‘Father Forgive Them’ – The Variant in Luke 23:34a

This is the last of a series of blog post [2018] on some of the textual variants found in the Passion narratives. The series discusses the rationale behind the text adopted in the Greek New Testament as Produced at Tyndale House in (1) Mt 27:16,17, (2) Mt 27:49, (3) Mk 14:30, 49, 72a, 72b, (4) Lk 22:31, (5) Lk 22:43-44, (6) Lk 23:34.

The previous variant that we discussed (Lk 22:43-44) was substantial and important. It makes quite a difference how Jesus is portrayed by Luke whether or not the episode of the strengthening angel and the sweat like drops of blood is present. The final variant of this series is, in my view, even more important and one with considerable theological ramifications. Come to think of it, I am not sure if there are many variants that have a bigger impact on New Testament Christology than Luke 23:34a.

It concerns the presence or absence of the following words

ὁ δὲ Ἰησοῦς ἔλεγεν· πάτερ, ἄφες αὐτοῖς· οὐ γὰρ οἴδασιν τί ποιοῦσιν.
And Jesus said, “Father, forgive them, for they do not know what they are doing.”

None of the other evangelists has any reference to Jesus prayer of forgiveness for those who are crucifying him; the presence of these words make a unique contribution, their omission changes Luke’s narrative considerably. And just to draw the modern battle lines: the THGNT has these words as part of the main text, though signalling the problems with a diamond in the apparatus. NA26-28 has these words in white square brackets, claiming that these words are certainly not part of the original text of the gospel but have been inserted at an early stage.

Here is the Greek evidence, and as far as the omission goes I believe it is complete:

omit P75 ℵ2a B D* W Θ 070 579 1241
text ℵ* ℵ2b A C D3 K L N Q Δ Ψ 0211 f1 f13 33 158 700 713 892 1071 l844 Maj

[IGNTP-Luke mentions 0124, but that witness is now combined with 070.]

There are at least two similarities between this textual variant and Lk 22:43-44, the angel and the sweat like drops of blood.
The first is found in the supporting evidence. This was the evidence for the omission of Lk 22:43-44

omit 22:43-44 P75 ℵ2a A B N R T W 0211 f13(but adds after Mt 26:39, as does a later corrector of C) 158 579 713 1071* l844.

The witnesses that omit at both places are P75 ℵ2a B W 579. The ones that omit in 22:43-44 and not at 23:34a are A N 0211 f13 158 713 1071 l844 (R and T are only extant at the first place) and those that omit 23:34a but not 22:43-44 are D Θ 1241 (070 only extant at 23:34a). The five witnesses that omit at both places form something of a solid core, it is not remarkable to see P75 B W 579 together (and on their combined testimony alone I am prepared to consider any reading quite seriously).

A second similarity is the nature of the longer reading. Neither in 23:43a or in 22:43-44 is there a clear source of influence. Yet there are plenty of thematic links with the Lukan corpus. Stephen’s words in Acts 7:60 (κύριε, μὴ στήσῃς αὐτοῖς ταύτην τὴν ἁμαρτίαν ‘Lord, do not hold this sin against them’) convey a similar sentiment as 23:34a but the wording is quite different. We have a shared theme rather than a source of harmonisation. The same is true for the shared notion of ‘not knowing what they are doing’ in 23:34a and Acts 3:17 (κατὰ ἄγνοιαν πράξατε ‘you acted in ignorance’). One could even argue that Acts 3:17 presupposes something like Luke 23:34a. Yet again, it seems unlikely that Acts 3:17 provided the the wording that we find in our passage.

So what are the arguments for or against?
• The main argument against the originality of 23:34a is that it is left out in a part of the earliest evidence.
• If these words were original, there does not seem to be a good motivation for leaving it out.
• A reconstructed background is that the words in question may be an agraphon (Metzger’s Commentary) which is subsequently made part of the gospel-tradition for numerological reasons as it brings the number of sayings on the cross up to seven (Whitlark and Parsons).

