The Synoptic Problem

The Doublets in Matthew

A DOUBLET consists of a pair of passages either of which is a repetition of the other. It is not necessary that the passages be exact duplicates. There are numerous doublets in the First and Third Gospels, as we now have these documents in critical editions. The Second Gospel in quite modern recensions presents, however, only a very few duplications. The doublets now found in the Synoptic documents consist usually of portions from discourse of the Saviour.

Doublets Brought Forward Many Years Ago

It has been imagined that the presence of doublets in two of the Gospels is indicative of their compositeness; and that the comparative absence of the phenomenon in Mark signifies that this Gospel is probably “not composite, at least in the sense that the two others are.”

If, indeed, the presence of doublets in Matthew means that this Gospel is a derivative from more than one source, then, when we take into account the additional circumstance that in quite a proportion of cases one part of the doublet is found in Mark and the other is not so found, it seems only fair to draw the conclusion that one of the sources primary to Matthew is this very Gospel of Mark. In view of this logical situation, it becomes my duty as an opponent of the hypothesis of the dependence of the First Gospel upon the Second to investigate the soundness of the proposition that the Matthaean doublets signify a plurality of sources back of the document in which they are found.

I propose, then, to establish the thesis that the doublets in Matthew—that is, those which require explanation— are probably due to a cause implying nothing as to compositeness. That is to say, I claim that the troublesome Matthaean doublets have arisen out of early interpolations made through agencies independent of the writer of the document.

This matter of doublets may be illustrated by a notable instance. I cite Mt. 5:29–30 and Mt. 18:8–9. The reader is referred to the third of a series of Matthaean doublets, whose Greek (WH) text is given in full later on in this present instalment. He will also find the text of Mk. 9:43, 45, 47, a portion of discourse which parallels the second part of the doublet.

A comparison of the Matthaean passages will show that while there are differences nevertheless either of the two substantially repeats the other. The pair constitute a doublet. The one member is paralleled in Mark, both in content and position. Neither member is paralleled in Luke.

It has been suggested that the two portions of text in Matthew are to be referred to different sources—the one to Mark, the other to some other document. If, indeed, this example of repetition in the First Gospel is only to be explained as a result due to the author turning now to one source, now to another, then we have in the doublet evidence tending to show that the Matthean writing is a compilation. A succession of similar instances—instances where one part, and one part only, of the doublet is a parallel, both in position and content, to a passage in Mark—would provide a very considerable basis for claiming that the First Gospel is a derivative of the Second.

But, are such Matthaean doublets to be explained as due to the use of Mark and one or more other sources by the writer who produced the Gospel of Matthew? Is this the explanation of the origin of these repetitions in the First Gospel? Perhaps the very first who thought so was C. H. Weisse, a writer who flourished in the earlier part of the nineteenth century. I give, in English, a passage from one of his writings.

Some of the sayings, here [at the end of Mk. 9] crowded together by Mark, they [the Evangelists Matthew and Luke] found, separated or incorporated in another context, also in the Source peculiar to them (the λόγια of Matthew). Thus, for example, the saying concerning causing the “little ones” to stumble [Mt. 18:6–7], which the First Evangelist here introduces in its complete form, but which Luke prefers to reserve for another setting [Lk. 17:lb-2]. Also the thoroughly dissimilar fragment of discourse of a paradoxical character, which is associated with it only lexically [i.e., by the word σκανδαλίειν], is to be found in that Source; and that our First Gospel, which had previously reproduced [Mt. 5:29–30] the same passage from that Source in the connection in which it was undoubtedly already incorporated there, nevertheless repeats it at the present place [Mt. 18:8–9] is a circumstance which appears as if expressly destined to afford final confirmation of the exposition of the true relation of the three Gospels to one another that has been undertaken by us. For, such a striking repetition, as precisely this passage, which the Third Evangelist has preferred to omit because of its rough aspect, is undoubtedly in need of an explanation; and what explanation could lie closer at hand than just this—that he had found it expressed in a different connection in two different sources?

“We have with this last remark perhaps not merely gained for this individual case a confirmation of our view, but unexpectedly have had opened to us yet a new variety of proofs, which we may use in favor of the way and manner of the composition of our Gospels that is maintained by us. In particular, in the First Gospel is disclosed an entire series, so to say, of doublets of single expressions of the Lord—and, in fact, of such where the one part belongs to that series of passages which this Gospel has in common with Mark, while the other part discloses itself as derived from that other chief source, from which the Gospel obtains its name. C. H. Weisse, Die Evangelische Geschichte (1838), Band I, S. 79–82.¹

Weisse finds in the view that the Matthaean writer had recourse to two sources and in consequence penned not only 5:29–30 but substantially repeated the passage in 18:8–9 a confirmation of the hypothesis that the First Gospel originated as a compilation based upon the Second Gospel and the λόγια mentioned by Papias.² “What explanation could lie closer at hand. . . . ?”

