)—while Matthean prioritists would argue that Luke used Matthew. Second, there are minor agreements between Matthew and Luke in triple tradition passages which suggest some kind of literary borrowing between these two—if so, then Markan priority is thereby damaged (for Matthew and Luke, in this case, would not have used Mark independently of one another).
Matthew and Luke have in common about 235 verses not found in Mark.52 The verbal agreements between these two is often as striking as it is between Matthew and Mark, Mark and Luke, or Matthew and Mark and Luke. Cf., e.g., Matt 6:24/Luke 16:13; Matt 7:7-11/Luke 11:9-13. Only two viable reasons for such parallels can be given: either one gospel writer knew and used the gospel of the other, or both used a common source. Lukan priority is virtually excluded on the basis of a number of considerations (not the least of which is his improved grammar, as well as the major gap in his use of Mark),53 leaving Matthean priority as the only viable option for intra-gospel borrowing. There are a number of considerations against this, however, as well as a number of arguments in favor of the existence of Q.
“One of the strongest arguments against the use of Matthew by Luke is the fact that when Matthew has additional material in the triple tradition (‘Matthean additions to the narrative’), it is ‘never’ found in Luke.”54 In particular, one ought to note the fulfillment motif of Matthew which is not duplicated in Luke (cf. Matt 8:16-17/Mark 1:32-34/Luke 4:40-41). There is a double problem for the Griesbach school in passages of this sort: (1) Why would Luke omit such rich material, especially since it would well serve the purpose of his gospel? (2) How can we account for the fact that both Luke and Mark omit this material? In the Holtzmann/Streeter hypothesis, however, Luke copied Mark as he had it, while Matthew added material. “If Matthew and Luke both used Mark independently, we would expect that their editorial additions to the account would seldom, if ever, agree with one another. Rather, they would appear as ‘Matthean additions’ and ‘Lukan additions’ to the narratives. And this is exactly what we find.”55
If Luke used Matthew, why does he never place the common (double tradition) material in the same context as it appears in Matthew? Matthew has five well-defined sections of sayings of Jesus which are, for the most part, absent in Mark but present in Luke. In each he concludes the section with “and when Jesus finished these sayings.” But Luke scatters these sayings throughout his gospel. The most common explanation is that Matthew has rearranged the Q material into five topics, while Luke has simply incorporated Q into his document.56 “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.”57
“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?”58 Again, this argument assumes that Matthew has rearranged Q and Luke has not, and it is supported by the premise that Luke’s arrangement is inferior. This argument cannot carry as much weight as Stein gives it if Luke’s structure is also highly artistic, as has recently been demonstrated. However, it still bears some weight: if Luke’s structure is highly artistic as well as Matthew’s, there is every likelihood that both authors have rearranged the material.
On the Griesbach hypothesis, if Luke used Matthew, we would expect Luke to have a more refined development (in theology, dominical sayings, etc.), and Matthew would evidence greater primitivity. Yet, there are times when Matthew is more developed theologically (e.g., the use of kuvrio" to describe Jesus, etc.). If, on the other hand, both used Q (and Mark) independently of one another, we would expect both gospels to alternate between primitivity and development. This, indeed, is what is found. For example, the Lord’s Prayer in Matt 6:9-13 is more elaborate than it is in Luke 11:2-4 (especially in that it adds “your will be done on earth as it is in heaven”). Yet, at the same time, it displays greater primitivity in that it says “forgive us our trespasses” while Luke has “forgive us our sins.” In Matt 7:9-11/Luke 11:11-13 we see that “your heavenly Father gives good things to those who ask him” in Matthew, while he “gives the Holy Spirit” in Luke. Further, in the Sermon on the Mount, Matthew speaks of “the poor in spirit” and those who “hunger for righteousness” while Luke simply refers to the poor and the hungry. Matthew clearly has a more developed articulation of the implication of Jesus’ words than does Luke.59
Overall, Luke has a greater number of harder readings than does Matthew in the common material. This is quite difficult to explain on the basis of Matthean priority. Especially is this so when Luke’s gospel involves certain motifs that would have benefited from Matthew’s articulation. Personally, I regard this as a very strong argument against Matthean priority.
Although (as it has been pointed out) Matthew’s and Luke’s complete lack of agreement in order whenever they disagree with Mark does not prove Markan priority, if Markan priority is assumed, this does establish that Matthew and Luke used Mark independently of one another. Thus, at least one permutation is negated by this evidence, viz., that either Matthew used Luke or Luke used Matthew as a secondary source.60
Finally, the fact that Luke lacks the ‘M’ material (material unique to Matthew) and, conversely, the fact that Matthew lacks the ‘L’ material, argues that neither knew the other. It should be carefully noted that this is not circular reasoning, though on the face of it it seems to be. As soon as we define ‘M’ as material unique to Matthew, then of course Luke would lack it! But that is not the real point of this argument. Rather, it is that there is so much material in Matthew—and rich material at that—which would in all probability have been utilized by Luke had he known of it, that for him not to have used it strongly suggests that he did not know of its existence. (The same can be said for Matthew’s lack of ‘L’ material.) For example, Luke lacks the coming of the magi to Jesus after his birth (Matt 2:1-12)—yet these are Gentiles (a key motif in Luke-Acts). The flight to Egypt, the Great Commission (again, picking up a motif relevant to a Gentile audience) are also missing. In fact, Luke has almost no narrative (as opposed to didactic) material that is not found in Mark. Further, Luke’s birth narrative is so different from Matthew’s that one wonders why he would not try to harmonize it better, assuming that he thought Matthew’s account was reliable.61
As Stein points out,
It is, of course, impossible to know what was going through the mind of Luke when he wrote and why he might have omitted this or that account from his Gospel. Such mental acts are beyond the capacity of the exegete to reconstruct with any certainty. Nevertheless, it is possible to discuss which procedure appears more probable in light of how an Evangelist handles the other material found in his Gospel. It would therefore appear that Luke’s use of Matthew is improbable, due to the lack of his incorporation of the M material into his Gospel.62
In other words, historical reconstruction belongs to the realm of probability vs. possibility, not truth vs. falsehood or certainty vs. uncertainty. To be sure, it is possible that Matthew’s Gospel was the first—and none of the arguments for Markan priority can completely erase that possibility. But whether it most probably is the first gospel is another issue. The arguments for Markan priority speak loudly against that supposition.63
To sum up, if Luke did not use Matthew (as the evidence seems to indicate), then why do Matthew and Luke share so much common material not found in Mark? The only solution is that they got their information from a common source. But was this source oral tradition or a written document? We will deal with this question in the next section, and finally conclude with arguments for the existence of Q.
