There is a large portion of text (usually referred to as the Triple Tradition) that all three synoptic gospels (Mark, Matthew and Luke) share, and it is generally considered to have been created in one of two ways: Either all three gospels depend on an unknown source for this material (and of course for some people God is that source), or the author of the first of the synoptic gospels wrote it (without defining how or where he obtained it!), and it was seen in his gospel by the authors of the other two. Additionally, there are approximately 4,500 words, or 24.5% of Matthew and 23% of Luke, that are common to Matthew and Luke but that do not exist in Mark, and which collectively are usually referred to as the Double Tradition (DT).
If Mark was the last of the synoptic gospels to be written the DT would be material common to Matthew and Luke that aMark (the author of Mark) chose not include in his gospel. While we have no knowledge regarding why aMark might have done this, it would be something that he certainly could have chosen to do, and on this hypothesis there is no problem regarding how the DT was created. If instead Mark was the first synoptic gospel to be written (so assuming what is generally known as Markan priority) then the DT is material not in Mark that both aMatthew and aLuke chose to put in their respective gospels. The questions that then arise are how did they both come to include the same material and what was its source, the answers to which are highly dependent on the assumed synoptic hypothesis. As the current majority opinion is that Mark was the first synoptic gospel to be written the issues surrounding these questions are explored below on that basis.
Stephen Carlson’s Synoptic Problem Website identifies many hypotheses regarding the actions of the authors of the synoptic gospels that have led to what we see today in those gospels. For example, the formulation of the Three-Source Hypothesis (3SH, or Mark-Q-Matthew model, described in Carlson’s Overview of Proposed Solutions) specifies that aLuke used Matthew as a secondary source of the DT material in addition to a source named Q, as Robert H. Gundry indicates in the introduction to ‘Matthew: A Commentary on His Handbook for a Mixed Church Under Persecution:’
The shared non-Markan tradition included not only the material usually designated Q, but also the nativity story and some of the materials usually regarded as peculiar to Matthew (M) and Luke (L)…
Because Mattheanisms occasionally appear as foreign bodies in Luke, we also have to think of Luke’s using Matthew as an overlay on his primary sources. That Matthew’s gospel did not provide one of those sources is shown by the disarrangement of Matthean material we would otherwise have to suppose. But that the Matthean foreign bodies come from our present Gospel of Matthew, not from an earlier source, is shown by their conforming to Matthew’s distinctive diction, style, and theology as evident elsewhere and by their frequently depending for their point on Matthew’s context (often stemming from Mark), whereas in Luke they lack contextual point. Therefore we need not fret over the numerous minor agreements of Matthew and Luke against Mark that do not fall into the category of Mattheanisms… They, too, may represent Matthean overlay in Luke, though apart from the Matthean foreign bodies we would not have known so with any confidence.
Note: In the 3SH the source referred to as Q can be thought of as a ‘second source’ (or SS) for Matthew and Luke (with Mark being the first source), and this nomenclature will be used here to refer to other potential source documents used by Matthew and Luke in addition to Mark (on the assumption of Markan priority).
The problem with the above description of the sources of the DT in the 3SH is that Q (a hypothetical source originally conceived as containing just sayings of Jesus that are not present in Mark) only exists as a source of the DT in hypotheses in which Matthew and Luke are independent, so that the DT cannot be created by non-Markan material in one being later used in the other. Therefore, in any hypothesis (such as the 3SH) in which Matthew and Luke are not independent a second source common to Matthew and Luke cannot be Q, although it may have similar characteristics to Q. Because on the 3SH the second source is only required to contain that part of the DT not created by aLuke using material he saw in Matthew, the rationale for stating that the second source in the 3SH is Q is invalid (See also What Exactly is Q? and Evidence of Q?).
In a similar manner to the 3SH, in the writer's Mark-Ev Hypothesis aLuke has Matthew as a source in addition to Mark, but instead of Q (which as currently accepted by most contains the DT and virtually nothing else), both aMatthew and aLuke have an form of Early Luke (is this case Marcion's Gospel, here Ev) as a second source. Because Early Luke itself depends on Mark it most likely would contain a significant portion of Mark, in addition to at least some Sondergut (unique) Luke and DT material, and as on the Mark-Ev Hypothesis at least some of the DT can be created by aLuke copying from Matthew (as on the 3SH), an Early Luke also does not meet the requirements of Q. The question of what is in a second source, and whether any particular second source could be Q, may seem unimportant, but in a blog post Mark Goodacre makes the following observations on Variation in the Reproduction of the Double Tradition and an Oral Q? by John S. Kloppenborg:
Perhaps given Kloppenborg's own extensive work on the Synoptic Problem, and given the article's focus specifically on Q, it is churlish of me to make the following remark, but I will make it all the same. A lot of the data gathered here is of interest and relevance more broadly in studies of the Synoptic Problem, and I find it a bit disappointing that the double tradition material is discussed solely in relation to the Q hypothesis, without any mention of competing theories. The issue is particularly focused in relation to verbatim agreement in the double tradition, where one is looking at the coincidence of independent close copying of a hypothetical document by both Matthew and Luke. In other words, it is even more remarkable that Matthew and Luke agree so closely in this double tradition material if they are both doing this independently of one another in relation to another entity (unseen by us). Kloppenborg is right to problematize the high proportion of verbatim agreement in double tradition material with respect to theories about an oral Q; I would like to take it a stage further and problematize the high proportion of verbatim agreement in double tradition material with respect to a written Q.
Goodacre makes a point that unfortunately is obscured by the use of the term ‘DT,’ because if Q was a written source then the DT is in reality another ‘Triple Tradition,’ but one for which the source text (Q) is non-extant. The Mark-Q hypothesis (also 2DH) is symmetrical with respect to Matthew and Luke, i.e. both have exactly the same sources (Mark and Q), and hence examining differences in how on this hypothesis aMatthew and aLuke used both Mark and Q may provide significant insight into the nature of Q in comparison with Mark. For example, are there differences in the proportions of verbatim agreement in Triple and DT material that suggest that Mark and Q have significantly different characteristics?
Are there differences in the proportions of verbatim agreement in Triple and DT material that suggest that Mark and Q have significantly different characteristics? If aMatthew and aLuke both used Q then the differences in their DT material would be due to changes that both made, whereas if aLuke used Matthew then the only differences would be due to changes made by aLuke. This suggests that comparing the degree of difference between the Markan and the DT material in Matthew and Luke could help understand the nature of Q, or whether it even existed.
Goodacre is correct in suggesting that the information presented by Kloppenborg “is of interest and relevance more broadly in studies of the Synoptic Problem,” and this is also true of other studies ostensibly about Q. In particular, ‘Excavating Q: The History and Setting of the Sayings Gospel’ (also by Kloppenborg) provides a wealth of information on the DT that is useful in studies of the synoptic problem in general. For example, on the 2DH Q (by definition) contains the whole of the DT, but as the order of the DT text in Matthew and Luke is different there is no immediately obvious reason for choosing any one particular order for that text in Q (or in any other second source that includes the DT text). However, Kloppenborg provides very useful statistics regarding the DT that help in this regard. For example:
The third significant set of Synoptic data is the considerable bulk of material found in Matthew and Luke but not Mark – the “double tradition.” This consists mainly of sayings or brief stories which feature sayings and amounts to about 4,500 words, approximately the size of 2 Corinthians.19...
19 Kloppenborg 1988a:209. According to my 1988 count, the verbatim agreement between Matthew and Luke amounts to 2,414 Matthaean or 2,400 Lukan words (the disparity deriving from the fact that in a few instances, Matthew or Luke has a doublet [e.g. Matt 9:32-34 / 12:22-24 / Luke 11:14-15]). The total word count for Matthew-Q and Luke-Q pericopae is respectively 4,464 and 4,652 words. Hence the degree of verbatim agreement is 54 percent (of Matthew-Q) or 51.6 percent (of Luke-Q).
Here Kloppenborg indicates that a little more than half of the text of the DT is essentially identical in Matthew and Luke, and as this text is not in Mark it begs the question, where did it originate? On the Mark without Q Hypothesis (MwQH – also Farrer-Goulder hypothesis) in which there is no ‘second source’ the DT can only have been created by aLuke copying from Matthew, while on the 2DH this text can only have come from Q. On both the MwQH and 2DH there is therefore only one source of the DT (Matthew and Q respectively), while on the 3SH both Matthew and Q are sources, and on the Mark-Early Luke Hypothesis there are also two sources (Matthew and an Early Luke), allowing for three paths by which the DT could have been created:
Matthew => Luke - Created or obtained by aMatthew, as on the MwQH;
Early Luke => Luke - Created or obtained by aeLuke (the author of Early Luke), as also on Mark-Q hypothesis;
Early Luke => Matthew => Luke - Created or obtained by aeLuke, and possibly modified and/or re-ordered by aMatthew.
The differences in the hypotheses noted above raise the issue of whether any of the DT text has characteristics that provide clues as to where it originated, i.e. in a non-Markan second source common to Matthew and Luke (Q, Early Luke, or other SS); and/or seen by aLuke in Matthew. For example:
On the MwQH Matthew is the only source for the DT text in Luke, hence the nearly 50% of that text that is not identical in Matthew and Luke is entirely due to aLuke changing non-Markan material he got from Matthew.