Perhaps there is no explanation lying closer at hand. But this has no point to one engaged in dealing with the phenomenon of doublets by scientific methods. What is wanted is not an explanation easily brought into service, but the one explanation which brings into view the cause actually at work when the doublet originated. Things close at hand are not necessarily things valuable to a scientific researcher. It is characteristic of true science, of scientific Biblical criticism, that it looks not for the easy road but for the right road—that it puts proposed explanations to severe tests and is by no means content with the setting forth of a cause which is competent to produce the observed phenomena. There may, in this and other cases, be a number of competent causes. Only one of these, however, was actually at work and was productive of the given effect. So here, what is the thing of real concern to a real critic is not whether a composite Matthew dependent upon two sources is an assumption capable of explaining the presence of the doublet under discussion, but whether this or perhaps some other assumption represents what actually occurred when the doublet was produced. I think that Weisse’s question as to what explanation could lie closer at hand stamps his state of mind as incompetent to the matter with which he was dealing.

A Competent Cause Not Necessarily The Actual Cause

It would be a very considerable mistake—in fact, an elementary blunder in logic—to imagine that, because an inquiry into doublets results in the discovery that the troublesome ones in Matthew may be explained on the assumption that Mark and a second source lay back of the First Gospel, therefore we have a completed proof of the truth of this assumption. That the consistency of facts with hypothesis is an aid to proof is not to be denied. But it is also evident, in a particular case, that the same facts may be consistent with some other and different hypothesis, though we may as yet be unable to formulate and define this other and different explanation. Until we have succeeded in showing that no second hypothesis is capable of explaining the facts, or else that for some other reason the first hypothesis must be the only explanation—until then, we have come seriously short of completing the proof. In other words, a given set of facts is to be understood as conceivably producible by any one of several independent competent causes; so that it is necessary by a process of exclusion or by some other procedure to single out the one true cause.

It is not enough, then, to say that the doublets in Matthew—that is, the group of troublesome doublets—are explicable as effects due to the use of Mark and one or more other documents by some compiler. The doublets may have originated from some other cause. And this is precisely the situation here. The Matthaean doublets that are represented by the one advanced by Weisse have never been shown, so far as I am aware, to be explicable only as the result of compilation.

In the foregoing, it has been assumed that the hypothesis of a plurality of sources, one of them Mark, provides a comeptent explanation of the facts. And it has nevertheless been shown that this competence affords an insufficient basis upon which to claim a proof of the hypothesis. In short, a competent cause is not necessarily the one true cause. I now proceed to challenge the competence itself—that is, the competence of the hypothesis to explain the doublets under consideration.

Uncertainty Of Competency A Damaging Characteristic

A reference to the excerpt from Weisse will show that he was aware of the fact that the second part of the doublet he was considering—that is, the passage, Mt. 18:8–9, concerning the offending hand, foot and eye— was dissimilar from the preceding context where the Saviour speaks of causing the little ones to stumble, and had merely a lexical association through the word Notwithstanding his perception of the unsuitability of the setting in which he found the second part of the doublet, Weisse raises no question as to whether the original writer of Matthew was responsible for its insertion at an unsuitable point. What reason is there for assuming that a Matthaean compiler who was aware of the true connection, having presumably already placed Mt. 5:29–30 next after Mt. 5:27–28, would nevertheless insert a repetition and insert it in a wrong setting, however some second source might present the passage? And even if we find the Matthaean compiler guilty, there still remains the question as to the rightfulness of assuming a source with a wrongly located passage in it.

It may be that, in 1838, at the time of the publication of Weisse’s statement, it was not easy to discern how such a passage as Mt. 18:8–9 could be present in all known copies and versions and at the point where we find it and nevertheless be there from some other cause than the will and purpose of the original writer of the book. At any rate, I propose eventually to show that there is now available abundant evidence to make understandable early interpolations of reliable material into the text of any and all of the four Gospels.

Some Opinions As To Doublets

The reader may, however, think that the advocates of a Matthew derived from Mark who have carried on their work during the many succeeding years up to the present have probably displayed a grasp of the logical situation better than that which was manifested at the beginning. Let us then have before us some representative statements of writers and investigators who have flourished at more recent periods.

“We have, however, obvious proof of this discourse-material having been drawn from a written source, in the duplicate sayings, at one time given by the Evangelist [writer of Matthew] in Mark’s connection and with adherence to his form, at another time in quite a different connection and in a modified acceptation; a circumstance which can only be explained on the assumption that the Evangelist looked on the sayings that lay before him in different forms as distinct utterances.” B. Weiss, A Manual of Introduction to the New Testament, vol. ii (1889), p. 268 (translated from the German by A. J. K. Davidson).

“A similar conclusion—the existence of a source used in common by Mt. and Lk. but different from Mk.—is indicated by the doublets, that is to say the utterances which either Mt. or Lk., or both, give, in two separate places.

“(a) In the majority of cases it can be observed that in Mt. the one part of the doublet has a parallel in Mk. and the other [part] in Lk. In these cases it is almost invariably found that in the parallel with Mk. not only the occasion but also the text is in agreement with Mk., and in the parallel with Lk. occasion and text are in agreement with Lk.” P. W. Schmiedel, Encyclopaedia Biblica, vol. 2 (1901) article Gospels, columns 1853–1854.