The arguments in favour of printing the passage are:
• The shorter text can be explained as a harmonization, this time by omission. And there are parallels elsewhere in the early manuscripts, and especially so in the Passion narratives. We have seen harmonization in the early witnesses in Matthew 27:49, and harmonization by omission in the variants in Mark 14, and I believe also in the two earlier discussed variants in Luke 22. And for those who accept the reading ‘Jesus Barabbas’ in Matthew 27:16, 17 (which I don’t) there is another example of harmonization by omission.
• Thematically and theologically it fits the Lukan writings.
• Metzger in his Textual Commentary mentions the destruction of Jerusalem as an event that seems in contradiction to Jesus’ prayer for forgiveness. One could go one step further and suggest that the omission is an anti-Jewish variant (in the sense that they should not be forgiven). However, as with many attempts to find a social or theological background to a textual variant, such reconstruction is rather speculative and perhaps more indicative of our desire to have a story behind a textual variant than that it provides us with a real argument. Admittedly, anti-judaism is not a strange sentiment in early Christianity (see Eubank who unpacks this line of argument).

For these reasons the Tyndale House Edition presents the text ‘Father forgive them, for they do not know what they are doing’ as part of the original text of Luke. There seems to be enough going on in the P75 B-03 group to throw some doubt over their testimony in the big variants in the Passion narratives. The omission has – what I would call – strong external support. But this is exactly why textual criticism cannot be reduced to choosing an algorithm or preferred group of manuscripts. The reality of historical transmission is more complex and messier than any simple solution.

Nathan Eubank, “A Disconcerting Prayer: On the Originality of Luke 23:34a”, Journal of Biblical Literature 129 (2010): 521-36

Jason A. Whitlark, and Mikeal C. Parsons, “The ‘Seven’ Last Words: A Numerical Motivation for the Insertion of Luke 23.34a”, New Testament Studies 52 (2006): 188-204 (see a discussion of this article on the ETC blog here)


Luke 22:43-44. Is the Angel and the Sweat like Drops of Blood an Early Addition?

This post is part of a series [2018] on some of the textual variants found in the Passion narratives. We will discuss the rationale behind the text adopted in the Greek New Testament as Produced at Tyndale House in (1) Mt 27:16,17, (2) Mt 27:49, (3) Mk 14:30, 49, 72a, 72b, (4) Lk 22:31, (5) Lk 22:43-44, (6) Lk 23:34.

Within the various variants found within the Passion narrative the variant found in Luke 22:43-44 is the most substantial. It is also one of the passages, together with Luke 23:34, where the Tyndale House Edition differs radically in its assessment from the NA26 – 28 editions. The THGNT prints these verses as part of the main text and signals the difficulties with the ‘diamond of uncertainty’. The Nestle-Aland editions enclose these verses in white square brackets ⟦ ⟧, indicating that, according to NA28, 55*, ‘the enclosed words, generally of some length, are known not to be a part of the original text. These texts derive from a very early stage of the tradition, and have often played a significant role in the history of the church (cf. Jn 7,53 – 8,11).’ The German version of the Introduction uses the term ‘mit Sicherheit’ (10*).

So what about Luke 22:43-44? These are the words under contention:

ὤφθη δὲ αὐτῷ ἄγγελος ἀπ᾽ οὐρανοῦ ἐνισχύων αὐτόν. καὶ γενόμενος ἐν ἀγωνίᾳ ἐκτενέστερον προσηύχετο. ἐγένετο δὲ ὁ ἱδρὼς αὐτοῦ ὡσεὶ θρόμβοι αἵματος καταβαίνοντες ἐπὶ τὴν γῆν.