On the 2DH both aMatthew and aLuke could have made changes to the text they saw in Q, and hence the differences we see between the Matthean and Lukan versions of the DT are the combination of changes made to Q by both aMatthew and aLuke (from perhaps 25% each to nearly 50% by one and almost none by the other).
On the Mark-Early Luke Hypothesis at least some of the DT text could have been created by the author of Early Luke, modified (and possibly re-ordered) by aMatthew in the process of adding it to Matthew, and finally aLuke could have seen and also modified and possibly re-ordered the two versions he saw in Early Luke and Matthew, allowing the possibility of even more variation in the DT text in Matthew and Luke.
As a result of the differences indicated immediately above the percentage difference between the text of individual DT pericopes in Matthew and Luke may therefore be significant in determining how the DT was formed, and on pp. 62-63 of Excavating Q Kloppenborg comments on the argument by Thomas Bergmann that: “most of the sayings in Luke 6:20b-49, where, he suggests, the agreement [between Matthew and Luke] averages only about 30 percent, were not part of Q,” writing:
What Bergmann proposes is conceivable, of course, but hardly required by the evidence. There is, moreover, a logical problem with his argument. It has to do with his premise that verbal agreement is the only significant ground for assigning a pericope to Q and that, consequently, low verbatim agreement provides a reason to deny a pericope to Q…
Alteration by either Matthew or Luke (or both) would create a disagreement; only when both chose to retain Q’s wording would there be an agreement. Of course, neither editor worked randomly; considerations of content and style, among other things, would have influenced editorial choices. Surely one of the factors that would have influenced their reproduction of Q was that Q was largely sayings of Jesus; but neither Matthew nor Luke displays a consistent or absolute fidelity to the formulation of sayings of Jesus in Mark; hence, there is no a priori reason to think that their reproduction of Q would be significantly better or worse.
Kloppenborg identifies a “logical problem" with Bergmann’s argument, but there is also a “logical problem" with his rebuttal, with him assuming that the sayings of Jesus in Mark do not overlap with those in Q. However, even the proponents of Q allow Mark and Q to overlap in a number of places (The Mark-Q Overlaps, numbering around 15-30 depending on the definition of Q), and Kloppenborg himself shows three such places in Figure 8 in Excavating Q, so opening the door to the possibility that other sayings (that we consider part of the Triple Tradition) were present in Q as well as Mark. Kloppenborg’s point that “neither Matthew nor Luke displays a consistent or absolute fidelity to the formulation of sayings of Jesus in Mark” is therefore invalid because an unknown number of those sayings could have had counterparts in Q. Indeed, a simple thought experiment would show just how unlikely it is that the sayings in Mark and Q did not overlap to a significant degree, especially as some Q proponents believe that Mark knew Q. Alternatively, there could have been yet another sayings source, but given that Q is hypothetical it is simpler to consider the source (Q or other second source) as containing some of the Triple Tradition sayings in addition to the DT.
These differences suggest that a detailed study of how the DT was formed could provide useful information pointing to a particular synoptic hypothesis, or perhaps make one ‘class’ of hypotheses less likely than another. For example, when there are multiple sources for what we know as the DT text the combination of the paths provides a mechanism that allows for bi-directional primitivity (In which in some places Matthew seems earlier or more ‘primitive’ than Luke, but in other places later), as Luke can contain not only some of the text that originated in the second source (e.g. Early Luke), which is earlier than that from Matthew, but also text taken from Matthew itself. It also naturally allows for the creation of doublets in Luke as a result of aLuke possibly incorporating different versions of parallel passages from Early Luke and Matthew into his gospel, in addition to doublets in Matthew resulting from aMatthew seeing different versions of those passages in Mark and Early Luke.
Without any document that we can identify as being Q, Early-Luke, or any other SS all we can do is to analyze the DT texts in both Matthew and Luke for clues as to how they came to differ. Kloppenborg explores this issue from the point of view of the 2DH, beginning with the question of whether Q was even in the form of a single document:
That Q was written is also suggested by the fact that Matthew and Luke concur in the relative sequence of more than 30 percent of the double tradition pericopae … This agreement exists despite the fact that Matthew and Luke combined Q and Mark differently. Such independent agreements in the sequence of Q’s sayings finds its best – perhaps only – reasonable explanation in the supposition that Matthew and Luke used a document and were thus influenced by its arrangement of sayings.
Here Kloppenborg deduces that as nearly one third of the DT sayings are in the same order in Matthew and Luke, but not in the same location relative to the Markan material in those gospels, the most likely explanation on the 2DH is that those sayings originated in a document (i.e. not oral traditions or something similar) that was used by both aMatthew and aLuke, and on the MwQH that document is of course Matthew itself. However, while this is so, it says nothing about the origins of the other two thirds of the DT sayings that are in different orders in Matthew and Luke. Kloppenborg then continues, identifying other agreements in order, again with comments that pre-suppose the 2DH, and using the convention in Q studies that verses in Q are identified by the location of the corresponding parallels in Luke, i.e. that the order of material in Q was taken over completely by aLuke:
There are also two other types of sequential agreements between Matthew and Luke. First is the agreement in the relative order of individual sayings even in clusters which Matthew and Luke place quite differently relative to Mark and to other Q material. For example, the several sayings on John the Baptist (Q 7:18-23, 24-26, 27, 28, 31-35) appear in the same order in Matthew [Mt 11:2-11, 16-19] and Luke even though they are differently placed in the overall outlines of the respective gospels; the parables of the householder (Q 12:39-40) and the faithful servant (Q 12:42b-46) appear in the same sequence despite the fact that Matthew uses them in his apocalyptic discourse and Luke includes then in his Travel Narrative. And Matthew, who connects the parables of the mustard (Q 13:18-19) and the leaven (Q 13:20-21) with the Markan parables discourse (Mark 4 = Matt 13), presents the parables in the same order as Luke, who includes them in his travel section.
Second, Matthew has clusters of DT materials that in Luke are scattered, but nonetheless, Matthew presents the sayings in Lukan order, as if he had scanned Q, lifting out and collecting sayings as he found them in Q. For example:
Matt 10:24-25 Luke 6:40
Matt 10:26-33 Luke 12:2-9
Matt 10:34-36 Luke 12:51-53
Matt 10:37-38 Luke 14:26-27
Matt 10:39 Luke 17:33
Such agreements in sequence are explicable only on the assumption of reliance on a common document, for it would be extraordinarily unlikely that two authors, drawing on a pool of oral sayings, would display so high a degree of sequential agreement, especially if nothing in the sayings requires a particular ordering relative to other sayings.
Although Kloppenborg here presents his argument on the assumption that verses in Q follow the order that we see them in Luke, his above “assumption of reliance on a common document” applies equally to any other hypothesis that includes a second source, such as the 3SH and the Mark-Ev Hypothesis. However, in other portions of Matthew 10 the text in Luke does not follow the same order as that in Matthew. For example:
Matt 10:16a Luke 10:3
Matt 10:16b -----
Matt 10:9-10b Luke 10:4
Matt 10:12-13 Luke 10:5-6
----- Luke 10:7a
Matt 10:10c Luke 10:7b
Matt 10:11 Luke 10:8
Matt 10:8a Luke 10:9a
Matt 10:7 Luke 10:9b
Matt 10:14 Luke 10:10-11a
Matt 10:15 Luke 10:12
Despite (according to Kloppenborg) the previous "agreements in sequence [being] explicable only on the assumption of reliance on a common document," there is ample evidence that in the DT Matthew and Luke do not always follow the same order. For example, earlier in Excavating Q Kloppenborg provides more information on the placement of the DT verses containing makarisms (beatitudes) relative to the Markan parallels:
The double tradition exhibits two important and seemingly contradictory features. First, while there is often a high degree of verbal agreement between Matthew and Luke within these sections, there is practically no agreement in the placement of these sayings relative to Mark. Matthew’s makarisms follow either the call of the four fishermen (Matt 4:18-22 / Mark 1:16-20, 21) or the description of Jesus’ preaching tour through the Galilee (Matt 4:23 / Mark 1:39), depending on how Synoptic tables are aligned… In Luke, however, the makarisms follow the healing of the man with the withered hand (Luke 6:6-11 / Mark 3:1-6) and the naming of the Twelve (Luke 6:12-16 / Mark 3:13-19). In Matthew the woes against the Galilean towns are part of Jesus’ attack on “this generation’s” rejection of John the Baptist and Jesus (Matt 11:2-30), while in Luke they form part of the commissioning speech directed to the seventy-two disciples (Luke 10:1-16). Similarly, the saying on serving two masters appears in the middle of Matthew’s Sermon on the Mount, while in Luke it appears toward the end of the gospel after the parable of the Unjust Steward and various sayings concerning money and greed. There is, in other words, little to suggest that Matthew was influenced by Luke’s placement of the double tradition or vice versa.
Here Kloppenborg is noting that Matthew and Luke very rarely agree on the placement of DT text relative to Mark, suggesting that their authors made independent decisions regarding its positioning. He continues:
Second, if one does not measure sequential agreement of these Matthew-Luke materials relative to Mark, but relative to each other, approximately one-third of the pericopae, accounting for almost one-half of the word count, are in the same relative order. That is, in spite of the fact that Matthew and Luke place double tradition materials differently relative to Mark, they nonetheless agree in using many of the sayings and stories in the same order relative to each other.