“The faithfulness in transcription from the ‘Collection of Sayings’ and the ‘Story of the Miracles’ is seen in the frequent reduplications in Matthew. For in this transcription is to be found the simple explanation of these repetitions. They were found by both Matthew and Luke in their sources, and taken from both.” T. C. Hall, The Messages of Jesus according to the Synoptists (1901), pp. 17-18.

“We have in conclusion two capital proofs that the complicated inter-relation of the gospels is really explained by the hypothesis of such a second chief source besides Mark: first the so-called doublets of Matthew and Luke. The case occurs in Luke nine times, in Matthew twelve times, that a saying of Jesus meets us twice over. The explanation is simplicity itself: it occurred in both sources.” P. Wernle, The Source of our Knowledge of the Life of Jesus (1907), p. 96 ( translated by Edward Lummis).

“It [the hypothesis of two principal documentary sources from which the Gospel history, as set forth in the Synoptics, was mainly derived] explains admirably the broad features in the relationship of the first and third Gospels to one another and to St. Mark. It also explains to a large extent the phenomenon of ‘doublets,’ that is to say the instances in the first and third Gospels of the repetition of Sayings where one member of the pair commonly has a parallel, both as regards its form and position, in St. Mark, while the other member, although the same in substance, differes somewhat in form and is placed in quite a different context, often in the midst of matter common to the first and third Gospels but not found in St. Mark,” V. H. Stanton, The Gospels as Historical Documents, Part II (The Synoptic Gospels (1909), p. 45.

“. . . its [Matthew’s] composite character, exhibiting as it does no less than twenty-two instances of the same incident or saying twice told in slightly different forms.” B. W. Bacon, An Introduction to the New Testament (1900–1924), p. 193.

Let the reader consider the foregoing excerpts and let him consider them a second time, seeking to distil from the whole of them or from any single one a proof that doublets in Matthew require the assumption of two or more sources. If he comes to the conclusion that the repetitions can be explained by this assumption—very well, what then? Has he in consequence attained to a proof? It may logically be asserted that the duplicate passages might have originated thus. But might is not must. And it would be rather silly to pretend that they are the same.

Calculation Of The Probability

We may appraise the logical situation with more exactness in the following way. We have, to begin with, one known explanation which we assume to be competent. The number of unknown competent explanations is indeterminate. There may be none. Then, there may be one, two, three, or more. On the assumption that the number is zero, the probability that the known competent cause—that is, a plurality of sources, Mark being one— is the true explanation is certainty, which is represented by 1. On the assumption that the number is one, the probability that the known competent cause is the true explanation is ½ . If we stop here, adopting the view that there could be no more unknown competent causes than one, we may combine the two assumptions already made and calculate that, on the assumption that either there is one known competent cause or else there are just two competent causes, one known and the other unknown, the probability that the known competent cause is the true explanation amounts to just ¾. That is to say, the two assumptions stand upon an equality—it is just as probable that there is just one known competent cause as it is that there are two competent causes one of which is known and the other unknown. We accordingly get for these two probabilities ½ and ½. The former of these multiplied by 1 gives the probability that the one known competent cause is the true one on the assumption that there is but one competent cause. The latter multiplied by ½ yields the probability that the one known competent cause is the true one on the assumption that there are just two competent causes, one known and the other unknown. The total probability that the one known competent cause is the true one is, accordingly, (½ x l) + (½ x ½)= ¾.

If we go on and assume that there are two competent unknown causes, we shall get for the total probability that the known competent cause is the true explanation the fraction 11/18. This is the result of adding ⅓ x 1, ⅓ x ½ and ⅓ x ⅓. This value 11/18 is less than ¾.

In short, if we continue and assume that there are three competent unknown causes, then four, and so on, we shall get smaller and smaller fractions for the total probability that the one known competent cause is the true one.

For the case, then, where there is one known competent cause and where it is not known that this is the only possible competent cause, the maximum value for the probability that this one known competent cause is the true one is ¾.

That is to say, when we assume that the troublesome doublets in Matthew can be adequately explained on the hypothesis of a compilation from Mark and at least one other source, and take into account that the question of other explanations, not yet known, is a perfectly open one, the maximum probability that the explanation advanced is the true one is precisely ¾, no more, no less. This is a long way short of certainty.

It would be absurd, in view of the foregoing analysis of the mathematico-logical situation, to claim that, because we have found an adequate explanation of the facts, therefore we have found the true cause which produced those facts.

Those of my readers, then, who understand the elements of mathematical probability and who have followed the preceding discussion, should have no difficulty in seeing that even the modest probability of ¾ will be further reduced, if it becomes clear that the explanation advanced is seriously lacking in its competency to explain the facts as to the doublets. Other readers who are unacquainted with mathematical probability may nevertheless understand the situation sufficiently to grasp the fact that it is not enough to show that an hypothesis is fully competent to explain the phenomena, unless directly or indirectly it may also be shown that there is no rival hypothesis just as competent to explain the same phenomena; and that the hypothesis advanced will become still less probable, if it is seen that it falls markedly short of affording an adequate explanation.