Scholars have presented four primary arguments that Q was a single, written document.
Many common pericopae between Matthew and Luke have identical or near identical wording, such as is common to triple tradition material. If the exactness of wording in the triple tradition argues that Matthew and Luke used a written document—namely, Mark—as the source, it would seem that double tradition exactness would argue for a written document shared by Matthew and Luke—namely, Q. However, two points militate against this to some degree: (1) There are not nearly as many pericopae in Matthew-Luke as there are in Matthew-Mark-Luke (or Matthew-Mark or Mark-Luke). (2) Several of the pericopae shared by Matthew and Luke have quite dissimilar wording. Thus, Q does not altogether parallel Mark either in quantity (number of pericopae) or quality (identical wording). The evidence, on the whole, argues that Q was both a written document and oral traditions.
Although there are several disagreements in order in the Q material between Matthew and Luke, there are also some general correspondences, and a few that are even striking.64 Still, “if Matthew’s and Luke’s use of Mark can serve as a pattern for how they used their sources, at least one of them did not use his Q source in the same way that he used Mark!”65 Overall, the argument from order still carries some weight, though there seems to be the distinct possibility that Q was both a written document and oral traditions.
Several scholars see doublets as the primary evidence of a written Q. A doublet “refers to the appearance of the same account or text two times in a Gospel.”66 Usually this is a saying of Jesus, though scholars have detected a small number of narratives that seem to be doublets. What is most significant about these doublets is that in almost all of them, one half of the doublet is paralleled in Mark and one half in Q. Altogether, scholars have detected eleven such doublets in Luke and twenty-two in Matthew.67 For example, Matthew records twice the dominical saying about cutting off the offending appendage. Matthew records this saying in Matt 5:29-30 and 18:8-9; it is found in Mark 9:45, 47. Yet, only one of these Matthean texts actually parallels Mark’s passage. Matthew 18:8-9 parallels Mark 9:45, 47 in (1) its arrangement in relation to other pericopae, (2) the amount of verbal agreement,68 and (3) the order and amount of offending body parts within the pericope (Matt 5 has right eye, right hand; Matt 18/Mark 9 have hand, foot, eye [‘right’ is not mentioned in either]).
As striking as these examples are, they stop short of proving that Q was a written source, though they do strongly suggest that Matthew and Luke had some common source besides Mark.
This last argument has suggested that there is a common vocabulary and style in the Q material, suggesting that it is more than mere oral tradition. However, snippets of dominical sayings are so guided by form-critical concerns,69 as well as by the possibility of ipsissima verba and certainly ipsissima vox, that these cannot prove a written document.70 Consequently, most scholars have abandoned this approach in the latter part of the twentieth century.
To sum up, that Q existed is a necessary postulate of Markan priority. For many scholars, this is the very weakness of that hypothesis. But given the severe problems of the other approaches to the interdependence of the gospels, Markan priority stands out as by far the most plausible. If it is true, then Q existed. But what shape did it take? I am inclined to think that Q represented both a written source and oral traditions. I do not think that it has been proved that Q was only a written source. This can especially be measured when one compares the use of Mark in Matthew-Luke with that of Q. If Q was a single written source, it was used in a way that is quite different from how Mark was used.
By way of conclusion, we want to address the arguments against the existence of Q (regardless of what kind of source Q really was).
There are three principle arguments against the existence of Q: (1) Why was it not preserved? (2) If it existed, it apparently consisted almost exclusively of dominical sayings, lacking the birth narrative, the resurrection, etc. Is it conceivable that such a document could have been produced? (3) “It requires a certain overlapping with the materials in order to explain such Matthew-Luke agreements as we find in the baptismal accounts . . . ; the Beelzebul incident . . . ; the parable of the mustard seed . . . ; and the mission charge . . . . The overlapping of the Q material with Mark has often been viewed as an embarrassment for the Q hypothesis and has even been sarcastically referred to as the ‘blessed overlap.’”71 We will address these arguments in chiastic fashion.
First, one should expect some degree of overlap between Q and Mark, especially in the dominical sayings. Not only do the doublets show this, but the fact that both are dealing with the same person would make zero overlap almost inconceivable.