In the above quotes Kloppenborg identifies several kinds of agreements in the sequence of DT material in Matthew and Luke, and in his Figure 14 he provides a complete list of all the DT verses, from which the 27 pericopae in Matthew and Luke that agree in sequence have been excerpted and listed below. Here the pericopae are split into two groups to highlight the fact that the great majority of these agreements in order occur in the first half of each gospel, with underscoring marking the notable large gaps in the sequences (notably with all of Matthew 10 in one of the gaps, despite Kloppenborg (above) using Matthew 10 as an example of some agreements in order (but not in a continuous flow of text):
Matthew Luke Matthew Luke
3:1-6 3:1-6 13:31-32 13:18-19
3:7-10 3:7-9 13:33 13:20-21
3:11-12 3:16-17 18:10-14 15:4-7
4:1-11 4:1-13 18:15-22 17:3b-4
4:13 4:16a 24:26-28 17:23-24, 37
5:1-2 6:20a 24:37-41 17:26-30, 34-35
5:3-10 6:20b-23 25:14-30 19:12-27
5:38-48 6:27-36
7:1-2 6:37-38
7:3-5 6:41-42
7:15-20 6:43-44
7:21 6:46
7:24-27 6:47-49
8:5-10, 13 7:1-10
11:2-11 7:18-28
11:16-19 7:31-35
11:20-24 10:13-15
11:25-27 10:21-22
12:22-30 11:14-23
12:38-42 11:29-32
Mt 4:13 / Lk 4:16a are shown above in red because unlike the other DT parallels they have very little in common apart from the name “Nazareth,’ with Jesus leaving Nazareth in Mt 4:13a, but entering it in Lk 4:16a. Although narratively they are located just after the temptation in both Matthew and Luke they are located at very different places with regard to Jesus’ movements. Textually they are very unlikely to both have Q (if it existed) as their source, and instead it is much more likely that Mt 4:13a and Lk 4:30-31a could instead have the same source, as discussed in Capernaum or Nazareth First?
Earlier Kloppenborg provides two lists of DT verses to show where “the double tradition materials in” Matthew “agree in sequence with those in” Luke (providing essentially the same information as shown above) and he follows it by a conclusion with which (given the information presented so far) it is hard to disagree:
This implies that even if Matthew and Luke are not in direct contact with one another, something has influenced them in the overall order of the double tradition.
In this Kloppenborg appears to be correct regarding the "overall" order, but not so regarding the details. Even if Matthew and Luke worked in total isolation we might expect at least some agreement between them due to the common subject matter, but an approximately 50% agreement in order in parallel verses that (by definition) have little to nothing in common with Mark makes some form of common influence almost undeniable. As already noted, assuming Markan Priority the ‘influence’ can only be because of a direct dependence of Matthew or Luke on the other and/or because both depend on another source, and as seen above Kloppenborg argues for a second source in the form of Q.
In addition to identifying the agreements in order between Matthew and Luke Kloppenborg also notes that: “there is practically no agreement in the placement of these sayings relative to Mark,” even though it is clear that both aMatthew and aLuke knew the contents of Mark (with some possible exceptions, e.g. The Great Omission in Luke). Of course, we do not know whether either aMatthew or aLuke created their respective gospels by beginning with the Markan text and inserting the DT and their own unique material into that, or by beginning with their own unique material or a source of the DT and into that inserting text from Mark, so it is equally valid to say that "there is practically no agreement in the placement of" the Markan sayings relative to the DT.
Because of the differences in the placement of the Markan material within Matthew and Luke, with Matthew diverging greatly from Markan order at a ‘macro’ (e.g. pericope) level and with Luke only rarely doing so but instead frequently diverging from Markan order at a ‘micro’ level (e.g. by swapping apparently insignificant details), and the similarities in the placement of the DT material within Matthew and Luke, it is not unreasonable to suggest that the overall framework of Luke is taken from Mark, while the framework of Matthew is taken from either aMatthew’s unique material or whatever he used as the source of the DT text.
Given the above it is a reasonable assumption that aMatthew and aLuke did not independently create the same collection of pericopes that we see as the DT. On this basis, and assuming that aMatthew did not know Luke, aMatthew created his DT text either:
Out of nothing (i.e. it was his own invention); or
From stories and/or sayings he had heard or seen; or
By copying from an existing document.
In the first two cases there would have been no single document used by aMatthew that aLuke could have also used as the source of the DT text, and hence aLuke would have had to merge three things:
Mark;
Some of Matthew (A re-ordered version of most of Mark, interspersed with the text that we see as the DT);
His own additions, that we see as Sondergut Luke.
In addition, aLuke would have had to exclude the text that we see as Sondergut Matthew.
This of course is the situation on the MwQH, in which aLuke has only Mark and Matthew as synoptic sources, which sounds simple when stated in that form. The problem is that simply stating the sources does not suggest HOW aLuke would have been able to merge two versions of the Markan text (from both Mark and Matthew) written in very different orders, his own additions, and some of the non-Markan text in Matthew that he did not otherwise know, while still maintaining the order of over half of the non-Markan text from Matthew (that we know as the DT) that he also incorporated into his gospel.
In a section of ‘The Four Gospels’ entitled ‘The Document Q’, B.H. Streeter (famously) addresses the issue of the procedure followed by aLuke, commenting on how very unlikely it is that the DT material could have been created by either aMatthew or aLuke copying from the other:
How are we to account for this common matter? The obvious suggestion that Luke knew Matthew’s Gospel (or vice versa) and derived from it some of his materials breaks down for two reasons.
(1) Sir John Hawkins once showed me a Greek Testament in which he had indicated on the left-hand margin of Mark the exact point in the Marcan outline at which Matthew has inserted each of the sayings in question, with, of course, the reference to chapter and verse, to identify it; on the right-hand margin he had similarly indicated the point where Luke inserts matter also found in Matthew. It then appeared that, subsequent to the Temptation story, there is not a single case in which Matthew and Luke agree in inserting the same saying at the same point in the Marcan outline. If then Luke derived this material from Matthew, he must have gone through both Matthew and Mark so as to discriminate with meticulous precision between Marcan and non-Marcan material; he must then have proceeded with the utmost care to tear every little piece of non-Marcan material he desired to use from the context of Mark in which it appeared in Matthew—in spite of the fact that contexts in Matthew are always exceedingly appropriate—in order to re-insert it into a different context of Mark having no special appropriateness. A theory which would make an author capable of such a proceeding would only be tenable if, on other grounds, we had reason to believe he was a crank.
(2) Sometimes it is Matthew, sometimes it is Luke, who gives a saying in what is clearly the more original form. This is explicable if both are drawing from the same source, each making slight modifications of his own; it is not so if either is dependent on the other.
Streeter very obviously favors what we know as the 2DH over the MwQH, and in The Existence of Q C.M. Tuckett similarly comments on the laborious process that M. D. Goulder (supporting Streeter) suggests that Luke (on the MwQH) would have had to have followed when copying text from Matthew:
However, in one recent article, Goulder has attempted to face directly the charge of Streeter (see above) that Luke’s ordering of the material, if derived from Mark and Matthew seemed so totally lacking in rhyme or reason that Luke could on this theory only be regarded as a “crank”. Goulder boldly sets out to account for the “order of a crank”. He claims that Luke first went forwards through the texts of Mark and Matthew, taking some of the material he wanted, and by the end of ch. 13 of his gospel had worked his way through to Mt 25; but then, according to Goulder, Luke realised that there was other material from Matthew which he still wanted to include; he therefore decided to go backwards back through the text of Matthew picking up non-Markan material from Matthew he had missed out up until now, sometimes using it directly, sometimes providing a “substitute” of similar material.
In very general terms such a procedure might seem plausible; however, the theory fails to fit the facts in an uncomfortably large number of cases …
Note the above references to Luke 13 and Matthew 25 as being the points at which Luke supposedly changed his copying process, which correspond closely with the writer's earlier observation (above) that the great majority of the agreements in order in the DT occur in the first half of Luke. However, although this appears to confirm that the formation of the DT did not occur in one step, it does not mean that Goulder’s explanation was necessarily correct, and at this point Tuckett continues with several examples that do not “fit Goulder’s explanation” of aLuke’s two-stage process, afterwards concluding:
These are some of the problems that seem to beset Goulder’s theory. His discussion of Luke’s order still provides no very convincing explanation for why Luke should have selected and divided up the material in Matthew in the way he must have done if he knew it in its Matthean form and order. When one couples this with Luke’s very conservative treatment of the order of Mark, the problem becomes even more acute. Why should Luke have had so much respect for the order of Mark, scarcely changing it at all, and yet change the order of Matthew at almost every point? Streeter’s comment that such a procedure seems like that of a “crank”, although expressed somewhat polemically, still has force. Not even Goulder’s defence of the “order of a crank” seems sufficient to meet the problem.
