Divorce and Doublets

Introduction

The synoptic gospels (Mark, Matthew, and Luke) contain a number of doublets, which are pieces of text that occur in two different places in the same gospel and which may or may not have parallels in the other synoptic gospels. For more information on doublets in general, and in particular how much doublets/formulas relate to the synoptic problem, see the Introduction to Doublets.

Where there is a doublet then the author of the gospel in which it exists chose to make essentially the same point in two places, or to more or less repeat something someone (typically Jesus) said. Either that gospel author edited (or redacted) one when writing the other, or he had two different sources, and an examination of the doublet text, its parallels in the other synoptic gospels (if any) may provide clues as to which was the case. The text below examines the doublet in Mt 5:31-32 // 19:9 (which Hawkins refers to as his ‘Doublet in Matthew No. 2)

Below ‘aMark’ refers to the author of Mark, and similar notation is used for the authors of Matthew, Luke, and any other author of a ‘named’ piece of text. In addition, ‘Marcion’ means the gospel attributed to Marcion the text of which is almost exactly a subset of the text of Luke, and ‘aMarcion’ means the person who wrote that gospel (who may or may not have been Marcion himself). This is included to allow for the possibility that Marcion’s gospel (essentially a ‘cut down’ Luke) might have been, or be related to, an earlier version of Luke.

In the text the symbol ‘//’ is used to denote a doublet (parallel text in the same gospel), and ‘/’ denotes parallel text in a different gospel. 

Mk 10:2-10, Mt 19:3-8 – Rules for Divorce Before Jesus

In the synoptic gospels the rules regarding marriage and divorce begin in Mk 10:2-12 and their close parallels at Mt 19:3-9, with the major issue being whether divorce was allowed, and if so under what circumstances. In Mark Jesus begins by reminding his audience that Moses had told them that a man could divorce his wife by writing her a bill of divorcement (Mk 10:3-5), but then tells them that (now) divorce should not be allowed (Mk 10:6-9). In Matthew these words are reversed: Jesus first says there should be no divorce (Mt 19:4b-6b), and then his audience question why Moses had previously allowed divorce (Mt 19:7). 

Although it is not intended to cover all the theological aspects of these verses here some discussion is necessary, in particular whether divorce was allowed, and if so for what cause. In ‘The Matthean Divorce Texts and Some New Palestinian Evidence,’ Joseph A. Fitzmyer comments on Mt 19:3-9: 

Closely related to Mk 10:2-12 is the similar pericope of Mt 19:3-9. In fact, Matthew has derived it from his "Marcan" source, but he has modified it to make it better suit his Jewish-Christian concerns. First of all, he has cast the Streitgespräch in terms of the Hillel-Shammai dispute [The House of Shammai held that a man may only divorce his wife for a serious transgression, but the House of Hillel allowed divorce for even trivial offenses, such as burning a meal - Wikipedia], by making the Pharisees ask whether it is lawful to divorce one's wife "for any cause" (kata pasan aitian). 

Fitzmyer then points out that some people see the gospel order Mark -> Matthew suggested above as wrong for at least two different reasons: First, Mk 10:2-12 is a composite, formed from two different answers, first to the Pharisees in Mk 10:2-10, and then to the disciples in Mk 10:11-12; Second, the question in Mt 19:3 could only have been asked in circumstances in which divorce was lawful for at least some reasons, and is therefore earlier than that in Mt 10:2 in which it appears that divorce was not allowed at all. Jesus was being asked a ‘trick question’ regarding whether there he believed were any circumstances in which it was (or should be) lawful, and the NET Bible notes: 

The question of the Pharisees was anything but sincere; they were asking it to test him. Jesus was now in the jurisdiction of Herod Antipas (i.e., Judea and beyond the Jordan) and it is likely that the Pharisees were hoping he might answer the question of divorce in a way similar to John the Baptist and so suffer the same fate as John, i.e., death at the hands of Herod (cf. 6:17-19). Jesus answered the question not on the basis of rabbinic custom and the debate over Deut 24:1, but rather from the account of creation and God’s original design. 

Fitzmyer notes the composite nature of Mk 10:2-12 but does not record that Mt 19:3-8 has some significant differences from Mk 10:2-10. In both Mt 19:1 and Mk 10:1 Jesus travels to a new location and crowds gather, and then in Mt 19:2 Jesus heals them while in Mk 10:1 he teaches them. Then in Mt 19:3 the Pharisees ask their question about divorce, and while Mk 10:2 has the related question it is not certain who asked it, being either the Pharisees or some unspecified people, about which the NET Bible has this long note (paragraph breaks added): 