“An angel from heaven appeared to him and strengthened him. And being in anguish, he prayed more earnestly, and his sweat was like drops of blood falling to the ground.” (NIV)

There is good evidence on both sides (note that the apparatus of THGNT failed to include 0171 in support for the presence of these words, a genuine error). Here is the Greek evidence, and the evidence for the omission is as complete as I can get—NA28 and IGNTP-Luke combined:

text: ℵ* ℵ2b D K L Q Δ Θ Ψ 0171(vid ]θρον[…]ς κα̣τ̣[.]βαι[…]π̣ι την γην) 0233 1071c 1424 Maj

omit: P75 ℵ2a A B N R T W 0211 f13(but adds after Mt 26:39, as does a later corrector of C) 158 579 713 1071* l844.

[Sinaiticus: The correction hooks and dots that were added by ℵ2a were later erased by ℵ2b]

[Tischendorf’s transcript and R-027]

In addition, a number of manuscripts that have the text obelized in the margin, which can indicate uncertainty whether to include the words or not. IGNTP Luke mentions Δ 230 1295 1424.


Depending on who you read, often the testimony of P69 is given as supporting the omission. We did not cite it as such in the Tyndale House Edition for the following reason. In P69 not just verses 43-44 are missing, but apparently also verse 42 (see for yourself here at the NT.VMR). So for all practical purposes, P69 misses our additional words because it is missing a larger section of text. Below I may suggest that P69 still might be relevant, but only in such a speculative way that it should not clutter an apparatus.

We have patristic references to this passage (and I recommend the discussion of Blumell in the TC journal, if you want to read more). The reference in Justin shows that the actual episode was known in the mid 2nd century.

The first thing with any textual variant is to see if the variant can be explained by some sort of scribal habit, things that can go wrong in the process of copying. Though text can drop out for any random reason, there is no ready scribal habit to explain its omission here. So we need to do some old-fashioned text-criticism here.

Comparable variants
Before going into specific internal reasons for the omission or addition of these words, it is good to ask the question if there are any comparable cases. Perhaps the following are the most pertinent ones:
• It is difficult not to think of Jn 5:3-4, where we have an explanatory comment added to the text concerning another angel, who disturbs the water so that the first ill person to reach it is healed. Though the similarity between our variant unit and that in John are clear (roughly similar length, involves an angel), there are also differences. In John the expansion fills a perceived gap in the story, but here in Luke the variant interrupts rather than explains the flow of the narrative.
• Mt 27:49, addition of piercing the side of Jesus, taken from Jn 19:34 – dealt with in a previous blog post. The cautionary tale of this variant is that even though the two oldest manuscript have the text (and some good additional support), one cannot automatically follow the rule that the oldest attested reading is therefore the best.
• Lk 22:31 small addition in a transition – dealt with in a previous blog post
• There is something of a parallel with Jn 7:53 – 8:11, the story of the woman caught in adultery, in that this story also appears in different locations (e.g. in f13 between Lk 21:37 and 22:1. This same family has also our passage moved.)
• Lk 23:17 (explanation that Pilate had to release a prisoner) – again an explanatory gloss.
• Lk 22:19b-20 omission only found in D-05 (and versions) – harmonization by omission
• Lk 22:64 small addition – influence from parallels
• Lk 23:34a the words ‘And Jesus said, Father forgive them for they do not know what they are doing’. Omitted by quite a few of the same manuscripts as omit the text here.

It seems that there is no clear parallel except for Lk 23:34a, and we deal with that variant in the next post. Unlike Jn 5:3-4 and Lk 23:17, our passage does not add an explanation, and neither is it based on a parallel elsewhere. Its varying location might be thought of as an argument against originality, but of course it is only an argument against its presence here in Luke in the source of that particular textual tradition. Wandering passages may be suspect, but are thereby not automatically condemned (1 Cor 14:34-35, anyone?)
If anything, the variants listed above raise the possibility that just as the earliest manuscripts have a big harmonization by addition in Mt 27:43, and a harmonization by omission in Mark 14, so it is at least possible that some of the early manuscripts also have harmonized by omission here in Lk 22:43-44. Since the story of the sweat like drops of blood and the comforting angel is not found in the parallel accounts of the passion, a part of the tradition which is not known to be collecting bits and bops anyway, left the passage out.