The combination of two uncertainties, (1) as to existence of rivals and (2) as to competency to explain, constitutes a poor basis upon which to found a claim that the hypothesis as to a plurality of sources reflects what actually took place when Matthew was composed.

Inquiry As To Competency

But let us make sure that this explanation is not really known to be adequate. First of all, let us consider the fact that quite a number of the Matthaean doublets require no explanation. They are simply repetitions of the same or a similar thing upon distinctly different occasions. That Jesus should, for example, under two sets of circumstances, use very similar language respecting divorce is nothing strange. There are, however, two kinds of doublets which cannot be disposed of in this way. Matthew contains, perhaps, eight repetitions where one part is certainly or probably in a wrong connection. The passage is out of place. Then there is one doublet which discloses a considerable duplication of narrative language. It is not permissible to explain these classes of doublets as phenomena arising out of the ordinary activities of a writer, as may very well be done when we have merely to do with pairs of short duplications of discourse material in which both members of each pair are embedded in suitable, though different, contexts.

Nine pairs of passages come into the field of consideration. For these we inquire whether we can say that we know that they are capable of being explained as phenomena due to a compilation of Matthew from Mark and one or more other documents. I now give these, together with any Markan parallels.

In one of the foregoing instances of repetition (4:23 and 9:35), the only case where we are concerned with narrative, we have not merely a similarity of import but a notable amount of identity of language. The first two words in both passages are the same. Then, beginning with διδάσκων, we have precisely the same verbal succession for a total amounting to eighteen words. Here is something calling for explanation.

The remaining eight doublets consist entirely of passages from the Saviour’s discourse. In the case of one member in each instance, there is more or less reason to think that the passage occurs in the Matthaean text in a connection other than that in which it was originally uttered by the Saviour. An explanation is needed for such cases.

That one of each pair of passages is a misplacement may be argued from the following considerations. Mt. 7:19 is concerned with a matter of judgment and accordingly seems out of harmony with verse 20, occupied as it is with the enunciation of a rule for the detection of false prophets (verse 15). In fact, it appears to interrupt the connection of this latter passage with the context preceding verse 19.—Mt. 18:8–9 lacks continuity with the preceding context, since the latter is concerned with little children and the question of causing to stumble, whereas the member of the doublet is occupied with the things which may cause one’s self to stumble.—Mt. 12:33 is perhaps a misplaced saying of the Saviour, inasmuch as it deals in its last clause with the formulation of a rule for distinguishing between the good and the corrupt trees, while the preceding and following contexts are occupied with other matters—the gravity of blasphemy against the Holy Spirit and the impossibility of offspring of vipers uttering good things.—Mt. 11:24 is probably out of position because we have ὑμἰν where only a single city, Capernaum, is concerned. Doubtless, the verse belongs—as the context preceding the Lukan parallel (Lk. 10:12) suggests—in some address given by the Saviour to disciples about to depart upon a missionary journey.—Mt. 10:22a and 22b are imbedded in a passage 17–22 which very exactly parallels in language Mk. 13:9, 11–13. The six Matthaean verses occur in an address delivered by the Saviour upon the occasion when He sent forth the Twelve. It seems rather improbable that at this juncture there was any need to utter a warning that the disciples would be delivered up to councils or that they would suffer scourging or that they would be brought before governors and kings or that brother would be delivering up brother to death or father his child or that children would rise up against parents and cause their death or that the disciples would have to endure the hatred of all men. But these things do fit into Mt. 24, as is shown by the Markan parallel already cited. In short, Mt. 10:17–22, and probably inclusive of 10:23, constitutes a passage belonging elsewhere than in the place where it is found. This would mean that 10:22a and 22b are both in wrong positions.—Mt. 10:38 seems to belong with the preceding verse, as is suggested in part by the repetitions of the words, οὐκ ἔστιν μου ἄξιος. In fact, these two verses are concerned with the matter of devotion to the Saviour. They accordingly seem out of place in an address to departing missionaries who had already left their homes, taken up their crosses and followed after Jesus.—Mt. 10:39 seems out of connection with the preceding context, once verses 37–38 have been removed.

In view of the foregoing considerations, it appears quite probable that in Matthew we have to do with a number of doublets—up to a maximum of eight—of which each consists, as to one of its halves, of a misplaced saying of the Saviour. There is also one other doublet, presenting a really considerable amount of exact duplication of narrative language. Do we have, in the hypothesis of a Matthew derived from a Markan and one or more other sources, a satisfactory explanation of the presence of these nine doublets in our First Gospel? I think not.

