A Gateway to the Research of the Jesus Seminar

 [Home] [About Site] [Complete Gospels] [Data Base] [Historical Quest] [Westar Institute
[Profiles] [Publications] [Reaction] [ Search ] [What's New?] [Weblog] [Network]


Mahlon H. Smith

This essay was prepared for a session of the Westar Institute focused on the question of what texts should be published in a modern scholars' canon of early Christian sources.

1.1. Clarification of the Issue

(a) A matter of definition.

To a student of the gospels who has worked for 40 some years with the hypothesis that the authors of Matthew and Luke derived the bulk of their material from two written 

sources---the gospel of Mark and a sayings collection of unknown title that modern scholars conventionally call "Q" ---the question of whether Q belongs in the canon of Christian scripture is not a moot issue.

  • If Matthew, Mark and Luke are "canonical" texts, and
  • if by definition "Q" indicates the source of the material common to Matthew and Luke that is not based on Mark,
  • then Q is, also by definition, already essentially in the canon.

In expressly endorsing a fourfold gospel, the greater church accepted what later scholars identified as "Q" material into its canon of scripture. 

(b) The consequence of censorship.

The only way now to exclude Q from the New Testament canon would be to expurgate the gospels of Matthew and Luke of all parallel passages that have no equivalent in Mark.

This, in effect, would eliminate from the church’s scriptures more than half of the lines ascribed to Jesus in the synoptic gospels, including 54 of the 90 sayings that the majority of the Jesus Seminar has agreed derive from things that Jesus actually said. While this would not totally silence the voice of Jesus it would severely muffle it behind the din of ideologies that early Christians imposed upon him ---a situation that only the most dogmatic apocalypticist would welcome.

To contemplate a canon without Q, in effect, confines all but the shadow of Jesus of Nazareth to the grave. For very little of his own distinctive outlook would have made enough of an impression on his admirers to have survived him. If that were the case, there would be good cause to doubt the historicity of claims that this particular Galilean, who was executed ca. 30 CE, was raised from the dead. While not every word of Q can be traced to Jesus himself, it is this corpus of sayings that, more than anything else, challenges people it confronts to consider the author of its extraordinary perspective on life and social interaction to be still a vital force in history.

(c) Refocusing the issue.

The question of Q’s canonical status needs to be refined since there is no dispute between between proponents and opponents of the Q hypothesis on two points:

(1) "Q" refers to material that is already canonical and 

(2) this material is such important evidence of Jesus that its exclusion from a Christian canon is not an option. 

The only issues that still need to be resolved to recognize Q as a legitimate canonical source are:

  • whether the Q hypothesis is dispensable or not;
  • whether a reliable text of Q can be isolated from material in Matthew and Luke, and
  • whether a reconstructed text can be treated as a canonical source.

(d) A matter of urgency

The recent publication of several reconstructions of Q has brought these questions to the center of debate on the development of the Jesus tradition./1/  

The first salvos in a war of myths over Christian origins have already been fired./2/ The 1996 release of the first fascicle of Documenta Q --- the monumental record of the result of years of collegial Q scholarship --- insures that the question of Q’s status will be kept alive in the public press for the next fifteen years or so./3/

The future of Q will in effect be decided by the current public debate. Since the question of Q’s canonical status has shifted from classrooms and scholarly journals to the marketplace, it can no longer be viewed as just an academic theory that is a matter of private opinion. It is as much a scientific description of phenomena that confront average people every day as Galileo’s solar system, Newton’s gravity, Darwin’s evolution or Einstein’s relativity. Whether the Q hypothesis, like these, comes to be generally accepted as the most plausible explanation of observable facts or is rejected as arcane speculation depends on public demonstration

  • that it is generally accepted by most experts, and
  • that it works better than other theories.

This task cannot be postponed, since the fate of objective biblical scholarship is at stake.

(e) The issue of authority.

Since Westar was founded as a public forum of scholars dedicated to promoting religious literacy, it has a constitutional responsibility to provide informed leadership on this issue.  

Our responsibility is  all the greater since many of the most visible proponents of Q were among the earliest active Fellows of the Jesus Seminar. Critics of the Seminar have tried to discredit its conclusions by claiming they are based on an unprovable fiction called Q. Silence or indecision on the Q question, therefore, would only convince a poorly informed public that this allegation is correct.

In making pronouncements about the status of Q, however, it must be clear that Westar is acting as a democratic forum of well-informed research to promote an increased level of public clarity on Christian origins, rather than as a self-appointed magisterium dictating what hypotheses scholars must teach in the classroom or what texts are to be read in church. As educators, we treasure freedom of debate as much as anybody. We would not accept a syllabus set by somebody whose authority we did not grant. Yet, as educators, we know the benefit of bibliographies produced by expert scholars and we ourselves regularly provide our students with required reading lists. The variety of canons of sacred scripture that have been published throughout Christian history are just such lists. The task that faces us here is to identify the primary ancient sources that we think are essential to know for a clear understanding of the historical origins of Christianity. Should Q be part of a scholars' canon?

Westar's only authority rests on public perception of a scholarly consensus reached through an open debate that allows all opinions a fair hearing. As in the voting on Jesus’ sayings and deeds, a Westar decision on Q’s place in the canon cannot prevent any Fellow from expressing personal insights that differ from the majority position, or from challenging particular reconstructions of Q and interpretations of the social history that generated it. But a series of votes to test our current consensus would clarify the substantive issues and could advance the canonization of Q in the public mind.

