The synoptic problem.

The literary relationships between Matthew, Mark, and Luke.


The canonical gospels of Matthew, Mark, and Luke are the three synoptic gospels. They are called synoptic because they can with relative ease be arranged into the columns of a synopsis so as to be compared pericope by pericope,* phrase by phrase, even frequently word by word. The gospel of John, by contrast, does not so easily flow alongside the synoptic three. Arranging it in a fourth column beside Matthew, Mark, and Luke usually invites an inordinate amount of white space either in its own column or in those of the other three.

* A pericope is a passage, an individual narrative unit within a gospel. It might be a saying of Jesus, a series of related sayings of Jesus, a healing or exorcism, an encounter with an opponent, or any of a number of other kinds of accounts.

There appears to be, in other words, a more intimate set of relationships between Matthew, Mark, and Luke than between John and any or all of those. The synoptic problem, then, is simply the question of the exact relationship of Matthew, Mark, and Luke with one other. How did those three gospels come to so closely resemble each other in ways not shared with John (or, for that matter, any of the noncanonical gospels)?

The literary nature of the problem.

Is there a synoptic problem? Before one considers the many and varied potential solutions to the synoptic problem one must first demonstrate that there is indeed such a problem. In that connection it is necessary to understand that the synoptic problem proper has to do with the literary relationships of Matthew, Mark, and Luke. If it turned out that all three were literarily independent of one another, their many similarities stemming from, say, a controlled oral tradition rather than one author borrowing from the text of another, then there would not actually be a synoptic problem. The problem of purely oral transmission of the Matthean, Marcan, and Lucan materials would form its own problem, to be sure, for students of those texts, but it would be a very different sort of problem from what we like to think of as the synoptic problem.

I have dedicated a separate discussion to the question of oral transmission and literary relationships as pertains to the synoptic gospels. Suffice it to say at this juncture that I for one am quite convinced that it will take a set of literary relationships to satisfy the data of the synoptic gospels. I am not at all against invoking oral tradition or other factors in specific cases, but the general scenario will require that Matthew, Mark, and Luke share literary connections beyond those instances, and that, moreover, there is some kind of literary connection between each pair of synoptic gospels, whether direct or indirect, mediated through hypothetical documents.

Terms and symbols.

Like other fields of inquiry, the synoptic problem comes with its own terminology and set of symbols. The most important for the present discussion are laid out below:

Term or symbol. Definition.
M 1. Material proper to Matthew, not shared with either Mark or Luke.
2. The Matthean version of any gospel material.
K 1. Material proper to Mark, not shared with either Matthew or Luke.
2. The Marcan version of any gospel material.
L 1. Material proper to Luke, not shared with either Matthew or Mark.
2. The Lucan version of any gospel material.
Q Symbol for the double tradition, often implying a distinct source copied by Matthew and Luke.
double tradition Material common to Matthew and Luke, but not Mark.
triple tradition Material common to all three of the synoptic gospels.
sondergut Material proper to only one of the synoptics, whether M, K, or L.

Hypotheses toward a solution to the problem.

I regard it as well established that there is indeed a literary relationship between Matthew and Mark, between Matthew and Luke, and between Mark and Luke. Each of these individual relationships constitutes at least one hypothesis (more if the relationship is mediated through a hypothetical source; see below). I basically accept the system of synoptic nomenclature proposed by S. C. Carlson in which it takes a set of (specific) hypotheses to form a (general) theory.

For my purposes, each synoptic hypothesis proposes a distinct literary relationship either (A) between any two of the synoptic gospels or (B) between any one of the synoptic gospels and any source, whether extant or nonextant, that serves as a connection between at least two of the three.

Option A requires very little explanation. There are only six basic relationships possible amongst any set of three texts. In the case of the synoptic problem those six direct relationships come out as the following list (call it list α):

List α.
  1. Mark copied from Matthew.
  2. Luke copied from Matthew.
  3. Matthew copied from Mark.
  4. Luke copied from Mark.
  5. Matthew copied from Luke.
  6. Mark copied from Luke.