Whether or not Luke did follow the process suggested by Goulder there certainly is a difference that indicates that two processes were involved, and the point is not only whether Luke could, in a methodical process, do what on the MwQH appears to be a wholesale destruction of the order of Matthew’s text, but also why? Of course, this is predicated on the assumption that Matthew was aLuke’s primary source, whereas we might just as (maybe more) reasonably suggest that his primary source was Mark, that he wanted to avoid changing the order of Mark to the order of the Markan material as presented in Matthew, and that he re-arranged what he saw in Matthew as a result. Nevertheless, whether Mark or Matthew was aLuke’s primary source, the fact remains that (on the MwQH) when creating the DT (by writing parallels to some of the non-Markan text in Matthew), aLuke appears to have done so by employing two very different and hard to explain procedures.
In the above table of parallel pericopae that are in the same sequence in both Matthew and Luke there are several large ‘gaps’ (marked by underscores). This does not mean there are no agreements in these gaps, because to a significant extent they are filled with other, smaller collections of agreements in order, some of which are identified by Kloppenborg, noting (also above) that:
“Nearly one third of the Double Tradition pericopae are in the same order (but not in the same position) in Matthew and Luke;”
“There are also two other types of sequential agreements between Matthew and Luke;” and
“The double tradition exhibits two important and seemingly contradictory features.”
Although on the MwQH these observations give rise to the peculiarities noted above regarding the convoluted procedures required by aLuke, Kloppenborg does not appear to recognize that they also indicate that on the 2DH there are multiple mechanisms at work in the ordering of the DT. It appears that after an initial selection and placement of the main sequence of verses in the order in which they saw them in Q, aMatthew and/or aLuke made one or more additional ‘passes’ through their Q material, re-arranging material they had already placed in specific positions in their narratives and/or selecting additional material from Q that they then placed ‘out of sequence.’ The problems this raises for the 2DH are particularly ironic as in On Dispensing with Q?: Goodacre on the Relation of Luke to Matthew, Kloppenborg himself notes an equivalent problem with the MwQH, writing:
Although Goodacre recognizes that Goulder is a far more able defender of the hypothesis than Farrer, he wishes to distance himself from some of the complexities of Goulder’s views, including his theory that each of the Synoptics was composed with a cycle of lectionary readings in mind, and Goulder’s speculation that in the latter part of his gospel Luke was working backwards through Matthew’s gospel.9 …
9 See Goodacre’s critique of Goulder’s lectionary theory in Goulder and the Gospels: An Examination of a New Paradigm (JSNTS 133; Sheffield: JSOT, 1996) 294–362. Goodacre (Case Against Q, 118) regards Goulder’s thesis that in the latter part of his gospel Luke was working backwards through Matthew’s gospel as the ‘most implausible element in Goulder’s thesis’. See Goulder, Luke, 581–3 and the criticism of this aspect of Goulder’s thesis by R. A. Derrenbacker, ‘Ancient Compositional Practices and the Synoptic Problem’ (Ph.D. diss.; Toronto: University of St. Michael’s College, 2001).
Note above Goulder’s speculation that aLuke was doing something different “in the latter part of his gospel,” apparently having in mind that (as noted above) the majority of the sequential parallels between Matthew and Luke occur in (roughly) the first two-thirds of both gospels. Both Kloppenborg and Goodacre recognize that there is something here that they cannot explain just on the assumption that aMatthew and aLuke used Q (Kloppenborg) or that aLuke used Matthew (Goodacre) as the source of the DT text, but perhaps because of their positions neither appear to question whether the problem can be resolved by allowing aLuke to use aMatthew, and aMatthew and aLuke to use a second source (which cannot then be Q because Q only exists in a hypothesis in which Matthew and Luke were independent). Instead, as Jeffrey Peterson notes in Order in the Double Tradition and the Existence of Q, Kloppenborg (and others) see no problem in aMatthew scanning Q multiple times:
Kloppenborg’s work offers no objection in principle to such a theory of multiple ‘scans’ of a source to yield a new arrangement. Indeed, Kloppenborg accepts as ‘brilliant’ Vincent Taylor’s similar account of the composition of Matthew’s discourses out of Q, with the caveat that ‘the more scannings [that] are required, the more cumbersome and the less convincing is this kind of solution’. On this criterion, Lummis’s solution, which accounts for Lucan order in the double tradition via three scans of Matthew, is more plausible than Taylor’s, which accounts for the Matthaean order via 15 scans of Q.
It is clear that even supporters of Q realize that the process that aMatthew must undergo in order to arrive at the order of the DT in Matthew is at best not straightforward, and at worst extremely cumbersome and unlikely. Even so, other solutions are seemingly not considered, perhaps because of similar issues with the process required on the assumption of the MwQH. If aMatthew must jump through multiple hoops on either hypothesis, there is no reason to switch, however unsatisfactory the result may be.
On the 2DH Q is required to contain all the DT text because there is no other way for non-Markan text common to Matthew and Luke to be created. However, if there is a second source (SS) and aLuke knew Matthew then any non-Markan text in Matthew that is not also in SS automatically becomes DT text if aLuke chooses to use it. As a result in this scenario SS does not need to contain the whole of the DT text (because some is in Matthew), and it can also contain non Markan text used by Matthew or Luke but not both, which we see as Sondergut Matthew or Sondergut Luke respectively (This latter point is also true of the 2DH, although this appears to be not generally accepted).
There are synoptic hypotheses in which both SS and Luke depend on Matthew, and in particular some in which SS depends on Mark. In these hypotheses SS could reasonably be expected to largely follow the order of Mark, in which case it could be considered to be an early version of Luke, and could contain non-Markan material later used in both Matthew and Luke, i.e. DT material. It is therefore reasonable to consider whether this SS could have similar characteristics to the ‘expanded Q’ suggested above, in particular whether the orders of the DT material in Q and SS could or could not be the same. Kloppenborg writes that:
… Lukan order for Q is often defended on the basis of an argument from general probability: since Matthew has rearranged and recombined Markan passages, it is likely that he has done the same with Q, and since Luke has in general preserved Mark’s order, he probably preserved Q’s order.
The problem with this is that there is no specific reason to believe that aLuke would treat both Mark and SS (Q or otherwise) in the same way. This also does not take into account either the agreements in order between the DT pericopae in Matthew and Luke or their placement relative to the Markan narrative in both gospels. Kloppenborg continues:
The case for Lukan order, however, can be founded much more securely (Kloppenborg 2987a:72-80). By beginning with (a) the twenty-seven pericopae that Matthew and Luke already have in the same order (see above figure 14), and adding (b) those pericopae that Matthew conflated with a Markan passage,66 (c) those that have been inserted into a Markan sequence,67 and (d) the ten sayings in Matt 10:24-39 mentioned above, one has already accounted for approximately 85 percent of Q. In the pericopae listed under (b), (c), and (d), we are certain that Matthew’s location of the text does not represent Q, since these pericopae are made to function in a Markan context, i.e. in a context that they could not have had in Q.
The 27 pericopae referred to by Kloppenborg (and shown above) show DT agreements in sequence in Matthew and Luke, as do “the ten sayings in Matt 10:24-39” (also shown above), about which Kloppenborg noted that: “Matthew presents the sayings in Lukan order, as if he had scanned Q, lifting out and collecting sayings as he found them in Q.” Kloppenborg also refers to “those pericopae that Matthew conflated with a Markan passage” and “those that have been inserted into a Markan sequence,” which he identifies in footnotes 66 and 67 (shown below combined, and reformatted for readability):
66/67 (those pericopae that Matthew conflated with or inserted into a Markan passage)
Q 6:40 (Matt 15:14 -> Mark 7:1-23);
Q 9:57-60 (Matt 8:18-22 -> Mark 1:29-34);
Q 10:2-12 (Matt 10:7-16 -> Mark 6:6-11;
Q 10:23-24 (Matt 13:16-17 -> Mark 4:12);
Q 11:39-44, 46-52 (Matt 23:4, 6-7, 13, 23, 25-31, 34-36 -> Mark 12:37b-40);
Q 12:10 (Matt 12:32 ->: Mark 3:28-30);
Q 12:11-12 (Matt 10:19-20 -> Mark 13:11);
Q 12:39-40, 42-46 (Matt 24:43-51 -> Mark 13:37).
Q 13:18-19 (Matt 13:31-33 -> Mark 4:30-32);
Q 13:30 (Matt 19:30 -> Mark 10:31);
Q 17:1-2 (Matt 18:7 -> Mark 9:42);
Q 17:6 (Matt 17:20 -> Mark 9:29);
Q 22:28-30 (Matt 19:28 -> Mark 10:29-31).
The point that Kloppenborg is making is that the Matthean verses shown here are either merged with or inserted into text aMatthew saw in Mark, while their Lukan parallels exist either within a sequence of other DT verses, or within material unique to Luke. However, Kloppenborg here is arguing from the viewpoint of the 2DH, because when he refers to “those pericopae that Matthew conflated with or inserted into a Markan passage” he has made an assumption that aLuke did NOT do this (either instead or as well).
What is not stated by Kloppenborg is that these are all examples of places where the text of Mark and Q overlap, usually known as ‘Mark-Q overlaps,’ and hence in all these instances both aMatthew and aLuke would have known the Markan versions of these pericopae in addition to the Q versions. Consequently, as the 2DH is symmetrical with respect to sources (because aMatthew and aLuke both have Mark and Q as their named sources), anything that aMatthew might have done regarding merging or inserting text from Mark into his own work could equally well have been done by aLuke, i.e. with aLuke “merging or inserting text from Mark into his own work.”