. . . on a purely theoretical basis, it would be most unusual if two sources concerning Jesus, such as Mark and Q, did not overlap in some way. After all, they do deal with the same person, with incidents in his life and sayings that he uttered, so that some overlap would be expected. The issue of overlap serves as an embarrassment for the Q hypothesis only if the hypothesis requires an inordinate amount of such overlapping and is inherently “unlikely” in individual instances. Overlap in the baptism accounts, for example, is by no means that surprising.72
Second, as strange as it may sound to modern ears to think that a document simply of sayings of Jesus might have existed at some time—and would have been meaningful to the early Christians—there are parallels to this.
(1) First, and most important, is the Gospel of Thomas which comprises 114 sayings of Jesus without any connection between them. Although this was a heretical document written at a later date, the analogy is not disturbed: a book of Jesus’ sayings had meaning in the early church.
(2) Second, if Papias’ statement about Matthew writing the logiva of Jesus in Hebrew is authentic in any way, then even Matthew himself might have written a book or several pamphlets of dominical sayings.73 In the Fragments of Papias 2:16 (preserved by Eusebius), Papias says this about Matthew’s Gospel:74 “And concerning Matthew he said the following: ‘Instead [of writing in Greek],75 Matthew arranged the oracles76 in the Hebrew dialect, and each man interpreted them as he was able.’”
(3) Third, there were several agrapha “floating” around in the first two or three centuries of the Church which many patristic writers felt were authentic dominical sayings. Several of them even crept into MSS of the Gospels. It is quite possible that portions of Q have been preserved for us in the agrapha. And if this is not the case, in the least something like Q has been preserved in these agrapha, though in an admittedly fragmentary way.77
Third, why was Q not preserved? As we have suggested, it may well have been preserved in part—either as part of Matthew (who may have authored some of it in the first place), or in the agrapha found in the gospel MSS and among patristic citations. But beyond this are three other considerations. (1) First, in light of Luke’s preface, he apparently used a lot of materials which were not preserved. Why should Q be any different? On any view, the canonical gospels absorbed the best of the previous written documents. (2) Second, transmission history reveals that non-canonical books did not get copied very much at all. (3) Third, patristic writers (and other ancient writers) frequently mention books—and very important and valuable books, judging by their descriptions—which are no longer extant. Why should Q be an exception to this? In light of all this, it is hardly surprising that we do not have Q (especially if it was fragmentary, and, in part, merely oral tradition). Indeed, it would be most surprising if Q was preserved past the end of the first century!78
In this final section on the synoptic problem, we will consider what has been termed as “the major stumbling-block for acceptance of the two-source hypothesis”:
Clearly the key question and major stumbling-block for acceptance of the two-source hypothesis . . . involves the issue of the various Matthew-Luke agreements against Mark. If these “require” that Luke knew (used) Matthew, then both the Q hypothesis and the priority of Mark become questionable. Q would then become unnecessary, for its existence is dependent on Matthew and Luke not knowing each other’s work. Also, although one could still argue for Markan priority if Luke used Matthew, many of the arguments for Markan priority would have been compromised and a Matthean priority would become more attractive.79
To be more specific, there are, in the triple tradition pericopes, four different kinds of minor agreements between Matthew and Luke that are not shared by Mark: (1) agreement in omission of details found in Mark; (2) agreement in addition of details not found in Mark; (3) agreement in expressions and wording against Mark; and (4) agreement in divergence from Mark’s expressions. Altogether, scholars have detected between 272 and 770 minor agreements.80 Our approach will be to look at three categories of minor agreements (organized on a somewhat different principle than above, though excluding none of it), with an attempted response from the Holtzmann/Streeter school.
The great majority of Matthew-Luke agreements belong to this category (180 of the 272, according to Stoldt). This is quite significant, because “if Matthew and Luke omit respectively 6,593 and 8,038 words of Mark’s 11,025 words, there would have to be numerous agreements in omission as a matter of course!81 Not only this, but the argument cuts both ways: If Mark were the last gospel, what is to explain his 180 additions—especially if his was an “abridged” version? Further, on the assumption of Markan priority, if Matthew and Luke wanted to add material from other sources (e.g., Q), many of their common omissions are quite predictable, in light of Mark’s redundancies, Aramaic expressions, etc.
There are scores of agreements between Matthew and Luke against Mark in terms of grammar and editing. On the assumption of Matthean priority, these may well look impressive. But on the assumption of Markan priority, they seem quite irrelevant. Four kinds are discussed below.