1.2. Formulation of the argument.

(a) Type of question.

Assessing the status of Q is quite different than deciding the fate of the gospels of Thomas or Peter or Egerton. In these cases there was discovery of a physical document that had not previously been considered canonical. 

These texts were apparently regarded as marginal by the Christian majority in antiquity, ceased to be copied and, thus, influenced no one until a few surviving fragments were unearthed in recent years. Ironically, it is relatively easy to convince the modern public of the importance of this type of material evidence, witness the almost universal acceptance of the explicitly esoteric Dead Sea Scrolls in current reconstructions of the social history of the second Jewish Commonwealth.

The status of Q is, rather, a matter of the state of scientific excavation of texts that have long been canonized. Here public recognition of progress in interpreting the facts is bound to be much slower. For there is an inherent intellectual inertia that resists altering customary patterns of viewing and handling familiar tools, which is magnified when the tools are considered sacred rather than utilitarian (witness: the reluctance of many to accept any English translation of the Bible other than the King James Version).

Here, as in the case of Darwin’s theory of evolution, the canonicity of a discovery needs to be measured by its ability to convince a smaller sampling of specialists and educated laity rather than by the popular opinions of those who fail to do their homework. It took 138 years for a pope to recognize that there is enough evidence to show that evolution is "more than just a hypothesis." Since pious opinions about sacred scripture are often confused with infallible dogma, it will take somewhat longer for the average Christian to come to the same conclusion about Q. There will always be those who regard the earth as flat on the basis of casual personal observation. When closer scientific examination with more precise instruments establishes evidence that it is not, the prior opinion cannot be considered authoritative, no matter how ancient or widespread or obvious it may seem.

(b) Scientific method

Though the question of what is canonical for well-informed modern scriptural research cannot be resolved by the criterion of universal consensus, a second criterion invoked to set the boundaries of the New Testament ---antiquity- -- can.  

As in any scientific investigation, here it is the age of the material itself rather than the theory that is important. The antiquity of any matter is not established by common opinion or its current superficial appearance but by repeatable tests that indicate its probable position relative to other datable objects. The naked eye of an untrained reader probably will not immediately discern the residue of an ancient collection of Jesus sayings (Q) in a casual reading of Matthew and Luke; but that does not mean it is not there. If tests demonstrating the relationship of the synoptic texts can continue to convince skeptics that two authors had independent access to a collection of sayings that was not utilized by the third, then the existence of Q can be held to have been scientifically established, regardless of what people who disregard those tests believe.

(c) The burden of proof.  

In cases of scientific judgment there is always room for re-analysis and adjustment of conclusions on the basis of more comprehensive or more refined tests.

But in research the demonstrable results of one set of tests are generally accepted until further investigation proves that they are flawed.

In any area of scientific investigation, a hypothesis does not have to be established beyond all question to be considered the norm. It merely has to prove more workable than other available options. Once a theory is established as the dominant working hypothesis among experts the burden of proof shifts to those who challenge it. Thus, if Q has become the normative theory among trained synoptic specialists -- as claimed by proponents and critics alike ---, then it has already been canonized and it is the dissenters who are in the position of challenging the current canon of scholarship./4/

(d) Dialectical format.

Given the regular and frequent reference to Q in standard NT textbooks, professional journals, monographs on the gospels and international academic forums it is easy to declare the canonization of Q a fait accompli.

But that would neither dispense with the objections of skeptics nor settle the issue of whether a current reconstruction of the Q text could claim canonical status. Fortunately, since it is plain that the contents of Q are canonical and it has been an established element of the dominant working hypothesis among synoptic scholars for more than a century, it is not necessary to rehearse the case for the existence of Q here nor to dispense with every conceivable objection. All one has to do is demonstrate that skeptics and critics of the Q hypothesis have not yet presented a strong enough case to eliminate Q from the scholarly canon.

To respond to challenges to a long established academic doctrine like Q, the dialectical format of the medieval scholastic disputed question is more efficient than the modern pattern of arguing a thesis. Instead of constructing an intellectual position that invites attack, the quaestio format effectively puts skeptics on the defensive by opening with their line of objections and then exposing the weakness of each. One refinement on the traditional quaestio is in order here, however. Rather than counter-balance lists of objections and rebuttals, objections will be dispensed with one by one, somewhat like the catalog of antitheses in Matthew 5. This method is adopted here with a bit of deliberate historical irony, since J. J. Griesbach used the antithetical scheme of objection/reply in his Commentatio of 1790 to respond to critics of his suggestion that the synoptic problem could be solved by what was then an almost totally novel hypothesis: that Mark had edited Matthew and Luke./5/


2.1. External evidence. 

(a) Objection


Critics argue that Q cannot be considered canonical since at least two other hypotheses --- the Augustinian and Griesbach --- are older and better supported by early external testimony about the composition of the gospels./6/

(b) Reply

While the theories of Augustine and Griesbach were proposed prior to the Two Source hypothesis, their authors accepted without question the patristic tradition that Matthew, as eyewitness testimony, was written first. 