Nevertheless, there does appear to be a difference between how aLuke and aMatthew used their two sources. For example, the Markan parallels to Mt 15:12a and 15:15b are Mk 7:17b and 17c respectively, with Mt 15:12b-15a ‘inserted’ into the Matthean parallels of the surrounding Markan text, while with the exception of Lk 6:39b (not 6:40 per Kloppenborg) there are no Lukan parallels at this point, being part of the Great Omission in Luke. In a similar fashion Mt 13:15 has a parallel at Mk 4:12, Mt 13:18 has a parallel at Mk 4:13, and Mt 13:16-17 have been ‘inserted’ into the Matthean parallel of this Markan passage.
As Kloppenborg notes that: “In only a few instances is there a consensus that Luke has reordered Q, and in most instances, it concerns rearrangement of components within a Q pericope,” it seems clear that for the majority of the DT it is safe to say that the order of verses in Q is largely that of the parallels in Luke, and of course this also would apply to any other document (not Q) that contains the DT text in addition to other material. However, the above is not the whole story, as Kloppenborg’s argument does not hold for the rest of the DT text for which the orders in Matthew and Luke are different, i.e. where the DT text in Matthew is not positioned where it would be ‘expected’ if aMatthew was following the order of Q as closely as did aLuke.
For example, because in the DT parallels at Lk 6:22-23 // Mt 5:11-12 and Lk 6:29-30 // Mt 5:39b,40b,42 both Matthew and Luke are following the main sequence, then if aMatthew was following the same sequence as aLuke between these verses we would ‘expect’ to see his parallels to Lk 6:27-28 somewhere after Mt 5:12 but before 5:39b. Similarly, because of the parallels Lk 6:32-33 // Mt 5:46-47 and Lk 6:36 // Mt 5:48a,c we would ‘expect’ the parallel to Lk 6:35d to be between Mt 5:47 and 5:48 instead of at Mt 5:45a. While these re-orderings of the DT text are only small (i.e. by just a few verses), sandwiched between them is Lk 6:31 with an ‘expected’ parallel between Mt 5:42 and 46, but with the actual parallel at Mt 7:12a. For some reason here it appears that aMatthew has lifted out a small piece of DT text and placed it much later in his narrative, and something similar happens with the Matthean parallels to 6:39b-40, 6:45a-45b which we would ‘expect’ to be at around Mt 7:2b-3 and 7:17-20 respectively, but which also appear much later in Matthew, at Mt 15:14b, 10:24a,25a and 12:35, 12:34b respectively. The table in the appendix shows where the verses in Matthew would have to be located to match the Lukan sequence.
Again on the assumption aMatthew did largely follow the order of the DT source, but in a relatively small number of places he chose to diverge both from that order and that placement. Also, it is particularly interesting to note where the DT text does NOT exist in both Luke and Matthew. Despite there being no parallels in Mark to Luke 1-2 and Matthew 1-2, these chapters are so different that they are not considered to be part of the DT, nor are they present in Q. Consequently, the 2DH leaves open the question of the source(s) of the infancy narratives in both Luke and Matthew.
Further on the assumption that the DT source was in the order in which we see that text in Luke, aMatthew makes other very different changes in order. For example, the DT parallels Lk 7:31b-35 // Mt 11:16-19 and Lk 10:13-15 // Mt 11:21-23a suggest that any parallels to DT verses in the range Lk 7:36 – 10:12 should have parallels where we see Mt 11:20. However, instead the Matthean parallels to what we know as Lk 9:57-60a, 10:2-6, 10:7b-11a, 10:12 (i.e. the majority of Lk 9:57-10:12) are split up, and instead of being located close to Mt 11:20 they are spread across three different earlier chapters of Matthew. Additionally, although the majority are in Mt 10 if aMatthew was following the ‘Lukan’ order in that chapter they would be ordered as Mt 10:16a, 9-10a, 12-13, 10b-11, 8, 7, 14. Even where the text in Luke is in consecutive verses (e.g. Lk 10:2-6), in Matthew the text can be spread across different chapters.
It has been suggested that if the gospel authors had multiple sources then they would most likely copy ‘blocks’ of text, first from one source, then the other. However, the above results are at odds with this, and assuming that the source of the DT was a single document that aMatthew and aLuke both followed for the majority of the text, one of them (most likely Matthew) nevertheless scattered small portions of that text across multiple chapters in his gospel. Although both types of sequences of DT text in which Luke and Matthew follow different orders occur in several different places the majority are of the second type, with (assuming Luke followed a single document containing the DT text) aMatthew taking a sequence of verses that we see close together in Luke and spreading them around several chapters (both before and after their ‘expected’ location) in his own gospel.
On this assumption aMatthew did largely follow the order of the DT source, but in a relatively small number of places he chose to diverge both from that order and that placement. Also, it is particularly interesting to note where the DT text does NOT exist in both Luke and Matthew. Despite there being no parallels in Mark to Luke 1-2 and Matthew 1-2, these chapters are so different that they are not considered to be part of the DT, nor are they present in Q. Consequently, the Mark-Q hypothesis leaves open the question of the source(s) of the infancy narratives in both Luke and Matthew.
Of course, the above evidence could be interpreted in a number of ways: that aMatthew largely followed the order of the DT source and aLuke diverged from it in some places; that both aMatthew and aLuke diverged from it in the some places; that where they diverge they were working from different sources; or even that these verses were in a different source to the rest of the DT. It should also be noted that there is no DT text in Luke 5, 8, 9 (except 9:57-60), 15 (except 15:4,5a,6b,7b), 20-24 (except 22:38-30), and Matthew 9 (except 9:32-38), 15 (because of Luke’s Great Omission, except 15:14b), 21, and 22 (except 2b, 3a, 5, 8-10), 26-28. The DT therefore in Luke ends with the parable of the pounds just before Jesus enters Jerusalem (with the exception of Lk 22:28-30), and in Matthew ends with the parallel parable of the talents that takes place in Jerusalem. Thus, in both gospels there is no DT trial or passion material, and hence none is hypothesized to be in Q.
It is also the case that the great majority of the ‘out of sequence’ DT parallels are in the last two-thirds of both Luke and Matthew, so being largely the ‘inverse’ of the earlier finding that the majority of the DT parallels that ARE in sequence are in the first halves of both Luke and Matthew, so strongly suggesting that at least one of aLuke and aMatthew did not select material to include in the latter part of their respective gospels in the same way that they did in the former part. Whether this reflects a change in or an addition to their sources; a change in the priority either gave to their sources; or something else; is another matter.
In Order in the Double Tradition and the Existence of Q Jeffrey Peterson comments on Kloppenborg’s arguments under the heading ‘The Lucan Order Axiom:’
The argument [from order] depends on a common presupposition in research on Q: whereas Matthew subjected the material he derived from Q to his own arrangement, Luke substantially preserved the order of the source. Kloppenborg has offered the most substantial recent defense of what may be termed the Lucan Order Axiom, effectively summarizing previous work on the question. Kloppenborg’s case has three fundamental contentions. (1) Matthew and Luke present a significant amount of the double tradition in common order, which can be credited to Q. (2) The placement of double tradition in Matthew is demonstrably redactional, and therefore secondary. (3) The order of the double tradition in Luke shows no clear signs of redactional arrangement, and as it is unlikely that Luke received this material in its Matthaean sequence and rearranged it to stand as it does in his Gospel, the Lucan order should be recognized as substantially derived from Q.
However, Peterson then writes:
This argument can be criticized under two general heads: (a) it does not afford sufficient grounds on the terms of the Two-Source Hypothesis to accept the Lucan order of the double tradition as the order of Q; and (b) it does not take sufficient account of the alternative possibility that Luke derived his double tradition material from Matthew’s Gospel rather than Q.
Peterson believes that Kloppenborg has not made his case, and also has not taken enough account of aLuke’s possible use of Matthew. However, it is clear that because Kloppenborg brings together his arguments for the order of the DT in Luke being “substantially derived from Q” in a chapter entitled “The Character and Reconstruction of Q,” he is already assuming that Q is the source of the DT, and is not evaluating other possibilities. At issue is therefore whether Kloppenborg’s arguments that Luke largely follows the order of Q are strong enough to withstand attack from other hypotheses, not whether he has done enough to present counter arguments. However, Peterson himself then supplies evidence that shows that whether Q or Matthew was the source of the DT text, the order of that text in Luke’s source largely followed the order we see in Luke:
In an intricate but neglected study, Edwin Lummis anticipated Farrer in holding to Marcan priority and to Luke’s derivation of the double tradition from Matthew, so dispensing with Q. In the 215 verses he ascribed to the double tradition, Lummis found three series of Lucan passages in Matthaean order, with 144 verses (67 per cent) in exactly common sequence and 170 verses (79 per cent) substantially reflecting common sequence. B. H. Streeter’s review, while of course dissenting from Lummis’s conclusion, endorsed his ‘insistence on the relatively large amount of fundamental agreement in order between the Q material in Matthew and Luke, if one looks below the surface.