Option B, however, is a bit more complicated. I worded it in such a way as to eliminate from central consideration any text, whether extant or nonextant, which does not form a link between our synoptic gospels. Any text, for example, thought to have served as a source for only one of our synoptic gospels1 is not going to help us make any decisions about relationships between each set of two synoptic gospels. Such a source will simply contribute to the special material, or sondergut, of the gospel that copied from it. Likewise, any text thought to have copied from one, two, or even all three of the synoptic gospels, but which did not serve as source for any of them,2 is going to be of no help in making the necessary literary decisions. Both of these kinds of texts will be peripheral to any proposed solution of the synoptic problem.

1 Two examples of such a text would be the M and L documents proposed by B. H. Streeter in The Four Gospels. It does not matter to the synoptic problem proper whether Matthew and Luke (or Mark, for that matter) got their respective special materials from a written document or from an oral tradition, or even created them ex nihilo. (Of course, Streeter was aiming at much more than a solution to the synoptic problem.)
2 The church fathers provide numerous examples of such texts. They frequently copy from all three synoptic gospels, but serve as the source for none of them. It does not matter to the synoptic problem proper how many subsequent authors copied from Matthew, Mark, and Luke. The patristic texts may be invaluable to the reconstruction of the respective texts of the gospels, but their own relationships to the gospels do not constitute individual synoptic hypotheses.

The kind of source on the table in option B, then, is one which either was copied by two or more of the synoptic gospels or both copied at least one and was copied by at least one more. There are, therefore, two distinct lists of relationships under option B (call them lists β and γ, and our source, whether extant or nonextant, will be X):

List β.
  1. Matthew copied from source X.
  2. Mark copied from source X.
  3. Luke copied from source X.
List γ.
  1. Source X copied from Matthew.
  2. Source X copied from Mark.
  3. Source X copied from Luke.

Any source that is worked into a synoptic theory will necessarily entail, on the principles outlined above, at least two hypotheses (because we have established that any source that relates to only one of the gospels lies outside the synoptic problem proper). On those same principles, at least one of those hypotheses must come from list β, and it is possible that both or all of these hypotheses will come from that list. What is not possible is that both or all of these hypotheses entailing an extra source should come from list γ, with none coming from list β.

The hypotheses above are all positive; that is, they all describe which author did indeed copy from which other. But it is also possible to construct negative hypotheses. One can state categorically that a certain synoptic gospel did not copy from a certain other. Indeed, such a statement is one way, though by no means the only way, of introducing an extra source into the theory;* if each synoptic gospel is literarily related to each of the other two, and if any two of them did not copy from each other, then there must be another source lurking behind or between those two gospels.

* The most famous example of this procedure is the two-source theory, at least on those conceptions of it in which the attempt is made to demonstrate that Luke and Matthew did not copy from one another; therefore they must owe their close similarities to an extra source (Q).

Such hypotheses of what could not have transpired can be very important in synoptic studies. However, I do not feel that they belong in the list of basic hypotheses required for a viable theory. Rather, they belong on the level of methodology, to be used as arguments for the relevance or existence of extra sources in the course of pressing a point such as (from list β) one synoptic gospel copied from source X or (from list γ) source X copied from one of the synoptic gospels. I do not think ill, though, of including a negative hypothesis in parentheses, particularly if it is the principal reason for bringing in an extra source.

Similarly, the very act of hypothesizing an extra source does not itself constitute a synoptic hypothesis.* I am not certain how one could demonstrate either the relevance of an extant source or especially the existence of a nonextant source without closely attending to the literary phenomena that are leading one to hypothesize an extra source in the first place. It ought to be the internal phenomena themselves that are pointing to a source outside the synoptic three. If there happens to be external evidence for the relevance or existence of such a source, all the better, but a statement such as source X is relevant or source X existed is not a synoptic hypothesis of its own. It is merely part of an actual synoptic hypothesis such as source X copied from one of the synoptics or one of the synoptics copied from source X.