The Western text (D it) and a few others have only καί (kai) here, rather than καὶ προσελθόντες Φαρισαῖοι (kai proselqonte" Farisaioi, here translated as “then some Pharisees came”). The longer reading, a specific identification of the subject, may have been prompted by the parallel in Matt 19:3. The fact that the mss vary in how they express this subject lends credence to this judgment: οἱ δὲ Φαρισαῖοι προσελθόντες (Joi de Farisaioi proselqonte", “now the Pharisees came”) in W Θ 565 2542 pc; καὶ προσελθόντες οἱ Φαρισαῖοι (kai proselqonte" Joi Farisaioi, “then the Pharisees came”) in א C N (Ë1: καὶ προσελθόντες ἐπηρώτησαν αὐτὸν οἱ Φαρισαῖοι) 579 1241 1424 pm; and καὶ προσελθόντες Φαρισαῖοι in A B K L Γ Δ Ψ Ë13 28 700 892 2427 pm. Further, the use of an indefinite plural (a general “they”) is a Markan feature, occurring over twenty times. Thus, internally the evidence looks rather strong for the shorter reading, in spite of the minimal external support for it. 

However, if scribes assimilated this text to Matt 19:3, a more exact parallel might have been expected: Matthew has καὶ προσῆλθον αὐτῷ Φαρισαῖοι (kai proshlqon aujtw Farisaioi, “then Pharisees came to him”). Although the verb form needs to be different according to syntactical requirements of the respective sentences, the word order variety, as well as the presence or absence of the article and the alternation between δέ and καί as the introductory conjunction, all suggest that the variety of readings might not be due to scribal adjustments toward Matthew. At the same time, the article with Φαρισαῖοι is found in both Gospels in many of the same witnesses (א Ï in Matt; א pm in Mark), and the anarthrous Φαρισαῖοι is likewise parallel in many mss (B L Ë13 700 892). 

Another consideration is the possibility that very early in the transmissional history, scribes naturally inserted the most obvious subject (the Pharisees would be the obvious candidates as the ones to test Jesus). This may account for the reading with δέ, since Mark nowhere else uses this conjunction to introduce the Pharisees into the narrative. As solid as the internal arguments against the longer reading seem to be, the greatest weakness is the witnesses that support it. The Western mss are prone to alter the text by adding, deleting, substituting, or rearranging large amounts of material. There are times when the rationale for this seems inexplicable. In light of the much stronger evidence for “the Pharisees came,” even though it occurs in various permutations, it is probably wisest to retain the words. This judgment, however, is hardly certain. 

Although the mss evidence appears to be against the shorter reading, it nevertheless makes sense because the Western text is acknowledged to be very old, while the above note betrays the author’s disdain for the Western text by its ‘paraphrase’ of the comments from Westcott and Hort: 

It’s chief and constant characteristic is a love of paraphrase, not generically different from the tendency to verbal modification exhibited by many scribes, but rather an extreme form of it. Words and even clauses are changed, omitted, and inserted with surprising freedom, wherever it seemed that the meaning could be brought out with greater force and definiteness. 

This characterization is of course circular, since it pre-supposes that the Western text is an edited form, whereas it would be equally valid to state that as the Western text is early other texts “are prone to alter the [Western] text by adding, deleting, substituting, or rearranging large amounts of material.” Unfortunately, the often-used term ‘Western non-interpolation’ suggests a common cause for the omissions in the Western text, whereas the equally valid, but almost unused, term ‘Non-Western interpolation’ would suggest an inclusion or other change in the other texts instead. Essentially, the creation of this one term by Westcott and Hort led to the overall character of the Western text being called into question, instead of each textual difference being evaluated on its merits. 

In ‘Marcan Usage: Notes, critical and exegetical,on the second Gospel’ C.H. Turner explains why he think the shorter form is correct because it is a use of the ‘impersonal plural’: 

The passage is included here with a little hesitation, but I believe both that this is the true reading, and that ἐπηρώτων αὐτὸν is the impersonal plural - not 'the multitudes asked Him' but 'the question was asked of Him'. It would not be reasonable to suppose that the question of divorce was the dominant one in the minds of the crowds: Peter simply remembered the question being raised at that time. There is no parallel in Luke: Matthew supplies προσῆλθον αὐτῷ Φαρισαῖοι, from which many authorities have borrowed προσελθόντες Φαρισαῖοι for the text of Mark. 