It should be clear from the above discussion that although portions of the DT share a common order in Matthew and Luke, and that for these portions a literary connection is the most reasonable explanation, the connections themselves do not suggest whether both Matthew and Luke depend on another document for this material, or whether one depends on the other. In addition, the portions of the DT that do not have a common order in Matthew and Mark indicate that the inclusion of the DT material in Matthew and Luke did not take place in a simple one-step process, but instead some of the material (either in Matthew or Luke) was added after the bulk of the material for which the order is the same in both gospels.
In Q, the Earliest Gospel: An Introduction to the Original Stories and Sayings of Jesus Kloppenborg discusses the ordering of pericopes in Mark, Matthew and Luke, stating that “the three often agree in relating the same incidents in the same relative order,” then continuing (parallels in Luke added in blue for completeness):
Although in the early part of Matthew (3-13), Matthew and Mark have a different order of events, from Matthew 14:1 and Mark 6:16 onward the two Gospels agree almost completely in sequence. The significance of the strong agreement cannot be missed. If Matthew and Mark were completely independent tellings of the story of Jesus, it is unlikely that the writers would choose to relate all stories and sayings in the same order, especially where there was no thematic or narrative reason to do so. For example, there is no special reason why the story of Jesus’ argument with the Pharisees about washing hands (Matt. 15:1-20 || Mark 12:13-17) should appear just before the story of the Syro-Phoenician woman’s daughter (Matt. 15:21-28 || Mark 7:24-30) [|| Lk 6:39 only], or why the controversies about payment of taxes (Matt. 22:15-22 || Mark 12:13-17 || Lk 20:20-26) comes just before the controversy about the resurrection (Matt. 22:23-33 || Mark 12:18-27 || Lk 20:27-38). Yet Matthew and Mark agree in these sequences and in many more pericopae. Luke and Mark agree in sequence even more strongly than do Matthew and Mark, and this indicates that copying has occurred.
Thus, we can conclude that some kind of literary relationship exists among the first three Gospels. At this point we cannot decide who is copying whom, but it is clear that both the wording and the sequence of the three Gospels is the result of literary interaction.
The problem with these statements is that, while the agreements in order are as stated by Kloppenborg, he couches his conclusion in terms of direct copying from one gospel to another, so obscuring the fact that the “literary interaction” could be via one or more unknown document, i.e. that either two or all three synoptic authors could have copied from other sources while keeping the order of pericopes that they saw in those other sources. Given that he very clearly favors an unknown document (Q) as the source of the DT, Kloppenborg’s failure to note such a possibility here is curious, to say the least.
Kloppenborg then argues that Mark is the earliest gospel, using in his arguments “the materials in the Synoptics where Matthew, Mark, and Luke have parallel stories,” i.e. the Triple Tradition, stating that: “While Matthew often agrees with Mark’s wording of a story or saying, and while Luke often agrees with Mark’s wording, it is relatively rare to find Matthew and Luke agreeing when Mark has a different wording.” He then uses as an example “the story of the healing of Simon’s mother-in-law in Matthew 8:14-15 || Mark 1:29-31 || Luke 4:38-39,” commenting:
What is important to note here is that Matthew and Luke do not agree with each other against Mark in any detail. It is true that Matthew and Luke fail to repeat some of the details in Mark: “immediately” (twice); “and Andrew, with James and John”; and “approaching he raised her up” (underscored). But they do not agree positively against Mark.
In other words, aMatthew and aLuke agree on removing text they saw in Mark but not in adding new text. Kloppenborg continues:
… If there had been a direct connection between Matthew and Luke, we should expect Matthew sometimes to agree with Luke against Mark. But agreements of this sort are in fact quite uncommon (although there some that we shall have to discuss later). The fact that Matthew and Luke tend to agree with Mark, but not against Mark, means that Mark is medial.
It is important to note that the above examples all come from the latter part of Matthew, and while it is reasonable to believe they show a connection between Matthew and Luke, Kloppenborg’s use of the term ‘medial’ with regard to Mark cannot be taken to preclude a direct connection between Matthew and Luke as well as via Mark. In particular, Kloppenborg is here only looking at part of the evidence, because as has been pointed out above the great majority of the DT pericopae in which Matthew and Luke agree in sequence are in chapters 3-13 of both Matthew and Luke, so we have the following two distinct sets of relationships:
In roughly the first halves of Mark, Matthew and Luke (Mk 1-6:15, Mt 3-13, Lk 3-13) “Matthew and Mark have a different order of events” while the non-Markan parallels in Matthew and Luke (i.e. the DT) largely follow the same order;
In the remainder of Mark, Matthew and Luke (Mk 6:16-16, Mt 14-28, Lk 14-24) Matthew and Mark “agree almost completely in sequence” (Kloppenborg) while the DT in Matthew and Luke largely follows different orders.
These observations suggest that the DT text constitutes the ‘backbone’ of the order of Mt 3-13, into which aMatthew wove heavily re-arranged text from Mark, while Mark constitutes the backbone of the order of Mt 14-28, into which aMatthew wove the remaining small portions of the DT. In contrast, aLuke kept as much of the order of Mark as he could in Lk 3-13 (so placing the DT sayings differently than in Mark), as he also did with the DT text. Whether aLuke was doing something different “in the latter part of his gospel” (Goulder) or not, it is clear that the relative lack of DT material in the latter halves of both Matthew and Luke had an effect on how they were constructed.
After his conclusion above that “some kind of literary relationship exists among the first three Gospels” Kloppenborg presents four ‘models’ of transmission of text from one gospel to another, which can be represented by the following:
a) Straight Line: Matthew -> Mark -> Luke
b) Straight Line: Luke -> Mark -> Matthew
c) Simple Branch: Mark -> Matthew and Mark -> Luke
d) Conflation: Matthew -> Mark and Luke -> Mark
However, Kloppenborg introduces these arrangements by writing: “In fact several arrangements of the Gospels are possible with Mark as the middle term” as if these four are the only possible arrangements of the gospels, but fails to point out that Mark can still be the ‘middle term’ in arrangements that include other links (e.g. Mark -> Matthew, Mark -> Luke, and Matthew -> Luke), perhaps because for him “middle term” allows for no links (direct or otherwise) adding other connections to Matthew and or Luke, so that the connections between Mark and Matthew, and Mark and Luke, are symmetrical.
Kloppenborg makes no reference to this or any other possible arrangement because he is only showing those in which: “there is no direct connection between Matthew and Luke, and hence, no possibility of them agreeing with each other apart from Mark, except by coincidence.” He is therefore excluding these other arrangements from his future arguments without having provided a rationale for so doing. After discussing arrangements a), b), and c) he writes:
The fourth model, conflation (d) … is more complicated that the others, since it presupposes that Mark had before him both accounts and moved back and forth between the two, picking elements from one, then the other. Such a model is not logically impossible, but examination of how other authors worked who combined two sources reveals that no known ancient author would have taken the trouble to compare sources so closely and to zigzag between them. An ancient conflator would more likely have taken over Matthew’s account or Luke’s but not bothered to micro-conflate them.
Of course, Kloppenborg here appears to be completely ignoring the fact that he elsewhere argues for the existence of a document Q containing the DT text, and that he therefore supports the following additional arrangement:
e) Mark-Q or Two-Source hypothesis: Mark -> Matthew and Mark - > Luke, and Q -> Matthew and Q -> Luke
This means that Kloppenborg supports the idea of both aMatthew and aLuke conflating Mark and Q even though he above states “that no known ancient author would have taken the trouble to compare sources so closely and to zigzag between them.” It also means that in the hypothesis that Kloppenborg himself favors both Mark and Q are ‘middle terms,’ and so it seems clear from the above that he has carefully selected just a limited subset of the possible arrangements, perhaps because his argument relies on there not being another possible source for Matthew and/or Luke. He continues:
A second set of data also points to the conclusion that Mark is medial. If we compare the sequence of episodes in each of the Gospels, an important pattern emerges. While Matthew sometimes relates an episode in a different sequence than Mark and Luke, and while Luke sometimes relates an episode in a different sequence than Mark and Matthew, Matthew and Luke never agree in locating an episode differently from Mark’s sequence… That is, even where both Matthew and Luke disagree with Mark’s sequence, they also disagree with each other.
These data reinforce the earlier conclusion: Mark is medial. In matters of sequence, we find Matthew agreeing with Mark’s order, and Luke agreeing with Mark’s order, but we never find Matthew and Luke agreeing to place a story where Mark has an entirely different placement. This datum suggests that there is no direct relationship between Matthew and Luke. Any of the possible arrangements of the three Gospels indicated in figure 1 [“no direct connection between Matthew and Luke”] might account for these data.