The narrative doublet calls loudly for an explanation of some sort. It would be well nigh absurd to attribute these two passages to two different authors. Sir John Hawkins, Horae Synopticae (2d ed., 1909), p. 92, in his section devoted to doublets, has no documentary explanation of his own to offer, but presents an hypothesis set forth by Dr. George Salmon, Introduction to the New Testament (9th ed.), p. 580. In accordance with this proposal by Dr. Salmon, the compiler of Matthew wrote down Mt. 5:29–30, following a text before him. He then broke off and turned to another source. Later on, he resumed work with the first source and, inadvertently or otherwise, substantially repeated what he had previously accepted. As we cannot very well conceive the compiler as having taken from his source, in either case, merely the text of one member of the doublet, we are to assume that when he accepted from his exemplar what we now find in Mt. 4:23, he also took over what at present occurs in the preceding context and that when he followed his source and wrote down the text found in Mt. 9:35 he likewise obtained from this primary document what now is observed to be in the succeeding context. The textual matter immediately preceding Mt. 4:23 is the Matthaean account of the Summoning of the fishermen. Upon comparison with the Markan parallel (Mk. 1:16–20), we find a very close resemblance between the two passages. The almost exact identity of the short discourse passage (Mt. 4:19 and Mk. 1:17) is perhaps nothing at which to be surprised. On the other hand, the almost identical mode of narrating the two calls is remarkable. Note especially the identity of the statement made, as by both authors, in explanation of the situation in which Jesus found Simon and Andrew—ἦσαν γὰρ άλεει`ς (Mt. 4:18 and Mk. 1:16). The details throughout are, in general, the same or similar; and the order of the corresponding details is substantially identical. We reach the highly probable conclusion, from the foregoing considerations, that wherever the Matthaean compiler got 4:23 he also got 4:18–22. That is to say, it is quite probable, once we assume a Matthaean compiler working with Mark and one or more other documents before him, that the source which he temporarily abandoned after having written down 4:23 was the Second Gospel. This conclusion is not disturbed when we compare Mt. 9:35 and succeeding context with Mk. 6:6b and succeeding context. But the result that we must choose Mark as the source used by the Matthaean compiler, when breaking off at the end of 4:23 and when resuming at the beginning of 9:35, is apparently fatal to Dr. Salmon’s conjecture. The Matthaean doublet cannot be explained as due to any conceivable use of Mark, since this Gospel does not contain anything more than a small portion of the eighteen-word narrative text duplicated with exactness in Mt. 4:23 and 9:35. It would seem, then, that we are not permitted to conceive that the Matthaean narrative doublet under discussion originated from a compiler’s double use of the same source. Nor, is it permissible to assign the two nearly duplicate halves of the doublet to different authors. In short, then, the hypothesis that Matthew is a derivative of Mark and one or more other sources is not found to assist us in explaining the presence in Matthew of the remarkable narrative doublet. Consequently, the fact of the existence of this doublet cannot be used as an argument that the cause must have been a derivative Matthew founded upon other documents one of which was Mark.

Nor can the remaining eight doublets be used for this purpose. The misplaced halves can be explained as interpolations into the original text, interpolations due to activities outside those of the writer. Nevertheless, let us grant—for the moment—that such doublets are conceivably producible by a compiler through the use of a plurality of documents. This is all very well as far as it goes. But a cause conceivably competent is quite different from a cause known to be competent.

Apparently No Actual Examples Adducible

I am not aware that anyone has shown that compilation from two documents ever did produce in the earlier centuries a doublet comparable with those with which we are concerned. That is to say, if we assume the general conditions of composition and transmission that obtained during the period when the Gospels were written and when documents were circulated only in the form of manuscripts, there appears to be no known example of a doublet of the kind under consideration which demonstrably came into existence under these general conditions because a compiler drew from two sources.

Examples Of Doublets Very Probably Formed By Interpolation

On the other hand, there are numerous instances of doublets formed as the result of interpolation—doublets found in Greek, Latin and other MSS. of the Gospels. Some of these had, apparently, a relatively short life; others seemed to have occurred only in a very few MSS.; while still others had a long and persistent life, and secured a place in what may be termed the Byzantine text, from which the so-called “Authorized Version” of 1611 was translated. I now proceed to cite from this version examples of doublets which almost certainly originated through interpolation. The instances come one from each of the four Gospels. There are numerous other similar cases of duplication which occurred in variously sized groups of MSS. And there are a few more which found a place in the Byzantine text. In all the instances cited, one member of the doublet is well authenticated, and is to be regarded as having originated with the autograph. The other member of the doublet, however, despite the fact that it was sufficiently supported to secure position in the Byzantine text, is omitted by our two most important authorities, the Greek Codices Vaticanus and Sinaiticus. We are to view the four instances of insufficiently supported passages as most probably examples of interpolation, though not necessarily purposed interpolation.

The foregoing are to be viewed as a few out of a rather numerous group of doublets that have arisen through ordinary interpolation. Accordingly, we must regard the autographs of the Gospels as having been free from these instances of duplication. If, now, we explain the eight Matthaean doublets as also due to interpolation, our hypothesis will have the great advantage of presenting a cause known to be adequate—not a cause conceived to be competent, but one ascertained to be so by an examination into the history of similar doublets.