Fitzmyer discusses two documents from Qumran, with particular reference to the meaning of the word zënût and its bearing on the question of order: 

In the OT zënût is used both of harlotry (e.g., Jer 3:2, 9; Ez 23:27) and of idolatrous infidelity (Nm 14:33). In the LXX it is translated by porneia (e.g., Jer 3:2, 9). Whatever one might want to say about the nuances of the word zënût in the OT, it is clear that among the Jews who produced the Damascus Document the word had taken on further specific nuances, so that polygamy, divorce, and marriage within forbidden degrees of kinship could be referred to as zënût. Thus, in CD 4:20 and 5:8-11 we have "missing-link" evidence for a specific understanding of zënût as a term for marriage within forbidden degrees of kinship or for incestuous marriage; this is a specific understanding that is found among Palestinian Jews of the first century B.C. and A.D… 

But now, in the light of the statute for the king in the Temple Scroll, which directly forbids polygamy (as does Dt 17:17) and goes beyond that to give a reason which at least implies the prohibition of divorce, the question put by some Pharisees to Jesus in Mk 10:2, "Is it lawful for a man to divorce his wife?" is not as "inconceivable" in a Palestinian milieu as might be supposed. Knowing about the Essene prohibition of divorce, a Pharisee could easily have posed the question to see where Jesus stood in the matter: Do you side with the Essenes or with the Pharisees? … 

Now if there is any validity to the interpretation of these divorce texts in the light of the Qumran material, we see that it does not support the position that the pronouncement-story and the dominical saying, as they are found in Mt 19, represent a more primitive form than that in Mk 10. In my opinion, it merely serves to accord to the Two-Source Theory its merited place as the most plausible solution to the Synoptic Problem. 

In Mk 10:2 the people or the Pharisees (depending on the ms) ask Jesus whether divorce is ever allowed, then after explaining why Moses did allow it in Mk 10:9 he tells them that God does not allow divorce, and later he tells just the disciples (who most likely had heard his earlier pronouncements) that the act of divorce and re-marriage constitutes adultery (Mk 10:11-12). In Mt 19:3 the Pharisees ask whether divorce is allowed for anything and in Mt 19:6b Jesus tells them that divorce should not be allowed, but then (in reaction to them asking why Moses allowed divorce in Mt 19:7) Jesus tells them that divorce and re-marriage (but only by a husband) is allowed in the case of fornication by the wife. Because the disciples comment on this in Mt 19:10 it is reasonable to assume that they heard what Jesus said to the Pharisees. Then, because Mark (no divorce) is more strict than Matthew (yes, but only for fornication) the relaxation of the rules supports the view that here Matthew is later than Mark.

Mk 10:11-12, Mt 19:9 // 5:31-32, Lk 16:18 – Jesus’ Rules for divorce (Hawkins: Doublet in Matthew No. 2)

As just indicated the initial statements in Mark and Matthew are followed by Mk 10:11-12 and Mt 19:9 respectively, in both of which Jesus lays down his rule that the act of divorce constitutes adultery, except that in Mt 19:9 an exception is allowed. None of the above text has a parallel in Luke, and instead there are much shorter sections on divorce at both Mt 5:32 / Lk 16:18 that have no parallel in Mark, and contain some significant differences from the rules in Mk 10:11-12 / Mt 19:9.

Assuming Markan priority we can reasonably assume that both aMatthew and aLuke saw Mk 10:11-12, and these verses might therefore be expected to form the core of the corresponding new rules in Matthew and Luke. Indeed, Lk 16:18a-c does follow Mk 10:11 regarding a husband committing adultery if he divorces his wife and re-marries, but both Mt 5:32b and 19:9d contain an exception to that rule in the case of fornication by his wife. Mt 5:32 adds other changes that state that it is the act of divorce (except in the case of fornication), not the re-marriage, that causes the adultery but, strangely, in Mt 5:32b the Markan rule is ‘flipped’ so that a husband divorcing his wife causes her to commit adultery, with no stated penalty for him if he re-marries. Additionally, in both Matthew and Luke Mk 10:12 is replaced by a rule that states that a man who marries a divorced woman commits adultery, although this rule is not present in Mt 19:9ef // 5:32d in some mss.

In the Old Testament it is clear that both divorce and re-marriage is allowed under some circumstances, for example as in Deut 24:1-2: 

When a man hath taken a wife, and married her, and it come to pass that she find no favour in his eyes, because he hath found some uncleanness in her: then let him write her a bill of divorcement, and give it in her hand, and send her out of his house. [24:1]  And when she is departed out of his house, she may go and be another man's wife. [24:2]

According to the NET Bible the Hebrew phrase עֶרְוַת דָּבָר (’ervat davar), translated above as ‘uncleanness:’ 

refers here to some gross sexual impropriety (see note on “indecent” in Deut 23:14). Though the term usually has to do only with indecent exposure of the genitals, it can also include such behavior as adultery (cf. Lev 18:6-18; 20:11, 17, 20-21; Ezek 22:10; 23:29; Hos 2:10). 