Griesbach dismissed the then novel (1786) insight of his student Gottlob Christian Storr that Matthew and Luke had edited Mark with this argument: "Now it is inconceivable that Matthew, an eyewitness (testem oculatum), chose as his guide for handing on the story of Christ a writer who had not been present at the events themselves."/7/  Griesbach, however, never bothered to show that the extant Greek gospel of Matthew actually represented "eye-witness" testimony of an author who had personally seen and heard Jesus.  He simply echoed a tradition traceable to Greek Christian writers of the 2nd c. CE. Such an uncritical blind spot was uncharacteristic of this historically sophisticated scholar who just a few paragraphs earlier claimed the unanimous patristic witness of Papias, Justin, Irenaeus, Tertullian, Clement of Alexandria, Origen, Jerome, etc. that Mark transcribed the words of Peter -- not Matthew -- "has to be a guess." (habendas esse coniecturis)./8/

Well before the end of the nineteenth century, the theories of Augustine and Griesbach had been abandoned by most western synoptic scholars in favor of Markan priority precisely because Matthean priority did not account for the internal evidence of the gospels as well. J. J. Griesbach’s views were resurrected in revised form only in 1964 in Wm. Farmer’s critique of Markan priority. The Augustinian position, while still favored by evangelical bible schools and other conservatives, has had little support among critically trained scholars since 1783 when Griesbach effectively created the synoptic problem by pointing out that the traditional position did not adequately account for Mark’s relation to Luke./9/ Griesbach himself did not pretend to justify his own hypothesis on the basis of patristic tradition. But later disciples have sought to justify his alternate "solution" against the ascendant Markan hypothesis with an even less critical acceptance of the opinion of Clement of Alexandria that the gospels with genealogies were written first./10/ Current support for the Griesbach or Augustinian hypotheses comes primarily from scholars concerned to validate patristic testimony as reliable historical reminiscence and not --- as Griesbach himself held --- pious speculations "which are today rejected by most scholars."/11/

Yet, the antiquity and reliability of these patristic reports need to be proven before they can be accepted as historical evidence of anything more than the beliefs of those who reported them./12/ Second-hand testimony, which is weak historical evidence at best, is diminished when found only in documents composed a century or more after the events reported, and is almost totally worthless when it can be easily falsified by primary texts --- as is the case here.

(1) Was Luke composed second?

Far from being an ancient tradition, the notion that the gospels with genealogies and birth accounts were composed before those without is an isolated statement in a single text that had no widespread currency before the late eighteenth century. 

The "testimony" credited to Clement of Alexandria (before 215 CE) is actually an indirect allusion by Eusebius of Caesarea (ca. 350 CE) to a passage in a no longer extant work that is alleged --- by whom is uncertain --- to be "a tradition of the original elders." /13/ Which elders at the origin of what is not specified. Also, Eusebius’ language leaves the nature of this "insertion" unclear (quotation from a lost source? parenthetical aside? popular rumor?). Even if one assumes

  • that Eusebius was himself accurately citing a written text and
  • that its ascription to Clement was authentic (neither of which can be proven, of course),

there is, in fact, no external evidence that this off-hand allegation of the priority of Matthew and Luke was actually known by anybody before Clement wrote it or endorsed by any ancient author after he did. Moreover, since this passage does not specify the order of composition of Matthew and Luke and expressly claims that Mark copied the words of Peter rather than either of the other synoptics, much less both, it hardly supports the the weight of the Griesbach-Farmer hypothesis./14/

Griesbach was the first to claim explicitly that Luke wrote second. But this sequence is undermined by Luke’s own assertion (1:1) that "many" (polloi) had written before him, unless one assumes that Luke knew at least two sources that are now lost. Since Luke’s testimony is internal to the gospels it must be accorded historical priority over Clement’s isolated opinion, which apparently no one except Eusebius valued until Griesbach’s disciples resurrected it more than fifteen centuries later.

(2) Was Matthew composed first?

Testimony that Matthew was written first is more widespread in patristic sources and was certainly well-known by the third century. Yet records of this claim are neither as universal nor as ancient or clear as generally presumed.

The only patristic voice before 180 CE to record a claim of Matthean priority was Papias of Hieropolis (d. ca. 138 CE) --- according to Eusebius of Caesarea, that is, who quotes (ca. 350 CE) this vague one-line assertion from Papias’ no longer extant five volume exegesis of the sayings (logia) of Jesus: "So Matthew arranged the sayings (logia) in the Hebrew dialect, and each translated (or interpreted) them as he could."/15/ Papias did not claim that Matthew wrote any story of Jesus’ birth, baptism, miracles, passion or resurrection; on the contrary, his choice of words describes the act of compiling a sayings collection like Q rather than composing a narrative gospel like canonical Matthew.

As far as we can tell, Irenaeus of Lyons (ca. 190) was the first to equate Papias' testimony with a "gospel" of Matthew./16/ Yet the nature of this "gospel" is uncertain, since (a) it is presented as a parallel to the oral "preaching" of Peter and Paul and (b) the contents are left unspecified. In antiquity a written "gospel" could be anything from a sayings collection (like Thomas) to a compilation of independent anecdotes (like Signs) to a connected biographical narrative (like Luke) to a theological treatise (like the Valentinian "Gospel of Truth"). So it cannot be taken for granted that the "gospel" Irenaeus refers to was coextensive with the contents of canonical Matthew.