* This decision is not designed to reduce the number of steps in theories involving hypothetical sources so as to make them look more streamlined or parsimonious than they really are. To the contrary, the relative complexity of such theories is highlighted in my system by the fact that each individual literary relationship receives its own hypothesis. Sticking to the three synoptics and no more will require only three hypotheses every time. Adding one extra source into the mix, if one recalls that each such source must be connected to at least two of the synoptic gospels, will necessarily require at least four hypotheses.

And one is always free, of course, to bring in more than one extra source. One could have sources X and Y, or sources X, Y, and Z, or even more.* It will probably be the case, however, that the more sources one finds necessary to explain synoptic relations, especially nonextant sources, the less probable others will tend to regard the resulting theory.

* Adding two extra sources will require at least five hypotheses, adding three will require at least six, and so on.

One may somewhat similarly hypothesize different versions of the same source, such as X1 and X2 or what have you. On the one hand, it is probably true that no two manuscripts of the same text were ever identical in antiquity; on the other hand, of course, the more a theory depends on knowing the nature of the variants between the versions, the less probable others will tend to regard the resulting theory.

Finally, one might wish to mitigate the hypothesized literary copying in some cases by taking recourse to oral tradition, borrowing by memory, or other contingencies. Any such considerations properly belong to the discussion of each hypothesized literary relationship. To explain one or some elements of the synoptic problem in light of oral tradition or the like is not at all the same as explaining all synoptic relations in that way (in which case one is saying that there is no synoptic problem).

Making allowances both for extra sources (such as X, Y, and Z) and for different versions of the same source (such as X1 and X2), and allowing that literary influence may be mitigated by oral tradition or borrowing by memory or the like, we see clearly that each of the hypotheses that we advance toward a viable and plausible synoptic theory will come (in one form or another) from the twelve options listed on the three lists above (α, β, and γ).

Theories toward a solution to the problem.

Any given hypothesis of the kind that I have been describing is only part of the whole picture. The combination of hypotheses sufficient to connect each of the three synoptic gospels to the other two comprises a synoptic theory. If it is correct that each synoptic gospel relates in a literary way to each of the other two synoptics, then a viable synoptic theory will require a combination of at least three hypotheses:

  • An hypothesis to relate Matthew to Mark.
  • An hypothesis to relate Matthew to Luke.
  • An hypothesis to relate Mark to Luke.

More hypotheses will be required as extra sources are added to account for the literary phenomena.

Synoptic theories are often named eponymously: Augustine, Greisbach, Farrer, Wilke, Büsching, Lockton... famous names representing particular sets of hypotheses. Some theories are named descriptively, such as the two-source and three-source theories. Others are named in yet other ways, such as that of the Jerusalem School.

I am not all that fond of most of these ways of naming the synoptic theories. I have largely accepted the system of synoptic nomenclature described by Stephen C. Carlson, who proposes to name each theory after the documents that it uses to account for the triple and double traditions, in that order. On my page detailing this novel and promising synoptic nomenclature I point out a couple of its weaknesses (at least one of which I remedy by adding a slash / to the hyphen - that Carlson uses to separate the document responsible for the triple tradition from that responsible for the double; the slash accomodates two or more documents responsible for the same tradition), but also its strengths, including but not limited to its apparent synoptic neutrality.*

* I might point out that identifying the double tradition as that which Matthew and Luke have in common against Mark is not purely neutral. Matthew and Mark also share material in common against Luke, and Mark and Luke share material in common against Matthew; are these not double traditions too? Yet scholarly history has decided, and not entirely without cause, that the double tradition of Matthew and Luke against Mark is of greater importance for solving the synoptic problem than the other two double traditions, if you will. No one to my knowledge has proposed that Mark and either Matthew or Luke are independent of each other, thus requiring a special text, tradition, or set of texts or traditions to account for what Mark has in common with either Matthew or Luke. The actual double tradition shared by Matthew and Luke against Mark appears to have a genius of its own, consisting almost entirely of sayings of Jesus and presenting certain repeated themes and motifs.

So, despite its perhaps inevitable shortcomings, I happily use a variant of the system that Carlson proposes. For the triangular theories* naming the theory amounts to naming the first and second documents supposedly written, in that order, separated by a hyphen.