This suggests that this exception is similar to that in Mt 5:32b // 19:9d, except that it is not completely clear what “fornication” (porneia - often translated as ‘sexual immorality’ or ‘unlawful sexual intercourse’) means in these verses. It is distinct from adultery (moicheia) because they are separately referred to in several places, for example as the first two items in a list of 17 “works of the flesh” in Gal 5:19-21, so indicating that one is not a form of the other: 

Now the works of the flesh are manifest, which are these; [Adultery, ]fornication, uncleanness, lasciviousness, [Gal 5:19] 

Because adultery specifically refers to actions within a marriage this suggests that porneia cannot mean sexual immorality or unlawful sexual intercourse within a marriage, but instead can only refer to something that took place outside a marriage, as confirmed by 1 Cor 7: 

… It is good for a man not to touch a woman. Nevertheless, to avoid fornication (porneias), let every man have his own wife, and let every woman have her own husband. [1 Cor 7:1-2] 

However, you cannot divorce someone unless you are married to that person, and if you are married when the immoral etc. act took place it would be adultery, not porneia, so that the exception in Mt 5:32b // 19:9d would seem to have no effect. It therefore appears that the exception could only refer to porneia prior to a marriage that is discovered later, or in a marriage that is found to be invalid for some reason. Fitzmyer writes as follows on the meaning of porneia: 

Etymologically, it means "prostitution, harlotry, whoredom," being an abstract noun related to porne, "harlot," and to the verb porneuein, "to act as a harlot." … Though it is differentiated from moicheia in Mt 15:19, Mk 7:21-22, 1 Cor 6:9, Heb 13:4, it is used of a variety of sexual activity: 1 Cor 5:1 (incest), 6:13 (prostitution), 2 Cor 12:21 (parallel to akatharsia and aselgeia); see further Col 3:5 and Eph 5:3. 

In Acts 15:20, 29 (cf. 21:25) porneia is used, however, in a specific sense, since it is lined up with several dietary tabus, which early Gentile Christians, living in close contact with Jewish Christians (i.e., in predominantly Jewish-Christian communities), were being asked to avoid: "what has been sacrificed to idols, blood, and what is strangled." The letter of James to the local churches of Antioch, Syria, and Cilicia forbids, in fact, four of the things proscribed by the Holiness Code of Lv 17-18, not only for "any man of the house of Israel" but also for "the strangers that sojourn among them" (ûmin haggêr 'âser yâgûr bëtôkâm, 17:8). These were the meat offered to idols (Lv 17:8-9), the eating of blood (Lv 17:10-12), the eating of strangled, i.e., not properly butchered, animals (Lv 17:15; cf. Ex 22:31), and intercourse with close kin (Lv 18:6-18). 

He then points out that “For many commentators, porneia is simply understood as ‘adultery,’" (incorrectly, as shown above) and continues: 

By another group of commentators the word is understood in the generic sense of prostitution or harlotry, as it seems to be used in most of the Pauline passages quoted above. This meaning, while not impossible, would be imposing on the word a predominantly Pauline and Hellenistic meaning in a passage which may have more Palestinian and Jewish concerns. 

A third group of interpreters prefer to use the specific meaning of porneia that is used in Acts 15:20, 29, understanding it to mean illicit marital unions within the degrees of kinship proscribed by Lv 18:6-18. This is preferred because of the Jewish-Christian problem envisaged in Acts 15 and the concerns of the Matthean Gospel itself. Of these three main positions I think that the last-mentioned is the one to be preferred, since there is now further evidence from Qumran literature to support it.

Lv 18:6-18 is focused largely on not ‘uncovering the nakedness’ of blood relatives, i.e. prohibiting incest, but also sex with anyone else ‘near of kin,’ such as in-laws. A small number of these we now accept (e.g. marrying a sister-in-law after the death of your wife), but the obvious problem here with Fitzmyer’s reference to ‘illicit marital unions’ is that if a marriage was illicit was divorce even necessary? Num 30:6-15 suggests that if the marriage was illicit could it be annulled, in particular: 

And if she had at all an husband, when she vowed, or uttered ought out of her lips, wherewith she bound her soul; [Num 30:6]  And her husband heard it, and held his peace at her in the day that he heard it: then her vows shall stand, and her bonds wherewith she bound her soul shall stand. [Num 30:7]  But if her husband disallowed her on the day that he heard it; then he shall make her vow which she vowed, and that which she uttered with her lips, wherewith she bound her soul, of none effect: and the Lord shall forgive her. [Num 30:8]

It is also the case that a wife could re-marry after the death of her husband, and 1 Cor 7:12-15 provides that a marriage could effectively be annulled if an unbeliever left a husband or wife. There are several other reasons why modern marriages might be nullified, but no evidence that any of these applied when the gospels were written. One other possibility is that a marriage was considered annulled because of porneia that took place prior to a marriage that was only discovered later, perhaps when it was consummated. Overall it appears that the exception in Mt 5:32b // 19:9d is more restrictive than most people believe.