Origen’s Commentary on Matthew (ca. 250) is the first patristic text that clearly identifies this primitive Matthean composition with the contents of the canonical gospel of the same name by identifying the author as a toll collector./17/ Yet neither these authors nor Eusebius equate this reputedly first apostolic composition with the extant Greek text of canonical Matthew, since all of them insist that that author wrote in a Hebraic dialect./18/

If one takes this evidence at face value, then canonical Matthew is at best a translation or---given the priority of Papias --- a completely revised Hellenistic version of this "Hebrew" source.  For it is virtually inconceivable that the Matthean stories of Jesus’ conception and temptation were composed in any language other than Greek. /19/If there ever was a primordial Semitic text ascribed to Matthew---as these four Greek fathers claim---it was not identical with our canonical gospel text and it was lost.  Thus, far from confirming the priority of canonical Matthew, these citations are solid evidence that

  • many texts known to early Christians were lost;

  • the Greek text of Matthew is a derivative work; and

  • there was a primitive gospel source that is no longer extant.

2.2. Non-extant sources.

(a) Objection


Skeptics argue that Q cannot be regarded as "canonical" since there is no need to posit a hypothetical lost sayings source. Competent synoptic scholars can better resolve the synoptic problem by hypotheses based only on extant documents.

(b) Reply. 

Appeal to non-existent sources is certainly unwarranted when extant documents and normal redactional tendencies can explain all the features of a text or texts. But that is not the case with the synoptic gospels.

If the Augustinian tradition that Mark got his material from Matthew and Luke got his material from both accounted for the details of those texts Griesbach would never have thought to propose a revised sequence. Likewise, if Griesbach’s hypothesis provided a satisfactory explanation of textual data, that hypothesis would have become dominant and the theory of Markan priority, which cannot appeal to any ancient tradition, would never have attracted wide support.

In fact, however, proponents of almost all synoptic hypotheses from Papias to followers of William Farmer have found it necessary to posit the existence of proto-texts that are no longer extant. Whether it is a Hebrew logia collection (Papias) or Greek Q (most 20th c. scholars), an Ur-gospel (Lessing) or proto-Mark (Holtzmann) or proto-Luke (Streeter) or proto-Matthew (Parker, Sanders, etc.) is not relevant here. Scholars on all sides of the synoptic problem repeatedly have found that the extant Greek texts of the synoptics are not mutually self-explanatory but indicate some source(s) for which there is no surviving unredacted exemplar.

Even William Farmer, who spearheaded the current assault on the two-source hypothesis by rejecting any appeal to a hypothetical document eventually admitted:

"There is nothing wrong with hypothecating the existence of an otherwise unknown source or sources if there exists evidence that is best explained thereby." /20/

Farmer’s reluctant recognition that Luke’s form of a saying is sometimes more pristine than Matthew’s led him to admit "the necessity of positing the existence of collections of ‘sayings material’ behind Matthew and Luke."/21/ Eventually, he went so far as to posit the existence of lost written sources behind every gospel:

"The Two-Gospel Hypothesis does not require the critic to deny the existence of earlier sources used by the evangelists, written and/or oral...Thus, according to the Two-Gospel Hypothesis, Matthew wrote first, making extensive use of existing sources (oral and written). Luke wrote second, making extensive use of both Matthew and extensive use of other source material (oral and written). Mark composed his Gospel making extensive use of both Matthew and Luke with a limited use of other source material (oral and written)."/22/

This passage shows that, contrary to proponents’ claims, the Griesbach-Farmer hypothesis---the chief rival of the Q hypothesis in current American scholarship---is not simpler but far more complex. Instead of positing a single missing written source containing all the passages common to Matthew and Luke but absent in Mark, proponents of this self-styled Two Gospel hypothesis have to posit at least two "extensive" lost written sources (one for Matthew and one for non-Matthean parts of Luke) and a third more "limited" one (to account for material unique to Mark). In admitting that Matthew used a written source that is no longer extant, Farmer conceded the indispensability of the Q hypothesis in form if not in detail. The major difference between his theory and the prevailing view is that Q can be objectively reconstructed from passages common to two extant texts, while the contours of Farmer’s hypothetical missing sources --- like those in other theories --- are defined largely by the individual scholar’s speculations about the redactional history of each gospel. The litmus test that determines whether such a theory is really superior to Q is this: would it receive wide interest or support from scholars, if there were no ancient tradition of Matthean primacy that a conservative audience was eager to defend?

2.3. Overlaps and minor agreements.

(a) Objection. 


Skeptics claim Q cannot be considered canonical because there is ample evidence that Luke knew and used the text of canonical Matthew.  If this is the case, then the Q hypothesis is indeed dispensable.  To support this claim they point to:

  • overlaps between Markan and "Q" material and
  • identical wording in the Matthean and Lukan versions of a triple tradition passage not paralleled in Mark.

(b) Reply. 

The strength of a hypothesis depends on its ability to predict the preponderance of evidence. In a world of concrete material phenomena one can find exceptions to any rule or definition proposed by the human mind, if one looks deep enough.

But the law of gravity was not repealed simply because it did not explain all the behavior of matter at the submicroscopic level; nor were the definitions of mammal and bird discarded with the discovery of a platypus. Nor are rules of spelling and grammar abandoned even though abnormalities are easy to find in everyday speech or writing. Rules and definitions remain in force when they provide the most practical description of the normal behavior of phenomena.

Such is the case with Q. While critics can call attention to several passages in Greek manuscripts where the text of Matthew and Luke contains some identical wording that is not found in any extant copy of Mark, it remains true that there is no general pattern of verbal agreement between Matthew and Luke against Mark in the triple tradition. It is a demonstrable fact in passage after passage that the Greek text of Mark is the mediating term in verbal agreements with either or both of the other synoptics. The converse is also demonstrable: where Matthew and Luke differ from the wording of Mark in the triple tradition they also usually differ from each other.