Mark uses the historical present 151 times; Matthew, 78; Luke, 9. The data can be used to argue for several different hypotheses. Stein summarizes well how they are used both by those of the Griesbach school and those of the Holtzmann/Streeter school:
Farmer has argued that Luke did not have a strong aversion to the use of the historical present since he has six examples of this in his non-Mark material. This argument is fallacious, however, because according to Farmer, Luke used Matthew. If Luke avoided at least seventy-two of the seventy-eight times in which the historical present is found in Matthew, certainly we cannot disallow his seeking to do the same if he used Mark! Actually the one clear example we possess of how the historical present is treated by a later gospel writer is found in the Gospel of Luke. Luke’s clear tendency, whether he used Matthew or Mark, is to eliminate the historical present. On the basis of the two-document hypothesis, all we need to do to explain these agreements is to presume that Matthew had a similar, although not as thorough, tendency in this area as Luke. On the other hand, according to the Griesbach Hypothesis, we must explain two equally strong but opposite tendencies: Luke sought to avoid the historical present in his Matthean source, and Mark sought to add the historical present to his Matthean source, even though his Lukan source avoided it. To explain the data, the Griesbach Hypothesis therefore requires two totally opposite tendencies on the part of Mark and Luke. The two-source hypothesis does not. The Griesbach Hypothesis also has difficulties in explaining why Mark, with his strong inclination toward using the historical present, did not follow Matthew in the following instances when he has the historical present in the triple tradition: Matthew 8:26; 9:28; 15:12; 17:20; 19:7, and 8. The theory that Matthew and Luke did not know each other does not encounter any real problem in this particular type of Matthew-Luke agreement against Mark.82
In over 30 instances Matthew and Luke use dev while Mark in the parallel passage uses kaiv. This is hardly an argument for Matthean priority, for (1) Matthew and Luke both use dev approximately twice as often as does Mark; (2) literary Greek tried to avoid simple paratactic constructions (especially the overuse of kaiv)—hence, a more literary author would tend to replace kaiv with other conjunctions; (3) it has been demonstrated that the apocryphal gospels based on Mark tended to replace kaiv with dev.83
As we have discussed earlier, Mark uses fevrw in the sense of “to lead” where Matthew and Luke use the more correct a[gw. This kind of agreement is, therefore, quite predictable, given Markan priority and given Luke’s and Matthew’s superior literary skills.
Again, there are a number of miscellaneous agreements between Matthew and Luke which are quite predictable given Markan priori. For example, Matthew and Luke have the more natural and chronologically correct “Moses and Elijah” while Mark has “Elijah with Moses” (Mark 9:4) in the transfiguration account; Matthew and Luke give Herod the more accurate title “tetrarch” while Mark calls him “king” (Mark 6:14); Matthew and Luke speak of Jesus’ resurrection as occurring “on the third day” rather than the more confusing “after three days” (Mark 8:31; 10:34); etc.84 Indeed, these minor agreements are so predictable, given Markan priori, that rather than supporting the Griesbach hypothesis, they strongly confirm the two-source hypothesis!
To be sure, not all the Matthew-Luke agreements are capable of such an easy explanation if Markan priority is true. The number of really significant agreements varies with different scholars: “Fitzmyer lists six: Matthew 26:68, 75; 17:3, 17; 9:7, 20 and their parallels; Hawkins lists twenty[-one]; and Stoldt apparently lists fifty-seven.”85 Among the most significant of these are the following (listed only by Markan reference): Mark 1:7-8; 2:12; 14:65; 3:24, 26-29; 5:27; 6:33; 9:2, 19; 14:72.
Rather than discuss these passages one by one (this paper is already too long!), we will suggest a four-fold complex of reasons as to why such agreements could take place. What should be noted at the outset is two things: (1) since the synoptic problem is not really solved on a single issue, but is rather based on strong cumulative evidence, the very paucity of significant examples of Matthew-Luke agreements is very telling;86 (2) the most significant kind of significant problem will involve places where Matthew and Luke are perceived to be more primitive than Mark. Yet again, even if one or two examples could be produced (and they can), this does not overthrow both the quality and quantity of examples produced on the other side: on almost all fronts Mark’s Gospel appears more primitive.
As we have argued, many of the less significant agreements between Matthew and Luke can be explained this way (e.g., the omission of the historical present), although few, if any, of the most significant agreements can.
Although one has to be careful not to appeal to Q simply to get out of a difficulty,87
it is inconceivable to think that along with Mark (or Matthew or Luke!) there were not also other collections of sayings or gospel-like collections that existed. Time and time again they must have overlapped. Hypothetically there is therefore no reason why Matthew and Luke could not have been influenced by such accounts in the writing of their Gospels. If Farmer can appeal to overlapping traditions, why cannot Streeter? . . . Possibly the Matthew-Luke agreements against Mark in the baptismal accounts can be explained in this manner. Other agreements that may be due to such overlapping are the temptation, the Beelzebul controversy, the parable of the mustard seed, and the mission charge.88
To elaborate on but one example given above: in John the Baptist’s preaching, after all three gospels record him as saying, “I baptize you with water, but he will baptize you with the Holy Spirit” (Mark 1:8), both Matthew and Luke add “and with fire” and then the threat about the winnowing fork (Matt 3:11-12/Luke 3:16-17). But in light of John the Baptist’s message being found in all four gospels, it is obvious that a common oral tradition was known to all four evangelists. There is, then, in this very pericope, evidence of the intermingling of Mark and oral tradition in Luke and Matthew. Whether Q was oral tradition or a written source in this instance, there should be no surprise about overlapping traditions.
This category has been argued in various ways. For example, Streeter felt that our present copies of Matthew and Luke (i.e., the current critical text of Westcott-Hort used in his day) may be corrupt. Hence, if we were to restore the text properly these minor agreements would disappear. Although there is merit in the text-critical principle of disharmony,89 one must never use this to the neglect of the external evidence. Otherwise, this approach looks suspiciously like the tail wagging the dog! For the most part, however, the modern critical texts have excellent credentials in the external evidence. Yet, a few significant Matthew-Luke agreements still remain (even where no MSS produce a disharmony). Thus, this explanation cannot handle all the data.90
In Stein’s argument for Markan priority in the face of the minor Matthew-Luke agreements, he lists overlapping oral traditions separately from Q. But if Q = a written source and oral traditions, then this really is not a separate category.91 Nevertheless, regarding oral traditions, a further point could be made: if certain oral traditions were known and well rehearsed from the earliest days, they would be more familiar to Matthew and Luke than Mark’s Gospel. Hence, if Mark’s Gospel deviated from the oral tradition, Matthew and Luke would be expected to follow the more familiar oral tradition.