Wieland Willker (Mark, TVU 217) suggests that the differences between Mark on the one hand and Matthew / Luke are because Mk 10:11-12 reflects Roman law while Matthew and Luke reflect Jewish law: 

Under Roman law a woman could initiate divorce. Under Jewish law only a man could initiate divorce and the woman was required to get her husband to initiate it... Initiating divorce was impossible for a Jewish woman. 

In the New Testament Mk 10:11-12 is the most restrictive regarding divorce, with Mt 5:32b // 19:9b providing the exceptions for fornication noted above. There is no reason to suspect that the changes to Mk 10:11 seen in Matthew and Luke are not original in those gospels, but the ‘replacements’ of Mk 10:12 in Mt 5:32d // 19:9ef are not secure, and Willker (Matthew – TVU 42) makes the following points regarding the omission of Mt 5:32d in mss D, pc, a, b, d, and k: 

It is possible that the omission is a harmonization to Mt 19:9, where D, a, b also omit. It is also possible that the omission is a reflection of local law. The support for the omission is not good. Interestingly it is better in 19:9. But there it is probably due to a clear case of h.t. (μοιχᾶται - μοιχᾶται). Weiss (Textkritik, p. 183) thinks that this omission is due to carelessness. It is possibly due to h.t. αι – αι. He also thinks (Textkritik, p. 77) that the ὁ γαμήσας is a conformation to the ὁ ἀπολύων in the same verse. 

Parker (Living text, p. 84) notes that the short form "makes much simpler and better sense". The words sound like an afterthought, an addition. But the clumsy style may also be a reason for an omission. Metzger: "The omission … may be due to pedantic scribes who regarded them as superfluous." 

The reading by B, pc is clearly a harmonization to 19:9. Compare the discussion at 19:9. It is probable that the Byzantine [i.e. longer] reading is the correct text in 19:9. 

Willker also notes what Augustine wrote in De conjugiis adulterinis 1.10, ca. 420 CE, regarding Mt 5:32d, and the latter part of which also applies to Mt 19:8ef: 

It may well be that some of the manuscripts, both Greek and Latin, do not have those last words, namely, the one who marries a woman divorced by her husband commits adultery, as part of the Lord‘s sermon on the mount. I think this is because what this says is implied by the earlier statement, he causes her to commit adultery. How can the divorced woman become an adulteress without the man who marries her becoming an adulterer? 

Willker concludes that the longer reading (with the new husband committing adultery) is probably original, with the omission being likely due to a scribes eye skipping from one use of ‘adultery’ to the other, and then in TVU 253 he adds to the above when discussing the similar omissions (with at least nine different variant readings in a larger number of mss) in Mt 19:9: 

If it is a harmonization it is harmonized to Mt 5:32 (so Weiss) and not to Mk or Lk. Only the above variant γαμῶν might be a reminiscence, a secondary variant reading to Lk. Also the singular reading of 579 is a harmonization to Lk. 

It is quite possible that the clause has been omitted due to h.t (μοιχᾶται … μοιχᾶται). Note that P25, B, C*, N, f1 read μοιχευθῆναι - μοιχᾶται, but this is due to harmonization and therefore is no argument here (as Metzger wants it). This has rightly been pointed out by Michael Holmes. Compare his convincing discussion: "The Matthean Divorce Passages" JBL 109 (1990) 651-664. 

Noteworthy is that D, it, Sy-C conform the previous passage to 5:32 (παρεκτὸς λόγου πορνείας), but omit the last clause. D etc. omit the clause in both passages 5:32 and 19:9. In both cases h.t. is possible. 

Willker above refers to Michael Holmes, who comments that Lk 16:18 “apparently has had very little impact on the text of the other passages,” and then writes regarding Mk 10:11-12 and their (lack of) influence on Matthew and Luke, and vice versa: 

The variants in the Marcan passage are extremely complex and interesting, but also largely peripheral to a discussion of harmonistic variants. This judgment is based on three observations: (1) the contents of v. 12 are unique to Mark; (2) it does not appear to have affected any of the other three passages; and (3) the variants in vv. 11-12, with the possible exception of the omission of ἐπ’ αὐτήν in a few witnesses, do not reflect any apparent harmonizing tendencies. This last point is quite remarkable; I find it amazing that the “Matthean exception,” for example, is not known to occur in any Marcan manuscript. 

Holmes is pointing out that the texts of Mk 10:11-12 and Lk 16:18 (each only with variants unique to themselves) appear to have had no effect either on each other or on Mt 5:32 // 19:9, and that no known ms of Mark contains a fornication exception. Whatever has influenced the variants in the text of Mt 5:32 // 19:9, it is not found either in Mark or Luke. Additionally, Mt 5:32b // 19:9b do not appear to have influenced either Mk 10:11-12 or Lk 16:18. Whatever variants exist in these verses they have not had any influence on the text outside the gospel in which they are found. 