Kloppenborg does not define what he means by ‘medial,’ but from the way in which he continues it appears that he takes it to mean that Mark alone mediates between Matthew and Luke, i.e. that the latter two have no other connection (direct or indirect) between them, even though he has not shown that other arrangements cannot result in the differences he describes above. In addition, Kloppenborg’s last point above is odd, since although his "Straight Line" arrangements a) and b) have Mark as a ‘middle term,’ that is only so in the sense that Mark appears between Matthew and Luke in the diagram in Kloppenborg’s book, because in arrangement a) Mark uses Matthew and Luke uses Mark, while in arrangement b) Mark uses Luke and Matthew uses Mark. Neither arrangement is symmetrical with respect to Mark and so in arrangements a) and b) Mark cannot be medial. This can be easily seen by adding to Kloppenborg’s text above so as to read:
In matters of sequence, we find Matthew agreeing with Mark’s order, and Luke agreeing with Mark’s order, but we never find Matthew and Luke agreeing to place a story where Mark has an entirely different placement because in a) aMatthew did not know Mark’s order, and in b) aLuke did not know Mark’s order.
We are therefore left with only c) "Simple Branch" and d) "Conflation" as arrangements in which Mark is medial in the sense that Kloppenborg appears to be using it, i.e. in an arrangement in which Matthew and Luke are symmetrical with respect to their connections with Mark. Even so, this says nothing about connections between Matthew and Luke and other possible sources, or to each other.
As has already been noted above the symmetry seen in arrangements c) and d) can be maintained if Q is added, but crucially only if there is no connection between Mark and Q. With the ‘classic’ view of Q this can be the case, but as defined by the IQP (and hence as supported by Kloppenborg) Q does have connections with Mark, in the form of the Mark-Q overlaps, so breaking the symmetry because either Mark depends on Q or Q depends on Mark (Because Q is itself hypothetical the possibility of the connection being via a hypothetical source of both Mark and Q is not considered). There is also the important point that although on the 2DH both aMatthew and aLuke used Q, they clearly did so in different ways because the DT material (that in the Mark-Q hypothesis came only from Q) was used very differently by aMatthew and aLuke, as shown above by Kloppenborg.
As also noted above in the first half of Matthew and Luke the basis of the order of the text is the order of the DT, whereas in the second half of both gospels the basis is the order of the Triple Tradition text as seen in Mark. As the vast majority (around 98%) of the DT text occurs in the first halves of both Matthew and Luke it seems reasonable to say that the DT source dominates the first half of both, supplemented by Triple Tradition material, whereas in the second halves (with almost no DT material) the Triple Tradition source (here assumed to be Mark) dominates.
This implies that on the 2DH aMatthew and aLuke each independently chose to place almost all of their DT material (interspersed with relevant Markan material re-ordered to suit) in the first halves of their respective gospels, and the remaining Markan material second with aMatthew keeping the majority of the remaining Markan material in order. In the first halves of Matthew and Luke it is not Mark that is ‘medial,’ but the DT. The question is then whether this indicates that the DT text came from Matthew or another source.
The evidence presented above shows that the order of the DT text in the source used by aLuke largely follows the order we see in Luke, whether or not that source was Q. However, if that source was Matthew then we know it contained more than just the DT, and it is being increasingly acknowledged that Q could also contain more than just the DT. It is widely acknowledged that Q contained some parallels to Mark (known as Mark-Q overlaps), and it could also have contained non-Markan text some of which aMatthew ignored and some of which aLuke ignored, so being known to us as Sondergut Luke and Sondergut Matthew material respectively. Including this material has no effect on the order of the DT material in Q, but the more text is included the less justification there is for calling that second source Q. However, simply adding this material does not require changing the order of the DT material, and so the above arguments for the order of the DT text in Q can apply equally to other second sources containing more than just the DT, provided that the DT material largely follows the order of that material in Luke. Clearly that can apply to any second source hypothesized as an early version of Luke, whether or not aLuke also knew Matthew.
As noted in the introduction there is a synoptic hypothesis generally known as the 3SH (Originally proposed by H.J. Holtzmann and more recently supported by Robert H. Gundry) which specifies “that aLuke used Matthew as a secondary source of the DT material in addition to a source named Q,” and about which Peterson comments:
A yet more striking use of the argument [from double tradition order] is made by Robert Gundry, who abandons a fundamental postulate of the Two-Source Hypothesis, that Matthew and Luke were written independently of one another, but continues to maintain the existence of Q because of diverging order in the double tradition. Surveying the Minor Agreements of Matthew and Luke against Mark, Gundry finds more than three dozen details of narrative or wording that comport with Matthew’s Gospel but not with Luke’s. From these ‘Matthean foreign bodies’ lodged in the text of Luke Gundry concludes that Luke was acquainted with Matthew. Yet Gundry, inhibited by ‘the disarrangement of Matthean material’ that must be posited if Luke used Matthew as one of his ‘primary sources’ rather than as a secondary ‘overlay’, does not follow Austin Farrer and Michael Goulder in dispensing with Q. He accounts for his Matthaean foreign bodies by positing Luke’s use of Matthew but for the arrangement of double tradition in Luke by retaining the evangelists’ common use of Q. The argument from double tradition order thus constitutes Gundry’s sole reason for continuing to assert the existence of Q.
However, as noted above, Gundry’s ‘Q’ is, by definition, not Q as defined in the 2SH, because in the 3SH Matthew and Luke are not independent. It is however most certainly a ‘second source’ (SS) in a synoptic hypothesis in which aLuke also knows Matthew, and on May 29, 2012, in a blog post on his NT Blog, Mark Goodacre commented on musings on The Holtzmann-Gundry Solution to the Synoptic Problem (Three Source Hypothesis) by Michael F. Bird:
On a basic level, the difficulty with this hypothesis is that it concedes defeat on the key premise for the postulation of Q. Normally speaking, the existence of Q is predicated on the basis of arguments that Luke could not have known Matthew's Gospel. If Luke knows Matthew, then there is no need to explain the double tradition material on the grounds that they both independently accessed a hypothetical document.
The idea that the 3SH “concedes defeat” is only valid if you begin with the belief that the 2DH is all that is required to explain the DT. However, as shown above there are problems with the proposed actions required of either aMatthew or aLuke (depending on the hypothesis) when creating the DT by copying material from his non-Markan source (Q or Matthew respectively). Goodacre continues:
In support of the solution, Mike notes E. P. Sanders's prophecy in 1969:
I rather suspect that when and if a new view of the Synoptic problem becomes accepted, it will be more flexible and complicated than the tidy two-document hypothesis. With all due respect for scientific preference for the simpler view, the evidence seems to require a more complicated one.
It is perhaps worth bearing in mind, though, how Sanders's own view changed over the subsequent twenty years. In Studying the Synoptic Gospels, co-authored with Margaret Davies in 1989, he accepts Goulder's hypothesis that Luke knew Matthew as well as Mark, with the important modification that he does think Luke has other sources too, a modification with which I agree and for which I have argued too.
There is a general issue here too that the discussion of "simple" against "complex" can mask. Scholars of yesteryear were often reticent to think seriously about issues of memory and oral tradition in the way that they configured the problem. Gundry's half-way house between Farrer and the 2ST is symptomatic of this -- he is thinking in purely literary terms as a means of configuring his solution.
An important point here is with regard to the “other sources.” It is true that in many synoptic diagrams sources other than those directly involved in the proposed solution (e.g. Q and the synoptic gospels themselves) are not shown. For example, by definition, Sondergut Matthew and Sondergut Luke cannot have originated in a different synoptic gospel (although as noted above their text could have existed in a second source such as an expanded version of Q), and some hypotheses include sources containing that text named M and L respectively. In addition, sources for Mark are rarely indicated, even though unless Mark is pure fiction then aMark did have other information (written and/or oral) as his starting point.
Goodacre refers to the 3ST as a “half-way house between Farrer and the 2ST,” and while in some ways he is correct, it should more properly be seen as combining the best features of two hypotheses that individually suffer from exactly the same problem: that either aMatthew (2DH) or aLuke (MwQH – Farrer) have to perform almost incredible feats of juggling texts in order to arrive at the orders of the DT material that we see in their respective gospels. Goodacre again:
The main difficulty, though, with the Holtzmann-Simons-Gundry approach is that once Luke's knowledge of Matthew is (rightly) conceded, there is no need for Q. The high verbatim agreement between Matthew and Luke means that we are not dealing with later, secondary overlay, but direct copying by Luke of Matthew -- Q actually causes problems for making sense of that high verbatim agreement.
The issue of order is similar. Gundry appeals to Q in order to explain Luke's order. This approach works with the notion that an evangelist's order was largely dictated by source constraints, the kind of perspective that made sense in the 19th and early 20th centuries, but makes less sense now. Further, as I have often pointed out before, postulating a Q to explain Luke's order only throws the problem back to Matthew's order. The different ordering of the double tradition is a fact; at least one person, Matthew or Luke or both, has done some rearranging.
With his last point here Goodacre has hit the nail on the head. The rearranging is the problem, as depending on the hypothesis it has been performed by “Matthew or Luke or both,” and unless the hypothesis allows for rearranging by both Matthew and Luke (which neither the 2DH nor the MwQH do) then that hypothesis requires extraordinary literary contortions by either aMatthew or aLuke. It is perhaps this very issue that prevents neither the 2DH nor the MwQH from being regarded as decisively ‘better’ than the other – they both suffer from essentially the same significant failing, and very few people seem to recognize that allowing aMatthew and aLuke to have access to a common non-Markan second source in addition to allowing aLuke to know Matthew solves this problem by providing two separate re-ordering mechanisms during the creation of the DT.