In sum, these four reasons for minor Matthew-Luke agreements can explain, to a large extent, why Matthew and Luke have those minor agreements. Indeed, they explain even the most significant Matthew-Luke agreements. However, there is one category of agreement that would seem difficult to explain on this hypothesis: material in the triple tradition in which Matthew and Luke have a more primitive expression than Mark does. Are there any such places where Mark’s phrase is more developed than both Matthew and Luke’s? The answer to this is a qualified “yes.” I have noticed one text in the triple tradition in which Mark is more developed than Matthew-Luke.92 In Mark 14:62, Jesus’ response to the high priest as to whether he was the Christ is “I am” while in Matthew-Luke his response is “you have said it/you say so” (Matt. 26:64/Luke 22:70).93 Although it is possible to see oral tradition playing a strong role especially in a text such as this, one still has to wonder why Matthew and Luke would not alter the text to the stronger affirmation found in Mark. Still, in the overall scheme of things, one text94 is hardly enough to overthrow Markan priori—especially when there are scores of passages in Mark which give the appearance of being much more primitive than either Matthew or Luke.
By way of conclusion, the evidence seems overwhelmingly to support Markan priority. With the labors of William R. Farmer et al., however, the issue is once more becoming alive in English-speaking circles. Perhaps a new breakthrough in how we view the literary relationships is on the horizon. Until then, one has to operate under some hypothesis. And mine is the two-source hypothesis.
The implications of this affect authorship, date, and purpose of the first three gospels. In particular, these areas are impacted once a fairly firm date for Acts can be established. If Acts was written toward the end of Paul’s first Roman imprisonment (c. 61-2 CE),95 then Luke must have preceded it. And if Luke preceded it, Mark must have preceded Luke (mid to late 50s seems most probable). Further, if both Matthew and Luke used Mark independently of one another, it is difficult to conceive of Matthew having been written much later than 62, even if he were cut off as it were from the literary fruits of the nascent Church. Mid-60s would seem to be the latest date for Matthew. Once such a date is assigned for each of these books, then their traditional authorship becomes virtually unassailable. And the purpose for each book would need to be found within the framework of such a date. There is one more implication which can be made from all this, in regard to date: if neither Matthew nor Luke knew of each other’s work, but both knew and used Mark, how long would it take before someone such as John would become aware of any of these books? Since Gardner-Smith demonstrated long ago John’s independence of the Synoptic Gospels, such independence becomes increasingly incredible with every passing year. There is the very distinct possibility that John, too, was written in the mid-60s.96
2 Indeed, I have found Stein’s book so helpful a synthesis of the issues involved, that to a some degree our comments here will be merely a distillation of his work. It should be mentioned, however, that his book is mistitled, for it is not really an even-handed approach to the synoptic problem, but a defense of the priority of Mark.
3 Remarkably, Bo Reicke, in the last book he ever published, argued that the interrelationship among the synoptic writers was that of oral tradition rather than literary (i.e., documentary) borrowing (B. Reicke, The Roots of the Synoptic Gospels [Philadelphia: Fortress, 1986]). As careful a scholar as Reicke has always shown himself to be during his career, it is difficult to see in this work much of substance. In the least the argumentation seems strained at several points, and is often built upon speculation, mere possibility, or argument from silence, rather than sound scholarship.
8 Streeter added that the material unique to Matthew (M) was a written source, and the material unique to Luke (L) was also a written source. Thus, the “Four-Source Hypothesis” (Mark and Q being the first two sources) was born.
9 For almost seventeen years I held to Matthean priority (Griesbach hypothesis), but have in recent years abandoned that view. Although the arguments used in this paper for Markan priority are well-known and will certainly not convince one predisposed to Matthean priority, what tipped the scales for me was greater weight given to internal (literary) considerations and less weight given to external considerations (especially early patristic comments about Matthean priority). The reason for this shift was threefold: (1) my text-critical approach was undergoing a similar metamorphosis, paving the way for me to see internal criteria as very important; (2) not only did the early patristic writers appear to contradict themselves as regards the time and motive of NT writings, but they also had a theological bias for preferring Matthew’s Gospel over Mark’s: it was written by an apostle (further, if their view that Mark got his gospel from Peter has any reliability to it, then Matthean priority is thereby dismantled); (3) this second point is confirmed by the testimony of the MSS: every gospel MS which has all four gospels starts with Matthew, in spite of the fact that the order of the other three varies. Some of the MSS (especially of the ‘western’ strain) place John right after Matthew, thus heading the NT canon with two apostles. Thus, if one were to take this datum seriously (as though it indicated literary interdependence or chronological sequence), he would end up with a view which is not found among any modern synoptic scholars (viz., Matthew-John-Mark-Luke)! Had the testimony of patristic writers been consistent, without built-in bias toward apostolic priority, coupled with rather inconclusive internal evidence, Matthean priority would still have held sway with me. The opposite situation, on all fronts, however, seems to be the case, rendering Markan priority by far still the most plausible view.
17 See especially Robert H. Gundry, Matthew: A Commentary on his Literary and Theological Art, 368 (on Matt 18:16). Gundry, however, takes this view to an extreme in thinking that Matthew at times creates one of the witnesses.
