It is beneficial first to note the modern history of Matt 17:21 in critical editions. The first in modern times to reject it was John Mill (1707). Yet none of the editors who immediately followed were persuaded, such as Bengel (1734), Wettstein (1751), Matthäi (1788), Griesbach (1796), Lachmann (1831), Scholz (1830), and Tregelles (1857), though he bracketed it. Only after the discovery of Codex Sinaiticus in 1844 did the scholarly consensus begin to change, although Lachmann's edition of 1842 anticipated the change and appears to be the first edition since Mill to have rejected it. Tischendorf's later editions rejected the verse, and with few exceptions since, most major editions have omitted it, including Westcott-Hort (1881), Nestle (189
, and von Soden (1911-1913), although it was bracketed by Merk (1938) and included by Bover (1943) and Vogels (1955).
Below follow the important opinions of two of the early masters of NT textual criticism, J. A. Bengel and J. J. Griesbach:
Bengel: "For this portion [i.e., Matt 17:21] actually especially answers the question. Mill adds Eusebius' canon, in which Mark proceeds alone at this place: but this method also removes verses 19 and 20. This canon not very accurately puts Mark alone, since it had already united Matthew with another place in Luke. In the end, some of the words are Mark's, others are Matthew's: therefore the words have not been brought over to this place from there [i.e., Mark 9:29]" (Johann Albrecht Bengel, Apparatus criticus ad Novum Testamentum [ed. Philipp David Burk; 2d ed.; Tubingae: sumtibus Io. Georgii Cottae, 1763], 125).
Greisbach: "Omitted by some and rejected by Mill, it more firmly rests on the consensus of the oldest Alexandrians with the most ancient Westerns, with which even the remaining families of manuscripts agree. It does not appear very probable that it has been brought in to this place from Mark; for instead of ουκ εκπορευεται he has εν ουδενι δυναται εξελθειν, neither has there been any reason apparent why an interpolator would have changed it into the former. Eusebius indeed assigned the parallel passage of Mark for us (Mark 9:28-29) to his tenth canon [i.e., "X"], in which the pericopes singular to only one Evangelist are specified, which more preferably ought to have been assigned to the sixth canon [i.e., "VI"], which has been set up for the places common to Matthew and Mark. And so from here Mill thought it could be deduced that our verse 21 had not been read by Eusebius. But far nearer are other reasons why our passage is missing from the sixth Eusebian canon. Without doubt the Ammonian section of Matthew (ροε) encompasses whatever lies between the end of our verse 18 and the beginning of verse 22. Therefore, Eusebius wrongly assigned this section to his fifth canon which exhibits the parallel places of Matthew and Luke, and this he connected, not very suitably, with the σ section of Luke, that is, with Luke 17:5-6. Having proceeded from here to the sixth canon, it was this one which he repeatedly was unable or unwilling to produce in common. The rest I do not pursue, such as what in the past could have presented the occasion for omitting this verse. There are those who suspect that at some time it was omitted in some manuscripts of the Latin version, which others followed in turn. But truly in my mind I cannot comprehend why is it that the Latins should be judged more probably and earlier than the Greeks, either to have caused this verse to leap over or to have slashed it away. Not to mention that not only the Vulgate but also the oldest manuscripts of Italy [i.e., the Old Latins] (with the sole exception of Corbeiensis I [i.e., ff2]) have preserved this verse, with Hilary of Poitiers and Juvencus joining in support" (Johann Jacob Griesbach, Commentarius criticus in textum graecum Novi Testamenti [2 vol.; J. C. G. Goepferdt, 1798, 1811], 1:146-7).
I now present my own argument, which will focus first on external and then on internal considerations.
I. EXTERNAL CONSIDERATIONS:
1. Only 10 Greek manuscripts of the 1700 or so omit the verse.
2. We know of only three (ℵ* B 0281) Greek manuscripts that omitted the verse in the first 700 years of Matthew's transmission, while eight manuscripts from the same period (ℵ2 C D E L O W Σ) include it.
3. Only two Old Latin manuscripts (e ff1) omit the verse, one from the fifth and one from the sixth century. On the other hand, the majority of the OL manuscripts (a aur b c d f ff2 g1 l n q r1) include the verse, including the oldest of them all, Vercellensis (a), written about the same time as Vaticanus and Sinaiticus in the fourth century. In addition are the important Veronensis (b), Corbeiensis (ff2), and the St. Gallen manuscript (n), all from the fifth century, as well as the important Colbertinus (c), the combination of all of which indicates a second or third century origin and thus a weighty early witness for the inclusion of the verse.