Holmes notes: “There are in Matt 5:32 four basic text-forms. Three of them … reflect varying degrees of harmonization to Matt 19:9,” and comments that “the exception clause … is without variation.” He then analyzes the “nine (possibly ten) different forms“ of Mt 19:9 and writes that “In the first half of 19:9” three out of the four variants “betray distinctive traces of the influence of 5:32.” After then analyzing the difficulties raised by a comparison of the variants in both Mt 5:32 // 19:9, Holmes summarizes his discussion of these variants as follows: 

When one analyzes the variants in both Matt 5:32 and 19:9 – two passages which clearly had a reciprocal effect on one another – the pattern of harmonization between them looks considerably different than if each is considered in isolation. In light of this new pattern, one is led to conclude that the original text of 19:9 is almost certainly that represented by the combination a + e [i.e. including 19:9ef] above, a combination which as a whole is found today almost exclusively among MSS of the Byzantine and (to a much lesser degree) secondary Alexandrian traditions. 

He ends with a comment on the variants (first those in Mark 14:70, which he had analyzed earlier as the lead-in to the verses discussed here): 

These conclusions will affect one’s view of the author’s meaning and method, of synoptic relationships, and of Matthean redactional activity… Further, the author of Matthew will now be viewed as having merely redacted a phrase found in his source, rather than having added or created this bit of explanatory detail. Similar conclusions likewise follow for Matt 19:9.

      Finally, these examples have demonstrated how atomization and isolation of the evidence can lead astray, particularly when dealing with harmonistic variants in parallel passages. A more comprehensive approach that takes into account simultaneously all the variants in all the parallels results in a more satisfactory and probable decision and explanation of both the text and the subsequent corruption of the passages examined. 

Willker also comments on the possibility of Mt 5:32 being harmonized to Mt 19:9 and Mt 19:9 being harmonized to Mt 5:32, but as harmonizations in both directions would be highly unusual it is worth considering whether either are actually likely, or even perhaps whether, as Willker comments regarding Mt 5:32, “the omission is a reflection of local law,” although he does not take this further. Of course, if this were so then it would be expected to apply equally to Mt 19:9, but the two verses are different. Unfortunately, Holmes and Willker both appear to have ignored another possibility, even though Holmes does seem to have almost touched on it: Neither Holmes (despite his mention of “synoptic relationships” and referring to Matthew “having merely redacted a phrase found in his source”) nor Willker appear to seriously consider the possibility that both Mt 5:32 / Lk 16:18 might actually depend on another, non-Markan, source, but if so then Q would be an obvious candidate. 

If this other source did not contain a parallel to Mk 10:12, but instead contained a parallel to (and so would be the source of) Lk 16:18 then we not only have a reason why aMatthew and aLuke did not also include a parallel of Mk 10:12 in Mt 5:32 / Lk 16:18, but also perhaps why aMatthew chose to largely ignore Mk 10:12 when writing Mt 19:9 and include a parallel to Mt 5:32b instead. As the text of Lk 16:18 is secure (it has no variants) it certainly has not been affected by any of the variants in Mark or Matthew, and so a different source for Lk 16:18 makes a great deal of sense. In this scenario it would then not be unusual for some scribes to not want to include Mt 19:9ef because it was different to Mk 10:12, but who did not feel they could add a parallel to Mk 10:12 instead, with the number of scribes wanting to harmonize Mt 5:32 to Mt 19:9 by excluding Mt 5:32d being much smaller because there was no Markan parallel with which it conflicted. 

Not surprisingly the suggestion of a non-Markan source has been previously considered. For example, Hawkins recognized that this doublet suggests “the use of two sources (probably Markan and Logian),” when he wrote: 

Mt A [5:32] corresponds with Lk and Mt B [19:9] with Mk… Probably therefore the latter have their source in the Marcan document, and the former in the Logia, though they are differently placed in Mt and Lk, as we shall see to be the case more often than not in the case of presumably Logian sayings.

As elsewhere Hawkins’ comments regarding a Logian source can be taken as suggesting that the double tradition verses Mt 5:32 / Lk 16:18 originated in a non-Markan second source, but not that it was necessarily the Logia, and on page 231 of The Biblical Theology of Divorce, Bruce Vawter comments that: 

Within the Gospel forms, even apart from the exceptive clauses in the two Matthean versions, development over against a presumed ipsissimum dictum is doubtless present in each case … 

He follows this with a footnote that assumes the Mark-Q hypothesis: 

It is usually argued that the Q form (Lk 16,18; Mt 5,32) is more primitive than the Marcan (Mk 10,llf., abbreviated by Mt 19,9). However, it is felt that Mt 5,32 has already added to the primitive logion by making the husband guilty of his wife's adultery merely by divorcing her, while on the other hand Mark and Luke may have explicated in the same direction by adding the condition of subsequent remarriage. Mk 10,12 may also be conceived as an adaptation to non-Jewish marriage customs. Cf. Rudolf Bultmann, The History of the Synoptic Tradition (tr. John Marsh; Oxford: Blackwell, 1963) 132, 148. For a somewhat different analysis, cf. Georg Strecker, Der Weg der Gerechtigkeit. Untersuchung zur Theologie des Matthäus (Göttingen: Vandenhoeck & Ruprecht, 1962) 130-132. 