The fact that these clear patterns represent the norm in pericope after pericope is what continues to persuade the majority of each generation of synoptic scholars that

  • Matthew and Luke depend on Mark rather than vice versa and
  • Luke does not depend on material in Matthew (or vice-versa).

The latter conclusion is confirmed by both

  • the sheer amount of non-Markan material in Matthew that is not contained in Luke (and vice versa), and
  • the number of instances in which shared non-Markan material is presented in a different place in the narrative sequence.

It is this preponderance of differences between Matthew and Luke in substance as well as style that is decisive here.  If there is direct literary dependence between the synoptic gospels -- as most scholars concede --, the only way to dispense with Q is to demonstrate that Luke is better explained as an edited version of Matthew./23/ Yet providing a detailed coherent explanation of why Luke deleted, rearranged and generally rewrote Matthew is a Herculean task that no rival hypothesis has yet accomplished to the satisfaction of the majority of synoptic scholars./24/

The theory of Austin Farrer, now championed by Michael Goulder --- that Luke also knew and valued the text of Mark ---, would account for Luke’s general pattern of abandoning Matthean constructions in most of the triple tradition. But it does not provide a cogent explanation of Luke’s retention of an occasional inconsequential Matthean word or phrase in a passage that otherwise is closer to Mark. It may explain the handful of pericopes where the substance of Luke’s text is closer to Matthew than to Mark./25/ But the Farrer-Goulder thesis does not provide a convincing explanation of why

  • Luke regularly departs from Matthew where Mark has no parallel and
  • Luke dismantles Matthew’s speeches of Jesus only to record many of the excised fragments in one lump --- the so-called "great insertion" --- with no discernable logical pattern.

If Luke knew Matthew, then in most cases he thought Matthean material needed to be reworked. If, as he alleges (1:1-2), he was trying "to compose an orderly account" (anataxthai diegesin) based on the "tradition" (paradosis) of those who were "the original eye-witnesses" (ap’ arches autoptai), and "ministers of the word," then his reasons for replacing the monumental monologues in Matthew --- of all sources --- with a thematically chaotic travelogue of his own design remain inscrutable. Indeed, one need only consult a standard synopsis outline to recognize that the mechanics of composing the core of Luke (ch. 10-18) without any prior template, out of original material and random bits the author deletes from Matthean passages --- both before and after this point in the narrative --- are mind-boggling, even in the age of modern computerized word processors.

Even skeptics grant that the real strength of the Q hypothesis lies in its ability to account for this section of Luke, and the sequence of double-tradition passages in general, more simply than any other redactional theory./26/ Compared with the problems other hypotheses have in accounting for the literary differences between Matthew and Luke, the instances of their agreement against Mark are minor indeed. The few substantive parallels are most simply viewed as evidence of Markan and Q versions of a common oral tradition similar to the occasional overlaps between John and the synoptics. Incidental verbal coincidences are easily explained by any of four facts:

  • as the least polished and most problematic of the synoptics, Mark invited emendation by later scribes in matters of grammar, style and content;
  • rhetorical patterns of a living oral tradition influenced ancient scribes’ reproduction of texts;
  • later scribes tended to harmonize synoptic texts; and
  • the original manuscripts of all the synoptics probably differed from modern reconstructions in some particulars.

Only if it could be demonstrated that the prototypes of Matthew and Luke displayed extensive substantial agreement against Mark would there be good grounds for rejecting the Q hypothesis. But that is virtually impossible. Moreover, as Christopher Tuckett has pointed out, the non-Markan minor agreements and overlaps in Matthew and Luke represent a two-edged sword that is more lethal to the Griesbach hypothesis than it is to Q./27/ For if Mark epitomized Matthew and Luke by concentrating on reproducing only their common material, then it is inexplicable why he would have altered or eliminated passages and phrasing that are identical in both.

Thus, far from being twin "Achilles heels" of Q, as E. P. Sanders has suggested, the minor agreements and few overlaps are actually a Trojan horse that subverts rival hypotheses./28/ More than a century after they were first noted consensus on the Q hypothesis is greater than ever, while remaining skeptics are divided and in disarray, launching alternate solutions to the synoptic problem, not one of which, Sanders concludes is "without objection."/29/


3.1. Verbal disagreement.

(a) Objection. 


Skeptics claim that Q cannot be considered a canonical text since there is not enough verbal agreement between parallel non-Markan passages in Matthew and Luke to prove that these sayings ever existed in written form before these two authors recorded them.

(b) Reply. 

Normal skepticism about invisible objects has kept many scholars who were convinced by concrete evidence that Mark was the first of the synoptics from immediately accepting the hypothesis that Q was a written document---myself included.  

But it has not, in the long run, prevented the persistence of that judgment as the eventual conclusion of most synoptic specialists.

The fact that Matthew and Luke do not generally preserve extensive verbatim parallels should neither be ignored nor obscured in any description of Q. But this is even more a problem for champions of the Griesbach hypothesis, who claim Luke copied Matthew, than it is for proponents of Q, who can demonstrate in the triple tradition that where Matthew is not verbally dependent on Mark, Luke is, and vice-versa.