The reader is, however, not to understand that the present discussion of this matter is to be taken as a full presentation. He is rather to see in it a preliminary indication of the method of solving the problem of Gospel doublets of the sort exemplified by the Matthaean eight, which he may expect to find set forth, later on, with some approach to completeness. That is, at the point further on, it is proposed to show more or less sufficiently the extent to which the Gospel MSS. have been interpolated and then to exhibit numerous instances where interpolations have resulted in the completion of doublets. When this exposition has been duly made, we will be in possession of a cause which has been ascertained to be competent to produce such doublets as the Matthaean eight. Such a cause may then with considerable confidence be brought forward as an explanation fully adequate to the facts. But, in such a case as that of the hypothesis that the eight doublets arose because of compilation from a plurality of sources, where the explanation put forth cannot be demonstrated to be competent and has to be viewed as only supposedly so, we have an hypothesis which is subject to a sharp reduction in respect to the probability that it was the cause actually at work, because of the possibility of competing explanations, and also to a further reduction, because of the uncertainty as to its competence.

If, by a resort to the Gospel MSS., or by some other procedure, it could be shown that doublets of the kind under consideration have actually arisen from compilation based on a plurality of documents, then the eight doublets in Matthew would have a more substantial value as evidence of a composite First Gospel dependent upon the Second. Can we find an example of a doublet of the kind represented by the Matthaean eight which has undoubtedly been formed through compilation from two sources? The Third Gospel is a document which contains quite a number of doublets and which may not unreasonably be supposed to have been produced by compilation from Mark and other writings. And certain of these pairs of duplicated passages consist in part of a member which is probably wrongly placed. Can any of these doublets be shown to be due to the presence of the same matter in more than one of the sources? Let us see.

Consider the following doublet:

Is this a doublet of the kind exemplified by the Matthaean eight? In all the eight instances in Matthew, one member of the doublet is assumed to be out of its proper place. That is also probably true in this Lukan instance of duplicate passages. The verses, Lk. 8:16–17, seem to be out of position, and in fact to break up the continuity between 8:15 and 8:18. If, now, we assume that 8:16 is really out of position, we have a doublet resembling more or less those belonging to the Matthaean group of eight. Let it next be noted that this wrongly placed member is parallel in content and position with Mk. 4:21. We may now claim considerable resemblance with that doublet of the Matthaean eight that is found in Mt. 5:29–30 and 18:8–9. In that instance, the wrongly placed 18:8–9 is parallel in content and position with a Markan passage.

The doublet half found in Lk. 8:16, which parallels in content and wrong position Mk. 4:21, may properly be claimed to have its source in the Second Gospel rather than in some one of the remaining and unknown exemplars upon which we are assuming the Third Gospel to depend. The other half of the Lukan doublet—that is, Lk. 11:33—is in a non-Markan setting and one very different from that which surrounds the half found in 8:16. If we can claim that this shows that the source of Lk. 11:33 was a document different from Mark, then we may also claim that the Lukan passages under consideration constitute an example of a doublet one-half of which came from Mark and one-half of which came from another source. However, upon examination of the context in which Lk. 11:33 is embedded, it may be seen that this passage can scarcely be in the position in the Saviour’s discourse in which it originated.

There is really no necessity to understand that the writer of Luke obtained it from a document in which it was found in its present position nor in fact that he was the one who set it down in this place. It is sufficient to see that it may very well have entered the text from the margin, where it had originally been put by a copyist because of a recollection of Lk. 8:16 and Mk. 4:21 and because of the thought that the passage had marginal value for the purpose of comparison with Lk. 11:34–36 on account of the similarity of ideas suggested by λύχνος (verses 33, 34, 36) and τὸ φῶς (verses 33, 35). In short, Lk. 11:33 may very well be an interpolation. Accordingly, it cannot confidently be asserted to be present in the Third Gospel because the author compiled from Mark and some other source.

Similarly, in the case of the Lukan doublet constituted by 8:17 and 12:2. The former passage and its Markan parallel, 4:22, are both very likely out of place, and it may be claimed that the Lukan text here is derivative from the Markan. This leaves Lk. 12:2. As it occupies a setting not found in Mark, it must have come into Luke by some route other than one passing through the Second Gospel. We are under no necessity to understand that 12:2 entered at the point where it now occurs because the Lukan author got it and its context from a non-Markan document. It is not at all necessary to bring in the Lukan author, since the passage cannot be claimed as very certainly in its right position, as Matthew gives a different one for the parallel, Mt. 10:26. Again, we have a failure to find in Luke a doublet which may confidently be asserted to owe its existence to compilation from more than one source.

What seems to be needed, from the point of view of practical logic, is one-half the Lukan doublet paralleled in Mark and situated in a wrong position in both the Gospels and the other half occupying a certainly authentic position. Under these circumstances, it appears possible to claim part of the doublet as derived from Mark and part from a non-Markan source used by the compiler. If the part from a non-Markan source is in a wrong position, then it cannot confidently be asserted to be a passage introduced by the compiler. We do not need to think of the compiler at all, since it is too easy to assume that it came in through the margin, especially if a good reason is assignable as to why it should have originally been inserted there in some MS. succeeding the autograph. That is to say, we require one member of the doublet to be fairly probably in a wrong position and one member in a highly probable right position. One of these, it is necessary to show, came from Mark and the other from a non-Markan source. The combination of a right position and a Markan source leads to little or nothing, because we must then assume, for the other member of the doublet, a wrong position and a non-Markan source. This combination can be satisfied by an interpolation from tradition or a non-Markan document—as, for example, from the other part of the Lukan doublet—and does not require that the compiler be made the agent by whom the passage was introduced. And the result of this is that the half in a wrong position and from a non-Markan source cannot with confidence be set up as having the compiler back of it. Consequently, the combination of a right position and a Markan source being excluded as infertile, we are left with the combination of a right position and a non-Markan source. This necessitates also the combination of a wrong position and the Markan source.