Vawter is suggesting that Q 16:18 was more primitive than any of Mk 10:11-12, Mt 19:9 // 5:31-32, and Lk 16:18, and did not contain any wording “making the husband guilty of his wife's adultery merely by divorcing her.” This could have been similar to what Jesus commented on in Mt 5:31, but even if so it does not affect the ‘trajectory’ of the text on the Mark-Q hypothesis, in which both aMatthew and aLuke saw Mark and Q. In contrast, in Two Shipwrecked Gospels MacDonald writes: 

Matthew contains doublets that prohibit divorce: Matt 19:9 clearly redacts Mark 10:11-12, while the other instance apparently derives from elsewhere. Here again one finds the pattern of Matthew’s non-Markan doublet being prior to the Markan one. 

The Matthean Evangelist clearly was responsible for the insertion of the phrase “except for a charge of fornication,” because a similar insertion appears in his redaction of Mark 10:11 (19:9). Otherwise, Matthew’s non-Markan doublet represents an earlier stage of the tradition than Mark insofar as it says nothing about a woman divorcing her husband, which seems to be a concession to laws concerning divorce in the Greco-Roman world. Mark seems to have taken a saying against divorce and expanded it into a full-blown controversy between Jesus and the Pharisees, a secondary contextualization. 

Fitzmyer also sees another source, stating that: 

The Matthean form of the prohibition of divorce recorded here differs from Lk 16:18, not only because of the added exceptive phrase parektos logou porneias, but in two other ways: (a) it lacks the second phrase, Lk 16:18b, kai gamón heteran; and (b) it relates divorce itself, and not divorce and subsequent marriage, to adultery. 

Whereas the Lucan form of the saying also expresses a judgment about the husband's subsequent marriage, the Matthean form regards divorce itself as the cause of adultery (poiei autên moicheuthênai, lit., "makes her to be adultered"). This is, I suspect, a Matthean reformulation of the original "Q" saying, which is found in a more primitive form in Lk 16:18a-b. One reason for regarding the Matthean form as a reformulation is the immediate context in the Sermon on the Mount, where in v. 27 Jesus' antithesis equates even the lustful look of a man at a woman with adultery, an antithesis that lacks a parallel in either Mark or Luke. Hence it is most likely Matthew who relates divorce itself to adultery. Once again, the prohibition is stated from the viewpoint of the man, as in the Lucan form of the saying. 

Mk 10:11-12, Mt 19:9 // 5:31-32, Lk 16:18 – Synoptic Implications

Willker largely ignores the synoptic relationships and Holmes fleetingly mentions them but does not explore the issue, while Hawkins, Fitzmyer, Vawter and MacDonald all suggest that a non-Markan source is involved, but do not agree as to what it was. Hawkins suggests a Logian source, Fitzmyer references Q, Vawter suggests a Q that is “more primitive” than Mark, while MacDonald re-constructs a ‘lost gospel’ twice as long as Q that he calls ‘Q+.’ Notably neither of the latter two are actually Q, with the term therefore being used to suggest some kind of kinship, but without some of the restrictions imposed by the Mark-Q hypothesis. It is therefore worth exploring what these verses may be able to tell us regarding their ‘synoptic relationships.’ 

From the above comments it is clear that how you view the differences between these verses is affected by the assumed synoptic hypothesis, and with regard to marrying a divorced woman we basically have two different forms (with variants as referred to above). In particular: 

Mk 10:12 And if a woman shall put away her husband, and be married to another, she committeth adultery.

Mt 5:32d [and whosoever shall marry her that is divorced                                                  committeth adultery.] – not in some mss

Mt 19:9ef [and whoso           marrieth her which is put away                                                doth commit adultery.] - not in some mss

Lk 16:18d and whosoever    marrieth her that is put away from her husband              committeth adultery.

As Mt 5:32d / Mt 19:9ef / Lk 16:18d are essentially the same then depending on the synoptic hypothesis either Mk 10:12 has the earlier form, or Mt 5:32d / Mt 19:9ef / Lk 16:18d do. In Mark a woman who divorces her husband commits adultery if she re-marries, whereas in Matthew and Luke it is the man who marries a woman who has been divorced that commits adultery. The two are therefore quite different, and the NET Bible has this comment on Mk 10:12: 

It was not uncommon in Jesus’ day for a Jewish man to divorce his wife, but it was extremely rare for a wife to initiate such an action against her husband, since among many things it would have probably left her destitute and without financial support. Mark’s inclusion of the statement And if she divorces her husband and marries another, she commits adultery (v. 12) reflects more the problem of the predominantly Gentile church in Rome to which he was writing. As such it may be an interpretive and parenthetical comment by the author rather than part of the saying by Jesus, which would stop at the end of v. 11. As such it should then be placed in parentheses. 