4. The two important Old Syriac manuscripts (s c), generally thought to be produced in or near Egypt, omit the verse, but all the manuscripts of the mainstream Syriac version (p) include the verse.
5. The Egyptian Coptic version is split, with the mainstream early dialect (s) and part of the later dialect (b-pt) omitting the verse, while the early and important Middle Egyptian dialect (mae) includes the verse along with part of the later dialect (b-pt).
6. The Diatessaron from the second century is said to contain the verse.
7. While apparently only Eusebius from the fourth century attests to the verse's omission (cf. Griesbach's note above, as such is not certain), most early fathers from nearly every important geographic center from which we have evidence attest to the verse's inclusion: Origen, Asterius, Hilary, Basil, Ambrose, Chrysostom, Jerome, Juvencus, and Augustine.
8. The Vulgate contains the verse, and it is said that Jerome in the fourth century used the oldest and best manuscripts of his own day (both Greek and Old Latin) to form the basis of his text. Thus it is fairly safe to say that the Vulgate in Matthew, when confirmed by the consensus of the Old Latin manuscripts, witnesses to a text current in the second and third centuries at the latest.
9. To sum up the external evidence: we have early, impressive, and manifold witnesses from the second century on, from every major geographic location of the early church, from nearly all of the early Fathers (not to mention later ones) to which we have access, and major representatives from every major early version (Old Latin, Coptic, Syriac), all strongly attesting the inclusion of the Matt 17:21. On top of all these is the fact that some 99.4 percent of all Greek witnesses contain the verse.
II. INTERNAL CONSIDERATIONS:
10. Some thought to omit the verse due to the similarity of Matt 17:20 with Luke 17:6 in order to bring Matthew's account in line with Luke's, where the verse in question is not recorded.
11. Some thought Matt 17:21 contradicted 17:20, in that if some demons do not come out except through prayer and fasting, then sometimes even faith like a mustard seed is not enough. The statement was thought by some to be difficult and thus removed.
12. Perhaps the strongest reason is that there was an attempt by some to remove certain texts that might pharisaically be used to promote certain ascetic practices (cf. also the omission of "and fasting" from Mark 9:29, and the omission of "to fasting and" in 1 Cor 7:5).
13. The wording of Matt 17:21 in Greek is completely different from that of Mark 9:29, indicating that Matt 17:21 did not come by means of scribes from Mark 9:29 (cf. also the notes of Bengel and Griesbach above):
Matt 17:21: τουτο δε το γενος ουκ εκπορευεται . . .
Mark 9:29: τουτο το γενος εν ουδενι δυναται εξελθειν . . .
14. Given the differences between Matt 17:21 and Mark 9:29, how is it that none of the 1700 or so manuscripts witnesses precise verbal harmonization to the Markan passage, if harmonization was so prevalent among scribes?
15. How did such a verbally different version of Mark 9:29 get into all the witnesses discussed above with such little meaningful variation unless this distinct form was present in the archetype of all those witnesses to begin with? In other words, what one must say is that a single scribe or editor alone managed to put a different version of Mark 9:29 into Matthew and that this verse managed to be copied precisely by all later scribes, covering all geographic areas, versions, fathers, etc. Who was this scribe, and how did he manage to do this? What is the basis of faith in the safety of our NT text in other places if a single force was able to pull off an alteration of this magnitude here?
16. It is practical to suggest that changes to the original text that occur some time later than the original text will not be able to overcome the multiplying copies of the original in all of the following categories simultaneously: Greek manuscripts; early versions, early fathers. We see, in fact, that the consensus of all of these strongly attests to the inclusion of Matt 17:21.
17. It is practical to suggest that various reasons, such as those mentioned in #10, #11, and especially #12, might have caused the omission of Matt 17:21 (the evidence indicates Egypt in the third century after Origen, since even he attests to the verse's inclusion). But whenever and wherever such occurred, it was unable to overcome the already multiplying copies of Greek, Latin, Old Latin, Syriac, Middle Egyptian, and other manuscripts, in addition to those older and more accurate copies in use by the vast majority of the early fathers.
Jonathan C. Borland