What has apparently been omitted from consideration is the case where both members of the Lukan doublet are in wrong positions. Such an instance would conform with the requirement that one member should be wrongly placed. Both members in wrong places would, however, lead to nothing, since the one from a non-Markan source can be assumed to be an interpolation made later than the composition of the autograph.

In our search amongst the doublets in Luke, we are, in view of the foregoing analysis of the situation, to be guided by the following criteria:

1. In the case of one member, we must show Tightness of position and non-parallelism with Mark.

2. In the case of the remaining member, we must show that it is not improbably in a wrong position and that it is paralleled in Mark both in position and in content.

Sir John Hawkins, in Horae Synopticae (2d ed., 1909), pp. 99 ff., lists eleven or twelve Lukan doublets. We have already considered Nos. 1 and 2. No. 3 is found in 8:18 and 19:26, Mk. 4:25 paralleling the former passage in position and content. However, 8:18 cannot, apparently, be shown as probably out of place. After we allow that both of the preceding two verses are to be considered as in wrong positions, the connection of 8:18 with the explanation of the parable of the Sower seems reasonable enough. Consequently, this doublet No. 3 appears unsuitable as an instance of the same kind as those exemplified by the Matthaean eight. It fails, because Lk. 8:18, though parallel with Mk. 4:24a, 25, is nevertheless apparently in its proper original position in the discourse as it came from the mouth of the Saviour. Criterion No. 2 is accordingly unsatisfied.

Doublet No. 4, found in Lk. 9:3–5 and 10:4–5, 7, 10–11, the former passage having a parallel in import and place in Mk. 6:8, 10, 11, consists of members both of which seem in proper position. Criterion No. 2 is again unsatisfied.

Doublet No. 5 has one member (14:27) without Markan parallel. Its other member (Lk. 9:23) is paralleled in position and content, even including narrative text, in the Second Gospel (Mk. 8:34), but this member appears to be in a very suitable setting; so that, once more, Criterion No. 2 bars the doublet from serving as an example of the kind of doublets illustrated by the Matthaean eight.

Similarly, the Lukan doublet, consisting of 9:24 and 17:33, has a Markan parallel as to content and position (Mk. 8:35) to its former half, and this portion of the doublet appears in proper place. There is still again a failure to satisfy Criterion No. 2.

And this same criterion serves to exclude Doublet No. 7, consisting of Lk. 9:26 and 12:9, because the former half, which is paralleled in place and substance by Mk. 8:38, seems in a very suitable setting.

Doublet No. 8, which is concerned with 9:46 and 22:24, presents an interesting case. The former passage is paralleled as to content and place by Mk. 9:34b. Sir John Hawkins (Horae Synopticae, 2d ed., 1909, p. 103) calls attention to the fact that the combination τὸ τὶς, occurring in both parts of the doublet, is a Lukan characteristic. This would tend to confirm the prima facie evidence that the presence of the passages in Luke is indicative that they owe their inclusion in this Gospel to the writer of it. But this cannot be pressed to signify that both were necessarily placed in their present positions by this writer, since, conceivably, either part of the doublet may be an interpolation originating from the citation of the other part in the margin.

If the earlier half of the doublet could be shown to be in a wrong position, then we would have an instance comparable with the doublets grouped in the Matthaean eight. For it can be made clear that Lk. 22:24 is very probably in its right place and consequently a passage whose import and placement are to be ascribed to the Lukan writer. And the Lukan authorship is confirmed by the presence of τὸ τὶς. We would then have two passages both incorporated in the text by the compiler—one from the Second Gospel and the other from a different source. That Lk. 22:24 is very probably in its proper historical position may be gathered (1) from its place in the text, and (2) from the fact that it affords an explanation of Jn. 13:1–17, where we find an account of how the Saviour washed the feet of the disciples, but where no occasion is intimated as affording an opportunity for the lesson in loving service. This is supplied by Lk. 22:24. We have, accordingly, in the case of half of the doublet, a passage introduced by the compiler presumably from a non-Markan source. As to the other half, 9:46, we have in it a passage from Mark, but one as to the suitability of whose setting nothing adverse appears to be in evidence. Accordingly, in the two passages constituting the doublet, we very probably have nothing more than accounts of two separate and distinct occurrences. This doublet is, then, not of the kind desired as an example. What is wanted is, in part, a doublet both halves of which consist of the same historical utterance or describe one and the same event.

Doublets Nos. 9 and 10 both consist of two passages one of which is paralleled in Mark as to place and content. And the portions of doublets paralleled in the Second Gospel are not to be considered as out of position. There is thus in both cases a failure to meet the requirements of Criterion No. 2. These doublets and the Markan parallels of portions are as follows: Doublet No. 9 is found at 11:43 and 20:46, and its second part is paralleled at Mk. 12:38b–39; Doublet No. 10 consists of 12:11–12 and 21:14–15. The latter portion has its parallel at Mk. 13:11.