18 Farmer, Synoptic Problem, 281: “It would only be with the greatest difficulty that an adherent of the Gospel of Matthew could convincingly argue that Mark was in balance unduly partial to the Gospel of Luke. Similarly, an adherent of the Gospel of Luke would have had little success in attempting to justify a complaint that Mark’s Gospel was unduly partial to the Gospel of Matthew. This is a very important point.” Indeed, Farmer is correct that this is an important point, for without it his liturgical hypothesis as the raison d'괲e of Mark’s Gospel does not work. Yet, as we have seen, Farmer’s point is not true.
19 It is not insignificant that both Matthew and Luke would be close to thirty feet long in a scroll and that the longest (wieldy) scroll was about thirty to thirty-five feet. Thus, Matthew and Luke had to trim some material in order to make sure all that they wanted to write would be included. Mark’s Gospel, on the other hand, would have been closer to twenty feet on a scroll.
20 G. M. Styler, “The Priority of Mark,” in C. F. D. Moule, The Birth of the New Testament (New York: Harper, 1962), 231. Cited by Stein, Synoptic Problem, 52. This dictum reveals one of the great weaknesses of the Griesbach hypothesis: what was the reason Mark was written?
21 Cf. N. Turner, Style, 11-30, on Markan style in general. Although Turner would rather argue that Mark’s style is largely due to a Hebrew mind-set, he does recognize that most scholars today would affirm that “Mark’s style is unpretentious, verging on the vernacular” (11).
23 This latter is a point not made by Stein, though it would have strengthened his argument. Of all Stein’s grammatical arguments for Mark’s primitiveness, the use of kravbatton is the only one we found convincing.
24 Cf. the complete discussion of “rude, harsh, obscure or unusual words or expressions, which may therefore have been omitted or replaced by others” in Sir John C. Hawkins, Horae Synopticae (2d ed., reprinted; Grand Rapids: Baker, 1968), 131-38.
25 Stein, Synoptic Problem, 58. Stein adds the further observation: “. . . careful writers of Greek avoided foreign words, which might explain why such better writers of Greek as Matthew and Luke would tend to omit the Aramaisms found in their source.”
29 This was most recently brought home to me at a recent Society of Biblical Literature annual meeting. At the “Two Gospel Source” Consultation (the name given to the Griesbach hypothesis group at SBL), in which I was in attendance, several papers were read defending Matthean priority. One of them dealt with the issue of Mark’s harder readings and suggested that Mark’s readings were not harder after all. An elderly scholar, who held to Markan priority, got a bit emotional during the discussion period and blurted out, “I cannot hold to Matthean priority because of Mark’s decidedly harder readings.” He proceeded to catalog several of the passages which are being discussed in this section. Neither W. R. Farmer nor J. B. Orchard had much to say on that occasion, even though Farmer had attempted a rebuttal of this kind of evidence in his Synoptic Problem, 159-68.
31 Farmer argues that since this text caused no offense to Luke, it is a poor example. Actually, this kind of example argues both for Markan priority and that Luke and Matthew used Mark independently of one another (on the other hand, are we to argue that Mark, having both Matthew and Luke in front of him, intentionally chose the more difficult reading?). Further, it is possible that Luke’s Christology intentionally builds to a crescendo through Luke-Acts. Whether Jesus Christ was God in the flesh is not answered in Luke 18—nor indeed clearly until one gets to Acts. But since Matthew wrote only a gospel, he would have to clear up the problem at this juncture.
32 Farmer disputes this, saying that “so far as is known such stories were not regarded as offensive in any sense” (Synoptic Problem, 167). But the evidence is quite scanty upon which to base such an argument. Further, the only other miracle of Mark’s not recorded by Matthew or Luke was the healing of the deaf-mute (Mark 7:31-37), in which Jesus’ saliva was also used (it may be significant that Matthew does parallel this healing, but only in a very general way—cf. Matt 15:29-31).
36 Lachmann’s argument was not simply an argument from order, but a reasoned discussion as to why Markan priority best fits the data. Consequently, it is an overstatement to speak of the “Lachmann fallacy.”
38 This Markan section is the one in which the most significant Matthew/Luke divergences take place. After Mark 6:7, Luke and Matthew almost always follow the Markan sequence.
47 On this score it should be noticed that never does Mark use the historical present in the parables of Jesus and Luke and Mark share only one historical present. His other three came from a different source. Cf. Hawkins, Horae Synopticae, 143-49. Further, on the Griesbach hypothesis, if Mark had both Matthew and Luke before him, would he double the historical presents found in Matthew, recognizing that Luke thought such a grammatical device was overused? Such a supposition could be stated in a cumulative way: Would any author with two fairly literary works in front of him alter them throughout into a less literary fashion? On a modern analogy, who would alter William F. Buckley’s language into something that was not only more colloquial, but also imbibed in grammatical solecisms? This would be like making Shakespeare say “ain’t”!
49 One argument concerning Mark’s harder readings which has been (as far as I can tell) completely overlooked is the probability that neither Luke nor Matthew had pristine copies of Mark at their disposal. In light of the fact that no two (of the more than 5000) Greek NT MSS are exactly alike (the closest two having between six and ten v.ll. per chapter!), it is rather doubtful that Luke’s copy of Mark looked exactly like Matthew’s—even if these were first generation copies. One must be careful, therefore, not to attribute every alteration between the gospels to the author’s redactional purposes. An intermediate scribe is probably responsible—either intentionally or unintentionally—for more than a few of the changes which ended up in Luke and Matthew.
50 Although I have not seen this in print, Markan posteriority is quite analogous to Tatian’s Diatessaron. The fact that that document was banned from the church—even though it contained nothing but material from the four gospels—suggests that if Mark came last, it too would have been banned (or, in the least, hardly copied).