Scholarly opinion
Nothing as nebulous as the consensus of the scholarly world. When somebody has published an article, and nobody writes a rebuttal in 10 or 20 years, it is a fallacy to assume that therefore everyone agrees with you (people who know the literature on this variant may recognize this). There have been voices in favour of its authenticity and also against. The NA26 – 28 editions are clear though, they do not regard these words as original by Luke as original.

So what could be the reasons for regarding the words as an addition, if they are not original to Luke?
1) Adding details – Embellishment of an existing narrative. If these words were part of the common, popular memory of the Passion narrative, they were bound to find their way into the biblical text.
2) They interrupt the flow of the narrative, there is no need for this heightening description of Jesus’ agony.
3) Both the appearance of an angel and the sweat like drops of blood have a folklore feel about them and are unnecessary supernatural expansions.
4) Better too much than too little. In cases of doubt, leave the words in.
5) Ehrman and Plunkett: These words were added as an anti-docetic improvement of the text.

What could be the reasons for seeing these words as original?
1) There is no obvious explanation for their origin other than that they are part of the original composition.
2) Luke has an angel motif throughout his writings. From the announcement of Jesus’ birth, all the way through Acts, finishing with an angel encouraging Paul before the shipwreck. Thematically this passage fits Luke.
3) As learned above, this could be a case of bringing Luke’s account into line by omitting an unknown episode (harmonization by omission).
4) The words are original, but were omitted because of theological embarrassment – Jesus is portrayed as too weak (a crude summary of Blumell’s argument).
5) The words are original, but were omitted because of an anti-Gnostic improvement of the text (Clivaz).

Without doubt there are many refinements and additions to these two sets of arguments, but this a blog post, not a full-blown review article.
I am wondering, though, what we can learn from P69. Often this papyrus is dated quite early, to the third century. It omits 22:42-44, uniquely so. Whether or not this was done by the scribe or copied from its exemplar is irrelevant, what is interesting though is that it raises the question that here we have an omission that clearly is secondary, nobody is going to defend this larger omission as being original. This could be because P69 copied a text without 43-44 and happened to omit another verse. Or it may have copied a text in which 43-44 were marked for deletion and simply deleted too much. Or, and this I find the most interesting possibility, P69 omitted roughly the same passage as is omitted in other manuscripts, and for similar reasons (whatever they may have been). Independently, P69 may have done the same (by and large) as was done elsewhere (and perhaps also earlier) in removing a section that was too unlikely to be correct.

In the end though, on one hand there is the relatively simple observation that manuscripts from any age and affiliation do harmonise, and I am fine to go with this. On the other hand there is the subsequent, more fraught exercise to come up with possible theological motivations behind an addition or omission. Therefore I am perfectly content to print 22:43-44 as part of the main text and signal the difficulties by means of a diamond in the apparatus.

Some literature:
Blumell, Lincoln H. “Luke 22:43–44: An Anti-Docetic Interpolation or an Apologetic Omission?” Textual Criticism 19 (2014): 1-35 (you can find it here).

Clivaz, Claire. “The Angel and the Sweat Like “Drops of Blood” (Lk 22:43–44): P69 and f13.” Harvard Theological Review 98, no. 4 (2006): 419-40 [and also her monograph on the issue – not for the fainthearted: L’ange et la sueur de sang (Lc 22,43-44): ou, Comment on pourrait bien encore écrire L’histoire (Biblical Tools and Studies 7. Leuven: Peeters, 2010)]

Ehrman, Bart D., and Mark E. Plunkett. “The Angel and the Agony: The Textual Problem of Luke 22:43-44.” Catholic Biblical Quarterly 45 (1983): 401-16.


‘And the Lord said’ – A Variant in Luke 22:31

This post is part of a series [2018] on some of the textual variants found in the Passion narratives. We will discuss the rationale behind the text adopted in the Greek New Testament as Produced at Tyndale House in (1) Mt 27:16,17, (2) Mt 27:49, (3) Mk 14:30, 49, 72a, 72b, (4) Lk 22:31, (5) Lk 22:43-44, (6) Lk 23:34.