Although in the quotes above from Kloppenborg he is specifically ‘Excavating Q’ his data and arguments also apply to any non-Markan second source of Matthew and Luke (SS, Ev) that may have contained more than just the DT. The difference is that if there was a second source and aLuke knew Matthew there would naturally have been two different possible rearrangements of the DT text. For example:
Someone edits Mark and adds to it new material, so creating a second source (SS), later used by aMatthew and aLuke;
aMatthew uses/reorders/edits text from both Mark and SS, including at least some non-Markan text from SS that becomes part of the DT as a result of later actions by aLuke .
aLuke uses/reorders/edits text from Mark and SS as above, but also includes text from Matthew that was not in either Mark or SS (i.e. was added by aMatthew), creating additional DT sequences that were not derived from SS.
These two possible sequential re-arrangements, the first by aMatthew, and the second by aLuke layered onto text some of which had already been re-arranged by aMatthew, or had been added by Matthew, may be the solution to the problems noted above by Kloppenborg and others, who were of course trying to make sense of the DT assuming re-arrangements by either aMatthew or aLuke but not both.
On the assumption that there was a non-Markan second source for Matthew and Luke and that aLuke used Matthew, there may have been agreements in order in the DT resulting from aMatthew and aLuke both selecting the same text from the second source and placing that text in the same order, but that aLuke later re-arranged when adding additional text from Matthew, including possibly more of the text from the second source that he saw was also in Matthew and that he had not previously used himself.
In addition, because on the 2DH the order of the DT in Luke is largely the same as the order of the same material in Q, then it is obvious that the same could apply to any other second source containing the DT. Consequently exactly the same arguments can be made regarding the order of the DT text in that second source, whether it is Q, an expanded version of Q (See What Exactly is Q?) or Ev). or any other source containing some or all of the DT.
R. A. Derrenbacker, R.A.: ‘Ancient Compositional Practices and the Synoptic Problem’ (Ph.D. diss.; Toronto: University of St. Michael’s College, 2001
Goodacre, Mark: The Case Against Q: Studies in Markan Priority and the Synoptic Problem – February 1, 2002, also Goulder and the Gospels: An Examination of a New Paradigm (The Library of New Testament Studies) 1st Edition – 1St Edition, The Library of New Testament Studies
Gundry, Robert H.: Matthew: A Commentary on His Handbook for a Mixed Church Under Persecution – 2nd Edition, 1994
Kloppenborg, John S.: ‘Excavating Q: The History and Setting of the Sayings Gospel’, 2000, also Variation in the Reproduction of the Double Tradition and an Oral Q?, The Formation of Q: Trajectories in Ancient Wisdom Collections , and Q, the Earliest Gospel: An Introduction to the Original Stories and Sayings of Jesus - Westminster John Knox Press, Oct 3, 2008
Peterson, Jeffrey: Order in the Double Tradition and the Existence of Q
Streeter, Burnett Hillman: The Four Gospels: A Study of Origins, Treating of the Manuscript Tradition, Sources, Authorship, & Dates, 1924
Tuckett, C.M.: The Existence of Q, in ‘The Gospel Behind the Gospels: Current Studies on Q’ edited by Ronald Allen Piper
In this table the verse numbers in the ‘Q’ column are the same as those in ‘Luke’ (per convention), while the ‘Expected’ Matthew column shows the locations (or at least the 'area') at which the Q parallels in Matthew that do not follow the same sequence as their parallels in Luke would have to be placed for them to be in the Lukan sequence. The verses highlighted below in blue are those noted above by Kloppenborg as “pericopae that Matthew conflated with or inserted into a Markan passage,” as none of them are in their ‘expected’ locations in Matthew, i.e. Matthew has taken these Q verses and moved them from their 'expected' location and places them somewhere else in his narrative.
Q Luke ‘Expected’ Mt Matthew
6:27-28 6:27-28 5:13-39a 5:44
6:31 6:31 5:43-45 7:12a
6:35d 6:35d 5:47-48 5:45a
6:39b 6:39b 7:2b-3 15:14b
6:40 6:40 7:2b-3 10:24a,25a
6:45a 6:45a 7:17-20 12:35
6:45b 6:45b 7:17-20 12:34b
9:57-60a 9:57-60a 11:20 8:18-22
10:2 10:2 c. 11:20 9:37-38
10:3 10:3 c. 11:20 10:16a
10:4 10:4 c. 11:2 10:9-10a
10:5-6 10:5-6 c. 11:20 10:12-13
10:7b 10:7b c. 11:20 10:10b
10:8 10:8 c. 11:20 10:11
10:9a 10:9a c. 11:20 10:8
10:9b 10:9b c. 11:20 10:7
10:10-11a 10:10-11a c. 11:20 10:14
10:12 10:12 c. 11:20 11:24
10:16a 10:16a 11:23b-24 10:40a
10:23b-24 10:23b-24 11:28-12:24 13:16-17
11:2a 11:2a 11:28-12:24 6:9a
11:2b (in D) 11:2b (in D) 11:28-12:24 6:7
11:2c 11:2c 11:28-12:24 6:9b
[11:2d] [11:2d] 11:28-12:24 6:10a
[11:2e] [11:2e] 11:28-12:24 6:10b
11:3-4a 11:3-4a 11:28-12:24 6:11-12
11:4b 11:4b 11:28-12:24 6:13a
11:9b-10 11:9b-10 11:28-12:24 7:7-8
11:11, 13 11:11, 13 11:28-12:24 7:9-11
11:14-15a 11:14-15a 11:28-12:24 9:32-34
11:16 11:16 11:28-12:24 12:38
11:24-26 11:24-26 11:31-38 12:43-45
11:33 11:33 12:43-13:30 5:15
11:34-35 11:34-35 12:43-13:30 6:22-23
11:39-41 11:39-41 12:43-13:30 23:25-26
11:42 11:42 12:43-13:30 23:23
11:43b-d 11:43b-d 12:43-13:30 23:7,6b,a
11:44 11:44 12:43-13:30 23:27-28
11:46b-c 11:46b-c 12:43-13:30 23:4a,c
11:47a-c, 48a 11:47a-c, 48a 12:43-13:30 23:29a,c,e, 31
11:49-51 11:49-51 12:43-13:30 23:34a,c,35-36
11:52 11:52 12:43-13:30 23:13
12:2 12:2 12:43-13:30 10:26b
12:3 12:3 12:43-13:30 10:27
12:4-5 12:4-5 12:43-13:30 10:28
12:6-7 12:6-7 12:43-13:30 10:29-31
12:8 12:8 12:43-13:30 10:32
[12:9] [12:9] 12:43-13:30 10:33
12:10 12:10 12:43-13:30 12:32a,b
12:10c [D] 12:10c [D] 12:43-13:30 12:32c
12:11a,b 12:11a,b 12:43-13:30 10:17b-18a
12:11c-12 12:11c-12 12:43-13:30 10:19
12:22b-28 12:22b-28 12:43-13:30 6:25-30
12:29-31a 12:29-31a 12:43-13:30 6:31-33a
12:31b 12:31b 12:43-13:30 6:33c
12:32 12:32 12:43-13:30 6:34
12:33-34 12:33-34 12:43-13:30 6:19-21
12:39 12:39 12:43-13:30 24:43
[12:40] [12:40] 12:43-13:30 24:44
12:42b-46 12:42b-46 12:43-13:30 24:45-51a
12:51-52 12:51-52 12:43-13:30 10:34, 36
12:53 12:53 12:43-13:30 10:35
12:54-55 12:54-55 12:43-13:30 16:2-3a
12:56 12:56 12:43-13:30 [16:3b]
12:58-59 12:58-59 12:43-13:30 5:26-26
13:24 13:24 13:34–18:11 7:13a, 14
13:26-27 13:26-27 13:34–18:11 7:22-23
13:28a 13:28a 13:34–18:11 8:12b
13:28b-29 13:28b-29 13:34–18:11 8:11b,12a,11a,c
13:30 13:30 13:34–18:11 20:16
13:34-35 13:34-35 13:34–18:11 23:37-39
14:5 14:5 13:34–18:11 12:11-12a
14:11 14:11 13:34–18:11 23:12
14:16b-24 14:16b-24 13:34–18:11 22:2b-3a, 5, 8-10
14:26 14:26 13:34–18:11 10:37
14:27 14:27 13:34–18:11 10:38
14:34-35 14:34-35 13:34–18:11 5:13
16:13 16:13 18:14 6:24
16:16 16:16 18:14 11:13, 12
16:17 16:17 18:14 5:18
16:18 16:18 18:14 5:32
17:1-2 17:1-2 18:14 18:7, 6da
17:6 17:6 18:23-24:25 17:20
17:33 17:33 24:39b-40 10:39
17:37b 17:37b 24:42-25:15 24:28
22:28-30 22:28-30 25:30-28:20 19:28
Note: in some places Kloppenborg is incorrect regarding the Markan passages that “Matthew conflated with or inserted” his own words into. For example, Kloppenborg has Mt 15:14 as the parallel to Q 6:40, with this text being “conflated with or inserted into” Mk 7:1-23. While this may be so for Q 6:39b // Mt 15:14b, it is not so for Q 6:40 // Mt 10:24b,25a.)