Before the invention of mechanical means of reproducing documents, complete word for word agreement between any two texts is the exception rather than the rule, and where it occurs it is rarely sustained for more than a few lines. Hand copying invites alteration, whether it be accidental errors or conscious emendation. For to be reproduced one written text had to pass through the womb of a human mind other than that of the author. Thus, variant wording is the norm at all levels of the transmission of gospel manuscripts. And verbal variation is inevitably greater when the second scribe is not just a clerk but an author intent on publishing his own version of the evidence --- as is the case in the gospels.

So, verbal variation is no proof that Q was not a text. The wording of Matthew and Luke’s versions of triple tradition passages diverges enough to obscure a literary relationship, unless compared to Mark. In fact, it is harder to use Matthew and Luke to reconstruct the text of Mark than Q. If Mark is indisputably a text, Q is even more so./30/ For these gospels have greater agreement in their presentation of the double tradition -- in both wording and structure --than in their treatment of the triple tradition.

Q was obviously composed of small blocks of material that originally circulated orally. But the lack of sustained oral mnemonic devices to combine these blocks of Q tradition in Matthew and Luke indicates that it reached them in written form./31/

3.2. Disputed borders.

(a) Objection.


Critics claim Q cannot be considered a canonical text since even Q specialists disagree about what materials this hypothetical source originally contained.

(b) Reply.

Today there is greater consensus than ever among synoptic scholars about the contours of Q. Its general boundaries are defined objectively by non-Markan parallels in Matthew and Luke, with additions limited to

  • six triple tradition passages in which Matthew and Luke’s wording displays substantial agreement against Mark;
  • single tradition passages framed by or next to Q material in one of the gospels.

Single-tradition pericopes beyond the border of the double tradition are few and generally restricted to units with obvious stylistic and/or thematic echoes of the adjacent section of Q. Scholarly debate about the internal tradition history of the Q material does not disturb this consensus, since Q is by definition the version of the text known to Matthew and Luke. Current scholarly debate about the contours of this document Q is no greater than disagreement about the original ending of Mark. Uncertainty about whether Mark originally ended at 16:8 or 16:20 or somewhere else has not kept it out of the canon, so questions about Q’s contours should present no obstacle to canonizing it.

3.3. Arbitrary contents.

(a) Objection


Skeptics claim that a reconstructed text of Q cannot be considered canonical since it depends upon modern scholars' arbitrary preferences for one reading or another in determining the original text.

(b) Reply

The whole Bible is a mammoth reconstruction from variant manuscripts by trained scholars. It always was; it always will be.

The task of reconstructing Q from existing texts of Matthew and Q is not qualitatively different from that of producing the text of any gospel from the extant manuscript tradition. Determining whether Matthew or Luke better represents the original is no more arbitrary than deciding whether the Egyptian, Byzantine or Western recension represents the oldest version of a particular passage. In each case objective criteria are employed to determine the version that is most likely to be older.

There is no claim to historical infallibility in determining the contents of the canonical text. Alternate versions can and do favor variant readings of the same passage. A verse or phrase or word that was excluded from one version may be reintroduced in a revision; an element that had long been included may be later suppressed as a corruption. Yet each version is equally canonical, tentative passages and all.

In other words, the concept of canon was never meant to define a single immutable text that can claim to be identical with the original autograph. Rather, it has always been a functional concept defining what is recognized as the textus receptus at this point in time. It is a fact that every current canonical gospel is a reconstruction that is not identical with any single ancient manuscript./32/ If a modern reconstruction of Matthew, Mark and Luke can claim to be canonical, so can a modern reconstruction of Q.

Thus, Q should be included in any scholars' canon of early Christian sources. Even those people who continue to doubt that such a text circulated prior to the composition of Matthew and Luke should find it a useful tool for identifying elements of the synoptic tradition that were not included in Mark. Whatever hypothesis one adopts to explain the literary relationship of the synoptic gospels, Q remains an essential source of information about the historical development of the Jesus tradition.




/1/ English editions of Q include those by J. Kloppenborg (Q Parallels, Q-Thomas Reader), A. Jacobson (The First Gospel), B. Mack (The Lost Gospel, pp. 71-102), M.Borg (The Lost Gospel Q, pp. 33-118) and Westar Institute’s Scholars Bible translation panel (The Complete Gospels, pp. 249-300). Earlier J. D. Crossan outlined a skeletal sequence of Q passages in an appendix to his study of Jesus’ aphorisms (In Fragments, pp. 342-345). .

/2/ Anyone who doubts what is at stake should read the counter-alarms sounded in the epilogues of recent books by B.L. Mack (Lost Gospel, pp. 245-258) and W.R. Farmer (The Gospel of Jesus, pp. 197-201). .

/3/ The recent article by Charlotte Allen "The Search for a No-Frills Jesus" [Atlantic Monthly 278,6 (Dec 1996) pp. 51-68] is an example of public coverage of scholarly debate over Q that should be required reading for journalists and lay people alike.  .

/4/ "In one form or another, this hypothesis has been accepted by most scholars" (H. Koester Ancient Christian Gospels, p. 129). "The dominant theory in the scholarly community" (S.C. Carlson Synoptic Problem Home Page: Synoptic Theories, Hypotheses and Models --- despite his own preference for a proto-Matthew rather than Q).  .

/5/ Commentatio qua Marci Evangelium totum e Matthaei et Lucae commentariis decerptum esse monstratur 3 (Orchard & Longstaff, ed. Griesbach Studies, pp. 83-100; ET pp. 114-133 by J. B. Orchard). Griesbach did not acknowledge the similarity of his position to earlier proposals of Henry Owen (1766); these were first pointed out in 1826 by his former student, W. M. L. de Wette (B. Reicke, "Griesbach’s answer to the Synoptic Question," p. 61 in Griesbach Studies).  .