If this is the case then Mt 5:32d // 19:9ef and Lk 16:18d can be seen as replacing a largely unnecessary statement in Mk 10:12 by an equivalent one regarding a man who marries a divorcee, in which case Mk 10:12 is clearly earlier than Lk 16:18d. However, this does not explain why in some mss Mt 5:32d and/or Mt 19:9ef have nothing corresponding to either Mk 10:12 or Lk 16:18d, basically removing Mk 10:12 but not adding anything in its place. The NET suggests that Mk 10:12 might have been “an interpretive and parenthetical comment,“ but the fact that equivalent text is omitted from Mt 5:32d and/or 19:9ef in some mss, but not from Mk 10:12 in the same mss, makes this very unlikely. It also makes it most likely that because Lk 16:18d is secure, in any mss containing both Matthew and Luke in which Mt 5:32d and/or Mt 19:9ef are omitted the omissions date back to an earlier stand-alone ms of Matthew, the scribes of which did not have Luke to hand when copying Matthew. 

The issue that is most in need of a solution here is that of the ‘competing’ variants in Mt 5:32d and 19:9ef, as in both verses essentially the same words are missing in a different selection of mss, only a few of which omit both Mt 5:32d and Lk 16:18d. Holmes notes that Mt 5:32 and 19:9 are “two passages which clearly had a reciprocal effect on one another“ and that Mt 19:9 most likely originally had the longer form “found today almost exclusively among MSS of the Byzantine and (to a much lesser degree) secondary Alexandrian traditions.” This conclusion is bolstered by the fact that the equivalent longer text is present in Luke in all mss extant for Lk 16:18d. Given this how do the various synoptic hypotheses account for this situation? 

As explained above it is very unlikely that Mt 19:3-8 is earlier than Mk 10:2-10, so instead assuming Markan priority, and on the MwQH (Mark without Q Hypothesis):

On this hypothesis aMatthew significantly changed what he saw in Mk 10:11-12, and aLuke then saw but ignored/rejected Mt 19:1-9 and most of Mk 10:1-12 and instead followed Mt 5:32 except for the fornication clause. Instead, on the Mark-Q hypothesis: 

Finally, the Mark-Q and MwEL (Mark with Early Luke) hypotheses both have Mark and a non-synoptic source for Matthew and Luke, but the major difference between them is that on the MwEL hypothesis Luke also has Matthew as a source, and so aLuke could also see Mt 5:32 // 19:9: 

Conclusions

The contortions required by aMatthew (and to a lesser extent aLuke) virtually rule out any synoptic hypothesis in which Matthew was first or had only one source. Although there are more complex hypotheses that can address some of these issues, the less complex hypotheses most likely to be correct all assume Markan priority and that aMatthew and aLuke had access to a suggested non-synoptic source, the most well known of which is Q. However, the Mark-Q hypothesis itself is not without its own problems, many of which are caused by the requirement that on that hypothesis aMatthew and aLuke did not know anything about each other’s gospel. Relaxing that requirement (e.g. as on the MwEL hypothesis) allows much greater freedom to solve problems such as the differences between Mt 5:32 and 19:9.

References

Dungan, D.L.: The Sayings of Jesus in the Churches of Paul: The Use of the Synoptic Tradition in the Regulation of Early Church Life, Philadelphia: Fortress, 1971, pp. 111-12

Fitzmyer, Joseph A, SJ: The Matthean Divorce Texts and Some New Palestinian Evidence , Weston School of Theology, Cambridge, Mass, 1976

Hawkins, Sir John Caesar: Horae Synopticae: Contributions to the Study of the Synoptic Problem

Holmes, Michael W: The Text of the Matthean Divorce Passages: A Comment on the Appeal to Harmonization in Textual Decisions , JBL 109 (1990) 651-664 

MacDonald, Dennis R: Two Shipwrecked Gospels: The Logoi of Jesus and Papias’s Exposition of Logia about the Lord , SBL 2012 

Turner, C.H.: Marcan Usage: Notes, critical and exegetical,on the second Gospel, 1924

Vawter, Bruce, C.M. The Biblical Theology of Divorce Kenrick Seminary St. Louis, Missouri. p 231, footnote 22

Westcott, Brooke Foss D.D. and Hort, Fenton John Anthony D.D.:  Westcott and Hort: The New Testament in the original Greek