A variant is a variant, but not all variants have the same effect on our understanding of the text. Today’s variant has only limited effect; it will not change anything within Luke’s narrative. It concerns the introduction to direct speech that we (do not) find in Luke 22:31:

Εἶπεν δὲ ὁ κύριος· Σίμων Σίμων, ἰδοὺ ὁ σατανᾶς ἐξῃτήσατο ὑμᾶς τοῦ σεινιάσαι ὡς τὸν σῖτον.

Are the word εἶπεν δὲ ὁ κύριος part of the text or not? And to me this is quite a thorny problem that I haven’t been able to resolve yet. Its impact on the text is not big, but if the text adopted in the THGNT is correct, we may have a pattern in a group of early manuscripts that is relevant for the two big variants later in Luke’s passion narrative.

First the Greek external evidence.

Omit: P75 B L T 1241 2542c l1231

Text: ℵ A D K N Q W Γ Δ Θ Ψ, all minuscules

Obviously the support for the shorter text is early and, barring any counter-arguments, my first inclination would be to go with it. And this is what almost everyone since Westcott-Hort till NA28 has done. However, the longer reading (in critical editions adopted only by Vogels and in brackets by Tregelles) has early witnesses at its side, from the fourth century onwards.

Before looking at any further arguments, what is the context of our passage? From 22:25 onwards Jesus is talking, first addressing the issue of who is the greatest (22:25-27), then moving to the promise of eating at the table in the Kingdom and judging the twelve tribes, which is introduced with stating that the disciples have stayed with him in his trials. From there it is a relatively small move to addressing Simon Peter and warning him about the trials Peter is about to face.
Our variant provides a separate, explicit, framing of the warning to Simon Peter that starts with the address ‘Simon, Simon’.

For what reasons could the introduction be a secondary development?
• The abrupt change of addressee called for a marker to signal this change, the variant supplied this.
• A new kephalaion starts at 22:31 and expansions such as ‘and the Lord said’ occur frequently at these breaks, especially at the start of a lectionary.
• In light of the synoptic parallels it seems that the whole of Jesus’ speech comes from two occasions, and therefore the variant is introduced to separate these out.
• The reference to Jesus as κυριος betrays it secondary origin; it is the language of the introduction to lections more than how the evangelists describe Jesus in narrative.

What about regarding the words as original?
• The introduction to Jesus’ direct speech is superfluous, he is already speaking, and therefore the words provide an unnecessary disturbance which led to their removal.
• Though it is true that the start of a kephalaion is a strong argument against originality of the longer reading, ℵ-01 has it, and there is no indication that the kephalaia were already part of the textual tradition in the fourth century.
• The reference to Jesus as κυριος within authorial narrative (so, not in quoted speech) is found elsewhere in Luke: 7:13; 10:1, 41; 11:39; 13:15 etc.

For me this is a real tricky variant. If someone could demonstrate to me that kephalaia were around by the time ℵ-01 was produced and could have influenced its text, the case for switching around the text and variant gets stronger. As things stand now, I would not put it past the P75 B-03 cluster to give us a text that is a bit too clean, and therefore the Tyndale House Edition prints the words in the main text, though the variant has the ‘diamond of uncertainty’.


A Rooster Crowing Once and Twice – Mark 14

This post is part of a series [2018] on some of the textual variants found in the Passion narratives. We will discuss the rationale behind the text adopted in the Greek New Testament as Produced at Tyndale House in (1) Mt 27:16,17, (2) Mt 27:49, (3) Mk 14:30, 49, 72a, 72b, (4) Lk 22:31, (5) Lk 22:43-44, (6) Lk 23:34.

The scene is familiar. The disciple Peter responds vehemently to Jesus’ announcement that all his disciples will fall away (οὐκ ἐγώ). Jesus responds that Peter will deny him three times in the coming night. But three times before what? Matthew has simply ‘before the rooster crows’. Luke says that the rooster will not crow until you have denied me three times. But what does Mark have? The text should read ‘before the rooster crows twice’, but there are a surprising number of variants relating to this number, both in Mk 14:30 and the subsequent unfolding of events.