/6/ For a comprehensive outline of various solutions to the synoptic problem with lists of scholarly support visit Stephen C. Carlson’s Synoptic Problem Home Page .

/7/ Commentatio 3.5 (p. 88 in Griesbach Studies; ET by J. B. Orchard, p. 120).  .

/8/ Commentatio 3.1 (p. 86 in Griesbach Studies; ET by J. B. Orchard, p. 117).  .

/9/ Kummel, W. G. NT: History of Investigation, p. 75. Only three 20th c. critics are on record as endorsing the Augustinian position without qualification: Jameson (1923), Butler (1951) and J. Wehnam (1992). See, Carlson, Synoptic Problem Home Page: Theories .

/10/ W. M. L. de Wette appears to be the first to have alluded to Clement to support Griesbach in his Introduction to the NT (6th ed., 1860): "According to ecclesiastical traditions, Mark wrote his book later than the other two" (cited by B. Reicke, "Griesbach’s answer to the Synoptic Question," in Griesbach Studies [ed. Orchard & Longstaff], p. 62). But W.R. Farmer is largely responsible for invoking "the tradition of the church" and Clement in particular to win conservative Christian support for his championship of Griesbach (compare the brief passing notices to Clement in The Synoptic Problem [1964; pp. 1, 8, 282] with the priority given this evidence in "The Two Gospel Hypothesis" [1990; pp. 125-130 in D. L. Dungan, ed., Interrelations of the Gospels] and The Gospel of Jesus [1994; pp. 15-18].  .

/11/ Commentatio 3.1 (Griesbach Studies, p. 84; ET 115). W. B. Farmer defends his preference for the Griesbach hypothesis thus: "advocates of the Two-Gospel hypothesis are prepared to argue that this hypothesis affords theologians and preachers a more adequate avenue to the earliest layer of the synoptic tradition, a more adequate avenue to New Testament christology, and a more adequate avenue to Church history." ("Two Document Hypothesis" in D. L.Dungan, ed., Interrelations of the Gospels, p. 147; italics mine).  .

/12/ Contrary to W. R. Farmer who claims "the burden of proof in this matter, in any case, rests upon the critic who would discount this tradition as having no historical value" ("The Two Gospel Hypothesis," p. 126).  .

/13/ "And in the same books [Hypotyposeis 6] Clement has inserted (tetheitai) a tradition of the original (anekathen) elders regarding the order of the gospels in this fashion: he said that those gospels that contain genealogies were written first." (Church History 6.14.5). [Cited as Farmer’s introduction to "The Two Gospel Hypothesis," in Dungan, Interrelations, p. 125 and misrepresented as Clement’s own words].  .

/14/ Eusebius contrasts Clement’s vague suggestion about Matthew and Luke’s composition with a detailed quotation of his testimony about Mark "...but [he said] that the gospel of Mark was arranged (oikonomia) thus: ‘When Peter preached the word to the people in Rome...those who were present called upon Mark, as someone who followed him for a long time and remembered the things (he) said, to record (his) sayings...’ So much for Clement." (Church History 6.14.6-7). The Greek syntax of this sentence implies the parenthetical pronouns not required by the passive participial construction of the last two verbs. .

/15/ Church History 3.39.16.  .

/16/ "Now Matthew also published a written gospel among the Hebrews in their own tongue" [quoted by Eusebius (Church History 5.8.2)].  .

/17/ "That according to Matthew ---who was once a tax-collector and later an apostle of Christ --- was written first. As it was composed in the Hebrew language, he published it for those from Judaism who became believers." (again cited by Eusebius Church History 3.31.6). The subject of the synoptic story of a tax-collector turned disciple is called "Matthew" only in Matt. 9.9. .

/18/ "Matthew, who first preached to the Hebrews; as he was going to go to others transmitted in writing in his native language the gospel according to himself" (Eusebius Church History 3.24.6).  .

/19/ The underlying logic of Mtt 1:18-23 that demonstrates Jesus’ birth literally fulfilled Isaiah’s prophecy (7:14) of the child born of a "virgin" presupposes the LXX translation of the Hebrew alma as parthenos Also, Jesus’ responses to the tempter (Matt 4:4,6,10) are virtually verbatim excerpts from the LXX version of Deuteronomy; and it is Mark who uses the Hebraic ha Satan in this and other scenes, while Matthew regularly prefers the Greek diabolos. .

/20/ "A Fresh Approach to Q," in J. Neusner, ed. Christianity, Judaism and Other Greco-Roman Cults,p. 46n4 (cited by A.D. Jacobson, First Gospel, p. 17n38).  .

/21/ "A Fresh Approach to Q," p. 46n4 (cited by A.D. Jacobson, First Gospel, p. 18). .

/22/ "Two Gospel Hypothesis," p. 132 (italics mine).  .

/23/ Luke’s explicit claim to be writing to set the record straight in the face of other reports (1:1-4) makes the hypothesis that Matthew edited Luke improbable. .