53 Surprisingly, there has become a trend of late to argue for Lukan priority (one or two papers were read in defense of this view at a recent SBL conference), though it has apparently fallen on deaf ears.
57 Ibid. This is probably an overstatement and one which, to some degree, can be tested. On the assumption of Markan priority, one can tell how Luke used Mark and, by way of analogy, see whether or not it corresponds to his use of Q. Since Luke’s use of Mark is not altogether consistent—i.e., he does not utilize all of Mark’s material, nor always arrange it in the same way that Mark has done (so much is left out that some have even suggested that Luke used a mutilated copy of Mark!), could he not have also done a similar thing with Q? But as soon as this is admitted, then Stein’s argument for Markan priority on this front becomes worthless. If Markan priority can be established on other grounds, then what this at least illustrates is that neither Matthew nor Luke is a reliable guide for the arrangement of material in Q—except, of course, where they agree.
61 Gundry sees such disparity between the two birth narratives that he supposes that Luke’s is the more primitive and that Matthew has altered it to the more sensational—so much so that the mandatory sacrifice of the turtle doves by the parents on behalf of the first-born son becomes, in Matthew, the slaying of the infants in Bethlehem by Herod’s henchmen! For what this is worth, in the least to argue that Luke used Matthew seems also to imply that he doubted Matthew’s credibility at one or more points. Furthermore, if one wanted to argue that Luke did, indeed, attempt to subdue Matthew’s material into a less sensational, more reliable gospel, then what is to explain how he handles the pericope about the raising of Jairus’ daughter/the healing of the hemorrhaging woman? For in Matthew’s account Jairus approaches Jesus with the news that his daughter had died, while in Luke’s (and, incidentally, Mark’s) account, the hemorrhaging woman interrupts Jesus’s trek to Jairus’ house and it is during this delay in which the little girl dies. Thus, Luke’s account is more sensational and dramatic. (On the Holtzmann hypothesis, Mark/Luke record accurately the event, and Matthew adapts Mark’s account, telescoping the event in order to get to his next dominical homily. This seems a far more reasonable approach, and implicitly preserves both the reliability of the evangelists as well as their regard for each other’s reliability.)
63 As a sidenote, I am reminded here of what one colleague (of the Griesbach persuasion) has suggested about writing commentaries on the gospels. He has felt that one should simply not address the issue since it is not yet fully resolved. In my view, that is not the best approach. If one embraces a particular historical reconstruction, the best way for it to be tested is in the crucible of exegesis. Further, something happened back then and to ignore this only makes our exegesis so much the poorer. (Is it possible to use the same argument on textual criticism—and simply not decide between variants?!) Always, of course, our exegesis needs to be done with humility, simply because we are not absolutely sure of all of the data. But real progress cannot be made in biblical studies until historical reconstructions are put to the rigorous test of exegetical detail.
64 See Stein, Synoptic Problem, 104-107, for examples. Although Stein does not clear up the problem, this kind of agreement in order must not be confused with the fact that Matthew and Luke never agree in order with one another when either departs from Mark. For when one departs from Mark’s order he might still employ Mark’s material. Further, even though Matthew and Luke might depart from Mark at the same time, they do not, at that time, jump to the Q material. Where they can agree with one another in using Q is in larger segments of material (more than one pericope strung together) which are nevertheless placed in different locations in relation to their material.
69 By this we simply mean that the oral tradition certainly lent shape to different kinds of forms, such as healing stories, pronouncement stories, miracles, etc. This is hardly saying any more than that a TV weather report will not resemble the headline stories in form, nor the sports update. By way of comparison, the apocryphal Gospel of Thomas contains 114 snippets of dominical sayings, each imbibing in a similar form—and that form coming very close to the form of dominical sayings found in the canonical gospels. This does not mean, however, either that Thomas has authentic sayings of Jesus or even that it was the work of one author.
70 One could see an analogy in oral traditions about famous people. For example, if I were to attempt to imitate Richard Nixon, I would stretch my arms over my head, flashing the victory sign, and begin with “My fellow Americans . . . .” Such a form, however, does not presuppose a written document—even though it would be universally known.
72 Ibid. We can add here an analogy. Over fifty years ago, P. Gardner-Smith virtually proved Johannine independence from the Synoptic Gospels. Yet, there are some common accounts found in all four gospels (e.g., the feeding of the five thousand, the cleansing of the temple, the statements about John the Baptist, certain dominical sayings, etc.). And in these common accounts there are occasionally remarkable verbal parallels. Yet, if John did not employ the synoptics to write his gospel, why are these parallels so remarkable? There must have been a common oral tradition that both John and the synoptics drew from. If this is so where all four gospels are concerned, then could it not also be so where only three are concerned? In fact, this creates the distinct possibility that, at times, Matthew and/or Luke altered Mark in light of the oral tradition with which they were more familiar.
73 It is my tentative opinion, though I cannot develop it in this paper, that Matthew might have written several pamphlets of dominical sayings in Aramaic. This is what Papias is referring to (logiva, after all, is not “acts” but “discourses, sayings,”). When Mark’s Gospel was published, Matthew’s audience wanted a framework for the sayings of Jesus. It would have been at this time that Matthew organized the sayings into five thematic units, and used Mark’s Gospel as a framework for them. One of the evidences of this internally is that the narrative material in Matthew is almost merely “stage setting” for the didactic material—each narrative section (except for the birth and passion narratives) concludes with a message by Jesus. The point is that Matthew himself may well have written a document very much like Q (is it even possible that he wrote Q?!).