* Id est, those six viable theories that require only the three synoptic gospels themselves, with no recourse to hypothetical sources. Their diagrams look like triangles; hence the name.

The six viable triangular theories are as follows, broken down into their necessary hypotheses:

Matthew-Mark (Augustine).
  • Mark copied from Matthew.
  • Luke copied from Matthew.
  • Luke copied from Mark.
Matthew-Luke (Greisbach).
  • Luke copied from Matthew.
  • Mark copied from Matthew.
  • Mark copied from Luke.
Mark-Matthew (Farrer).
  • Matthew copied from Mark.
  • Luke copied from Mark.
  • Luke copied from Matthew.
Mark-Luke (Wilke).
  • Luke copied from Mark.
  • Matthew copied from Mark.
  • Matthew copied from Luke.
Luke-Matthew (Büsching).
  • Matthew copied from Luke.
  • Mark copied from Luke.
  • Mark copied from Matthew.
Luke-Mark (Lockton).
  • Mark copied from Luke.
  • Matthew copied from Luke.
  • Matthew copied from Mark.

There are too many nontriangular theories to list here, but Carlson offers a very extensive list on his Synoptic Problem Website, as well as an enumeration of 1488 viable synoptic theories. I will list only three, again broken down into their constituent hypotheses:

Mark-Q (two-source).
  • Matthew copied from Mark.
  • Luke copied from Mark.
  • Matthew copied from Q.
    (Matthew did not copy from Luke.)
  • Luke copied from Q.
    (Luke did not copy from Matthew.)
Mark-Q/Matthew (three-source).
  • Matthew copied from Mark.
  • Luke copied from Mark.
  • Matthew copied from Q.
  • Luke copied from Q.
  • Luke copied from Matthew.
Mark-Q/Luke (Hengel).
  • Luke copied from Mark.
  • Matthew copied from Mark.
  • Luke copied from Q.
  • Matthew copied from Q.
  • Matthew copied from Luke.

It ought to be fairly clear from those lists how the various hypotheses come together to form a theory. It ought also to be clear which are the simpler and which the more complex theories based on how many hypotheses each has to employ in order to reach viability. The triangular theories are the simplest available. The nontriangular theories add various degrees of complexity.

One might ask, why add complexity at all? Why not stick with the simpler theories? The answer lies with the literary phenomena. When we view the problem in the abstract, there is no reason to prefer a more complex solution over a simpler solution. It is only when we begin to apply concrete literary methods to the texts themselves that the possibility of a nontriangular theory being the most probable begins to arise.

The matter of methodology.

One decides between synoptic theories by deciding which has the most probable set of synoptic hypotheses, but how does one decide which hypotheses are more or less probable? To ask how is to ask about methodology.

The methodological key to determining which text copied from which other text is directionality. That is, one must find indicators that tell us the direction in which the copying took place.

These indicators must of course be unidirectional, or irreversible. For example, pointing out that two of the synoptic gospels share the same passage verbatim does not prove which text copied the other. Nor does explaining what one author had to do to the other text in order to arrive at his own. Doubtless someone else could explain how the opposite process could have happened. Explanations tend to be bidirectional, or reversible.

Any criterion used, therefore, ought to have directionality built into it. What follows is an admittedly incomplete list of criteria that have been tried before:

  • Christology. The text with the higher Christology probably copied from the text with the lower Christology.
  • Theology. The text with the more orthodox theology probably copied from the text with the less orthodox theology.
  • Semitisms. The text with fewer Semitisms probably copied from the text with more Semitisms.
  • Jewishness. The text that is less Jewish probably copied from the text that is more Jewish.
  • Grammar. The text with the better grammar probably copied from the text with the worse grammar.
  • Length. The longer text probably copied from the shorter text.
  • Detail. The most detailed text probably copied from the least detailed.