/24/ Graham Stanton notes (Gospel Truth?, pp. 70-71): "scholars who claim that Luke has used Matthew must accept that it is always Luke who has changed Matthew's earlier form of the tradition. Their attempts to defend this view often look like special pleading." Michael Goulder's Luke: a New Paradigm (1989) has not yet converted many synoptic scholars to the view that Luke's version of the double tradition is an edited version of Matthew's. It is still too early to assess the achievement of the recently published volume Beyond the Q Impasse - Luke's Use of Matthew (ed. by A. J. McNicol et al, Valley Forge PA: Trinity Press International, 1996). Future editions of this essay will take that selection of studies and Mark Goodacre's recent Goulder & the Gospels: An Examination of the New Paradigm (JSNT Supplement 133; Sheffield: Sheffield Academic Press, 1996) into account. .

/25/ In only six passages is Luke substantially closer to Matthew than Mark: the message of John the baptizer, the testing of Jesus in the wilderness, the Beelzebul controversy, Jesus’ mission instructions to his disciples, the sign of Jonah and the parable of the mustard seed. In doublets, where either Matthew or Luke present two versions of the same material one is usually closer to Mark. .

/26/ E.g., E. P. Sanders and Margaret Davies, Studying the Synoptic Gospels, pp. 66, 114. .

/27/ The Revival of the Griesbach Hypothesis, pp. 64-75. .

/28/ Studying the Synoptic Gospels, p. 79. .

/29/ Studying the Synoptic Gospels, p. 112.  For an on-line analysis of problems raised by  leading source hypotheses (Augustinian, Griesbach, Farrer-Goulder, and Two Source) in accounting for literary details of the texts of  Matthew, Mark and Luke see my Synoptic Gospels Primer.   .

/30/ J. S. Kloppenborg (Formation of Q, pp. 43-44) cites Carlston and Norlin’s cautious conclusion from their extensive statistical comparison of Q and Markan passages in Matthew and Luke: "[Our samplings] are surely large enough to establish beyond reasonable doubt that Matthew and Luke used Q, as far as wording of their material is concerned, at least as conservatively as they used Mark. There seems to be no reasonable explanation of this phenomenon except a second written source for Matthew and Luke."  .

/31/ J. S. Kloppenborg, Formation of Q, p. 44-51. .

/32/ See R. W. Funk, Honest to Jesus, pp. 94-120 for a recent survey of the process and problems of producing a canonical text.



Allen, Charlotte. "The Search for a No-Frills Jesus" The Atlantic Monthly 278,6 (Dec 1996) pp. 51-68.

Bellinzoni, Arthur J. The Two Source Hypothesis: A Critical Appraisal. Mercer University Press, 1985..

Borg, Marcus. The Lost Gospel Q: The Original Sayings of Jesus. Berkeley CA: Ulysses Press, 1996.

Dungan, David L., The Interrelations of the Gospels: A Symposium. Mercer University Press, 1990.

Eusebius. The Ecclesiastical History (Loeb Classical Library) . vol. 1. (ET by K. Lake) & 2 (ET by J. E. L. Oulton). New York: G. P. Putnam’s Sons, 1926, 1932.

Farmer, William R. The Gospel of Jesus: the Pastoral Relevance of the Synoptic Problem. Louisville, KY: Westminster/John Knox Press, 1994.

_____, The Synoptic Problem: a Critical Analysis. New York: Macmillan Company, 1964.

Funk, Robert W. Honest to Jesus: Jesus for a New Millennium. San Francisco: HarperCollins, 1996.

Jacobson, Arland D. The First Gospel: An Introduction to Q. Sonoma CA: Polebridge Press, 1992.

Kloppenborg, John S. The Formation of Q: Trajectories in Ancient Wisdom Collections. Philadelphia: Fortress Press, 1986.

Koester, Helmut. Ancient Christian Gospels: Their History and Development. Philadelphia: Trinity Press International, 1990.

Kummel, Werner G. The New Testament: The History of the Investigation of its Problems. Nashville: Abingdon Press, 1972.

Mack, Burton L. The Lost Gospel: the Book of Q & Christian Origins. San Francisco: Harper Collins, 1993.

Orchard, Bernard & Thomas R. W. Longstaff, ed. J. J. Griesbach: Synoptic and Text-Critical Studies 1776-1976. Cambridge: Cambridge University Press, 1978.

Sanders, E. P. and Margaret Davies. Studying the Synoptic Gospels. Philadelphia: Trinity Press International, 1989.

Stanton, Graham. Gospel Truth? New Light on Jesus and the Gospels. Valley Forge, PA: Trinity Press International, 1995.

Tuckett, C. M. The Revival of the Griesbach Hypothesis. Cambridge: Cambridge University Press, 1983.


Carlson, Stephen C. The Synoptic Problem Home Page.

Goodacre, Mark. The Case Against Q.

Longstaff, Thomas R. W. The Two Gospel Hypothesis.

Smith, Mahlon H. A Synoptic Gospels Primer.


copyright © 1997-2008 by Mahlon H. Smith

This is a revised edition of a paper presented at the Spring 1997 meeting of the Westar Institute & posted on this site in March 1997. Special thanks are due to Westar Fellows & friendly critics like S. C. Carlson & Mark Goodacre for calling attention to points that needed correction or clarification. Second revision 12 April 1998. 

Hypertext links to this web page and brief quotations in scholarly reviews and publications are invited. But the text as a whole may not be posted or reproduced elsewhere without express written permission of the author.

- Last revised  05 April 2008 -


Website designed by Mahlon H. Smith
copyright © 1997- 2008

Visit  since February 1997 on our Web Counter.