14:30 πρὶν ἢ δὶς ἀλέκτορα φωνῆσαι – omit δίς (with some other variation)
‘before the rooster crows twice’ – omit ‘twice’

14:68 καὶ ἀλέκτωρ ἐφώνησεν – omit
‘and the rooster crowed’ – omit

14:72a καὶ εὐθὺς ἐκ δευτέρου ἀλέκτωρ ἐφώνησεν – omit ἐκ δευτέρου
‘and immediately the rooster crowed for the second time’ – omit ‘for the second time’

14:72b πρὶν ἀλέκτορα φωνῆσαι δίς, τρίς με ἀπαρνήσῃ – omit whole phrase or omit only δίς.
‘before the rooster crows twice, you will deny me three times’ – omit whole phrase or omit only ‘twice’.

All these variants are related to one another in that each of them moves away from a double crowing, either by removing δίς (‘twice’), or removing one of the two mentions of the bird’s noisy activity. Is this shared concern found in related manuscripts? More or less. Here is the Greek evidence:

14:30 πρὶν ἢ δὶς ἀλέκτορα φωνῆσαι

omit ἢ δίς ℵ C*(has ἤ) D W 579
πρὶν ἀλέκτορα δὶς φωνῆσαι Θ f13 565 700 2542
πρὶν ἢ ἀλέκτορα φωνῆσαι δίς C2
text A B K L N Γ Δ Ψ 083 f1 892 1241 1424 Maj

14:68 καὶ ἀλέκτωρ ἐφώνησεν

omit ℵ B L W Ψ* 579 892
text A C D K N Γ Δ Θ f1 f13 33 565 700 1241 1424(και ευθεως) 2542 Maj

14:72a καὶ εὐθὺς ἐκ δευτέρου ἀλέκτωρ ἐφώνησεν

omit ἐκ δευτέρου ℵ C*vid L 579
text A B D K N W Γ Δ Ψ f1 f13 33 565 700 892 1241 1424 2542 Maj

14:72b ὅτι πρὶν ἀλέκτορα φωνῆσαι δίς, τρίς με ἀπαρνήσῃ

omit completely D
omit δίς ℵ C*vid W Δ 579(απαρνηση με)
… φωνῆσαι, ἀπαρνήση με τρίς 2542 (so does omit δίς)
… φωνῆσαι δίς , ἀπαρνήση με τρίς A K N Γ f1 f13 33 1241 1424 Maj
… δίς φωνῆσαι, τρίς με ἀπαρνήση B
… δίς φωνῆσαι, ἀπαρνήση με τρίς Θ 565 700
text C2vid L Ψ 892

Note on 14:72b: Though I have presented the seven readings as a single variation unit, I take it that there are in fact three units: (a) the wholesale omission, (b) [δις] φωνησαι [δις], and (c) the word order of τρις με απαρνηση. The last one is the least interesting, but it may not be completely unrelated to what goes before.

So what order can we bring to this flood of data?
ℵ C* D W 579 have no reference to Jesus’ announcement of a double crowing, both in 14:30 and 14:72b the word δίς (or in D-05’s case the whole clause) is lacking. These manuscripts differ though in how they describe the actual crowing. Both ℵ-01 and minuscule 579 are consistent and have only the second crowing and without the words ἐκ δευτέρου, W-032 has only the second crowing but with ἐκ δευτέρου (so still gives the game away), and C-04 has the rooster crowing twice, though the second time without ἐκ δευτέρου. D-05 has two crowings, and the second with ἐκ δευτέρου.
This leaves us with some manuscripts that have the double announcement δίς but leave out the first crowing: B L Ψ 892. Of these B Ψ and 892 still have ἐκ δευτέρου at the second crowing, but L-019 lacks the phrase. In light of the previous paragraph I take this as a partial spill over from the concerns reflected in the text of the other witnesses.