In my judgment, none of these criteria is free of problems. Each is, or at least can be, bidirectional, or reversible. To wit:

  • Christology. What if the author of the text with the lower Christology is actually correcting the higher Christology of the other text?
  • Theology. What if the author of the less orthodox text is actually correcting the theology of the more orthodox text?
  • Semitisms. What if the author of the text with more Semitisms is Semitic himself and finds that Semitisms come naturally, while the author of the other text is not?
  • Jewishness. What if the author of the more Jewish text is writing for a Jewish readership, while the author of the other text is not?
  • Grammar. What if the author of the text with the poorer grammar wishes to sound picturesque or rustic?
  • Length. What if the author of the shorter text is writing an epitome, or abbreviation, of the longer text?
  • Detail. What if the author of the less detailed text felt that the details distracted from the main point?

Notice how important the author is in this listed counterpoint to the criteria. Not one of those criteria takes into account what the author himself may have intended by his piece of writing. The author, I think, is the key to directionality. We must ask ourselves what we think the author would or would not have done with this or that text in front of him. We are reduced, in other words, to editorial or redactional arguments. (An editor, or redactor, is what an author becomes while dealing with sources.)

Redactional arguments for unidirectionality presume, of course, that the author was writing with some purpose or purposes in mind, and that we are capable of discerning those purposes by closely reading his text. They also presume that an author was striving for some measure of clarity, not writing in an arbitrary or ciphered manner. And they allow that the author might have had his own set of habits or quirks. To get hold of these authorial characteristics is to at least hypothetically be able to determine what he would or would not do with a given text in front of him. And to apply such an insight to both authors of the pair of texts in question might well inform us which procedure is the more likely, that this author copied from that author or vice versa.

A redactional argument is not a panacea, at least not in the simplistic way in which I think those other criteria have at times been used. One cannot simply take a one-size-fits-all criterion, apply it mechanically to the textual evidence, and wait for it to spit out a yea or nay response. One must instead painstakingly look at the textual evidence from both possible perspectives (id est, this author copied that one or vice versa), looking for the plausibility of one scenario to contrast with the implausibility of the other. In most cases such a contrast will not exist. Neither author will have left a clue as to his source or sources. But sometimes one or both of the authors will let a clue slip by as he writes; it is up to us to pick up on it and correctly interpret it.

I have elsewhere divided this kind of argument into two categories...:

  • Redactional tendency.
  • Editorial fatigue.

...which are then combined into a single working principle which I have dubbed (perhaps somewhat cumbersomely) the principle of the real and the ideal. That kind of combination, I think, is the best argument available for deciding which text or tradition copied the other.

However, I also think that either of the two above kinds of argument can work on its own. For both redactional tendency and editorial fatigue take into account what the respective authors are doing or trying to do in their respective texts, a step that is absolutely necessary to figuring out which is working with existing material and which is perhaps composing freely.

I take a closer look at redactional tendency and editorial fatigue on another page.

The synoptic project.
Inventory of the synoptic gospels.
Synoptic nomenclature.
The four-color synopsis.
Synoptic code-switching.
Major and minor agreements.
Raw list of agreements.
Synoptic vocabulary.
Repeated and transposed formulae.
Synoptic doublets.
Synoptic intercalations.
Synoptic mixed parallels.
Synoptic summary statements.
Matthean discourse finales.
Marcan chiasms.
Marcan endings.
Lucan parallels to Olivet.
Oral and literary tradition in the synoptics.
Redactional tendency and editorial fatigue.
Gospel manuscripts.
Textual variants.

The Synoptic Problem Website, by S. C. Carlson.
The Synoptic Problem and Q, by Mark Goodacre.
A Synoptic Gospels Primer, by Mahlon H. Smith.
A Website for the Two Gospel Hypothesis, by Thomas R. W. Longstaff.
The Logia Translation Hypothesis, by Brian E. Wilson (a development of the Greek Notes Hypothesis).
The Three Source Hypothesis, by Ron Price.
The Existence of Q and The Priority of Mark, by Peter Kirby.
The Synoptic Problem, by John McVay.
The Synoptic Problem, by Daniel B. Wallace.
The Present State of the Synoptic Problem, by William R. Farmer.
Goulder and Q, by Allan J. McNicol.
A Statistical Approach, by Dave Gentile.
The Modified Augustinian Hypothesis, by Jim Deardorff.