76 taV logiva—if Papias had just spoken of Mark’s gospel, then the reference is to the same thing (i.e., oral tradition about the life of Jesus). But if Eusebius is merely quoting without giving us a proper context (i.e., if Eusebius has juxtaposed two statements by Papias about the gospel writers which, when originally written, were in different contexts), then taV logiva could refer to the sayings of Jesus. (The problem with seeing Papias’ statements as coming from different contexts is both the subject matter [composition of the gospels] and the connective meVn ou .)
Nevertheless, in light of the possibility that Papias was speaking about the sayings of Jesus, I suggest the following hypothesis about the composition of the first two gospels. Mark recorded Peter’s messages about Jesus while Peter was still alive. At about the same time, Matthew published isolated sayings of Jesus in Aramaic for his and other Jewish-Christian communities. He would, therefore, have been unaware of Mark’s work, just as Mark would have been unaware of Matthew’s. Over the next few years, the dominical material of Matthew would have been translated into Greek. At the same time, Matthew’s own community wanted a framework for these sayings, in light of the publication of Mark’s Gospel. Mark was at hand for the framework, and some of Mark’s material duplicated Matthew’s (e.g., the Olivet Discourse) and was already in Greek. Hence, Matthew used Mark as his basic framework, even where sermonic material was found in Mark. Then, Matthew reorganized these isolated sayings of Jesus into five great sermons (though one was already found in Mark—viz., the Olivet Discourse). For the rest, Matthew simply supplemented Mark with a fulfillment-motif, birth narrative, etc. This hypothesis both affirms Markan priority and Papias’ statement about Matthew’s ‘Hebrew.’ As well, it strongly affirms that Matthew implicitly recognized the reliability of Mark’s Gospel. Still, it does leave several questions open: (1) Does Papias really mean ‘sermons’ when he writes taV logiva for Matthew, but oral tradition or worse, historical narrative when he refers to Mark’s logiva? (See Lampe, A Patristic Greek Lexicon, s.v., logivon , for definitions of Papias’ usage as an argument for our hypothesis.) The quotation by Eusebius seems to imply that the same idea is meant for both authors. (2) If Papias really meant Jesus’ sermons for Matthew’s logiva, then he seems to be privy to a very short-lived tradition (i.e., from the time of the writing of the sermons to their packaging in a Greek gospel), without evidencing knowledge of the Greek gospel itself. (3) Internally, there are still problems between the two gospels: the overlap of the Olivet Discourse seems especially to abandon Ockham’s razor for our hypothesis. Still, if this hypothesis (or a modification of it) has validity, it satisfies not only Markan priority and Papias’ reliability, but also gives strong precedence for something like Q in that Matthew himself would originally have been interested only in the sayings of Jesus.
78 Once the gospels were produced, why would anyone want to make copies of Q? If not, then the most recent copies of Q would have been from the first century. And since we have no extant New Testament MSS from the first century, why should we expect copies of Q to survive?
81 Ibid., 116. Stein quotes Streeter to the same effect: “. . . it would have been quite impossible for two persons to abbreviate practically every paragraph in the whole of Mark without concurring in a large number of their omissions.”
86 I am reminded here of the very few places where the Byzantine text has a better claim to originality than does the Alexandrian text (cf., e.g., the v.l. at Phil 1:14). If, on occasion, the Byzantine does claim to be original, this in no way overthrows the whole weight of evidence either against its general inferiority or its secondary nature as a texttype dependent on Alexandrian and Western traditions.
90 Our earlier comment about textual corruption also does not help out in this regard, for if neither Matthew nor Luke had a perfect copy of Mark to work from, then they did not have identical copies. Actually, if textual corruption took place before Matthew and Luke produced their literary works, this would produce disharmony between them, not harmony—except, of course, in predictable variations. But this does not explain most of the significant agreements.
91 We have listed it separately because (1) Stein lists it separately, and he is more agnostic about the make-up of Q; and (2) this highlights oral tradition as a possible source of minor agreements—especially if one is predisposed to seeing Q as a written source, but does not care to read the contents of each paragraph.
93 One could appeal to textual corruption in this case, for a number of Caesarean MSS insert into Mark’s account, “you have said that” in front of “I am.” But this is not only a harmonization to Matthew’s account, but lacks serious external support.
94 One other passage has plagued me over the years, though it is not found in the triple tradition. In Mark 15:39 the centurion at the foot of the cross claims that “surely this man was a/the Son of God.” In Luke’s parallel account (23:47) we read instead, “surely this man was righteous.” On the face of it, Mark’s Christology (on the lips of the centurion) is higher than Luke’s, suggesting that Mark borrowed from Luke, not vice versa. But there are three considerations which might explain the phenomena: (1) Mark’s expression is ambiguous: ‘a son of God’ might indeed be a lower Christology than ‘righteous’; (2) ‘Son of God’ in Mark seems to be a functional category equivalent to “righteous,” rather than an ontological statement of Christ’s Deity; (3) as we have mentioned before, Luke’s Christology seems to build toward a crescendo, culminating in the middle of Acts; hence, to follow Mark’s wording here might well “let the cat out of the bag” before Luke, in his literary purposes, intended to do so.
96 What is most remarkable in this regard is to note certain authors who believe that the synoptics were written pre-70, but that John was written in the 90s. The vast bulk of NT scholars who date John in the 90s (or later) do so because they date the synoptics in the 80s.