Doublet No. 11 has for neither half a parallel in Mark. It consists of 14:11 and 18:14b.

There are a number of other pairs of similar passages which may be considered. Thus, Lk. 8:8b and 14:35b constitute a doublet. The former half is paralleled in place and import at Mk. 4:9. But it is in excellent position. Accordingly, we do not get in this doublet an example of the kind sought.

As to the Lukan pairs found at 5:29–30 (=Mk. 2:15–16) and 15:1–2; 6:9 (=Mk. 3:4) and 14:3; and 8:21 (=Mk. 3:34–35) and 11:28, it may be claimed that it is more or less doubtful whether they are to be classed as doublets. Let us not hesitate here, but grant that we have true duplications. In each case, the half paralleled in Mark in content and position is apparently in its proper place. None of these pairs of passages, even though we concede that they are properly considered doublets, furnishes us, consequently, with an example of the sort exemplified by the Matthaean eight.

I have now covered—perhaps with some approach to thoroughness—the possibility of securing from the Third Gospel, a document which may tentatively at least be regarded as a derivative of the Second, some examples of doublets which could confidently be shown to owe their existence to compilation from this Second Gospel and some one or more other sources. But no success has attended this effort, though numerous doublets have been considered.

Success would have meant that the explanation of the Matthaean eight which makes them the result of compilation from two or more sources could be regarded as somewhat certainly a cause competent to the effect observed. Lack of success, on the contrary, signifies that this proposed explanation is still not known to be competent. The logical situation is this: The explanation by compilation, assuming it to be competent, is not known to be the only competent one. The probability that this hypothesis reflects the facts is, accordingly, no greater than ¾. If, now, we take into consideration the additional fact that the competence of the hypothesis is uncertain, no actual instances being adducible of the operation of compilation from diverse sources in producing doublets of the kind exemplified by the Matthaean eight, then this modest value of ¾ must be still further cut down.

There is another hypothesis, one which does not involve the assumption that the writer of Matthew had anything to do with the formation of these doublets, but assigns a cause that is independent of the author and that can be shown by numerous actual examples to have been operative in the days of the transmission of the Gospels by means of manuscripts. A few examples have already been cited, but the evidence of the extent and character of the assimilation and interpolation to which the text has been exposed has not been set forth with due fulness and critical care. The reader may expect in what is to follow a discussion of the manner in which many more or less transient Gospel doublets have arisen and a presentation of an adequate number of actual examples.

(To be continued)

1) I supply the original German: Einige der von Marcus hier zusammengedrängten Aussprüche fanden sie, abgetrennt oder einem andern Zusammenhange einverleibt, auch in der ihnen eigenthümlichen Quelle (den λόνια des Matthäus). So namentlich den Ausspruch über die Verführung der “Kleinen,” welchen der erste Evangelist in seiner vollständigen Gestalt hier einschaltet, Lukas aber einem andern Zusammenhange vorzubehalten vorzieht. Auch der durchaus heterogene, nur lexikalisch daran geknüpfte, paradoxe Ausspruch fand sich in jener Quelle, und dass unser erstes Evangelium, welches denselben bereits aus jener Quelle in dem Zusammenhange, dem er unstreitig schon dort einverleibt war, gegeben hatte, ihn nichts destoweniger an gegenwärtiger Stelle noch einmal bringt, ist ein Umstand, der wie ausdrücklich dazu bestimmt erscheint, der von uns unternommenen Darlegung des wahren Verhältnisses der drei Evangelisten zu einander die letzte Bestätigung zu geben. Denn eine so auffallende Wiederholung, wie gerade dieses Satzes, den wegen seines schroffen Ausdrucks der dritte Evangelist lieber ganz hat weglassen wollen, bedarf unstreitig einer Erklärung; und welche Erklärung könnte näher liegen, als eben diese, dass er ihn in zwei unterschiedenen Quellen, in verschiedenem Zusammenhange ausgesprochen vorgefunden hatte?

Mit dieser zuletzt gemachten Bemerkung haben wir nicht etwa nur für diesen einzelnen Fall eine Bestätigung unserer Ansicht gewonnen, sondern es hat sich uns unerwartet noch eine neue Gattung von Beweismitteln eröffnet, die wir für die von uns behauptete Art und Weise der Zusammensetzung unserer Evangelien benutzen können. Vorzüglich im ersten Evangelium lässt sich eine ganze Reihe so zu sagen von Doubletten einzelner Aussprüche des Herrn nachweisen, und zwar von solchen, wo das eine Exemplar derjenigen Erzählungsreihe angehört, welche dieses Evangelium mit Marcus gemein hat, während das andere sich aus jener andern Hauptquelle geschöpft erweist, von welcher das Evangelium seinen Namen trägt.

2) Eusebius, Church History, 3:39:16. See Nicene and Post-Nicene Fathers, New Series, vol. 1, p. 173.

3) We have in this example of duplication from Mark what is really a triplet.