The origin of all this confusion is influence from the parallel accounts in Matthew and Luke: only Mark mentions the double crowing. Each of the four variant units has then a similar explanation, which is that it brings Mark in line with the number of crowings found in the other gospels. That there is a singular motive behind these four variants gains further force in that the main manuscripts involved are not unrelated to one another (not in strict statistical terms, but in the ‘you should not be surprised to see these together’ sense).
Of course, we could take each of these variants in isolation, taking the atomistic approach encouraged by the piecemeal way of presenting an apparatus unit by unit. Thus, on the basis of external evidence alone, I can see the case against the text reading in 14:68. On the other hand, knowledge of the manuscripts and the manuscript tendencies described above causes me to adopt the reading as printed in the THGNT.
If this is correct, we have an excellent example in which harmonisation takes place by omitting text, and that already in some of the very early manuscripts (ℵ C* D and also B is involved once) — a useful phenomenon to keep in mind when dealing with some of the big variants in the Passion Narrative in Luke.


Matthew 27:49 Was Jesus Pierced before His Death?

Variants in the Passion Narrative (2)

This post is part of a series [2018] on some of the textual variants found in the Passion narratives. We will discuss the rationale behind the text adopted in the Greek New Testament as Produced at Tyndale House in (1) Mt 27:16,17, (2) Mt 27:49, (3) Mk 14:30, 49, 72a, 72b, (4) Lk 22:31, (5) Lk 22:43-44, (6) Lk 23:34.

In many ways the following variant is salutary, as it will correct any slavish tendency to think about the ‘earliest and best’ attested reading as an almost pleonastic collocation. The ‘best’ reading is not always the ‘earliest’, there may be good reasons not to follow the earliest manuscripts, and our variant is a good example. And, yes, we could, and probably should have mentioned the variant in the Tyndale House Edition, but we did not.

Mt 27:49
οἱ δὲ λοιποὶ ἔλεγον· ἄφες, ἴδωμεν εἰ ἔρχεται Ἠλίας σώσων αὐτόν.
But the others said, “Wait, let us see whether Elijah will come to save him” (ESV).

The early variant ἔλεγον / εἶπαν should not distract us here, what is interesting is the addition we find after the final word of this verse:

ἄλλος δὲ λαβὼν λόγχην ἔνυξεν αὐτοῦ τὴν πλευρὰν καὶ ἐξῆλθεν ὕδωρ καὶ αἷμα.
And someone else, taking a spear, pierced his side and there came out water and blood.

Those who know their gospels will suspect that we might have influence from one of the other gospels, and indeed, in John 19:34 we have (without relevant variation):

Jn 19:34
ἀλλ᾽ εἷς τῶν στρατιωτῶν λόγχῃ αὐτοῦ τὴν πλευρὰν ἔνυξεν, καὶ ἐξῆλθεν εὐθὺς αἷμα καὶ ὕδωρ.
But one of the soldiers pierced his side with a spear, and at once there came out blood and water. (ESV)

The addition in Mt 27:49 and the undisputed text in Jn 19:34 are not identical, but they share the same vocabulary: ‘spear’, ‘to pierce’ (same form), ‘his side’ (same word order), ‘to come out’ (same form), ‘water and blood’ (reversed word order).
The differences in the first words of the addition in Matthew are explained by the immediate context of Matthew. The non-specific ‘someone else’ (ἄλλος) is in line with the equally non-specific designations in Mt 27:47 ‘some’ (τινές), 27:48 ‘one of them’ (εἷς ἐξ αὐτῶν), and 27:49 ‘the others’ (οἱ λοιποί). The participle λαβών in the phrase λαβὼν λόγχην comes from the earlier Mt 27:48 λαβὼν σπόγγον. However, in Jn 19:34 this piercing happens after Jesus’ death, whilst in Matthew the death occurs only in the next verse.

On transcriptional grounds (influence of parallel account) and internal grounds (unlikely that Matthew would associate the loud cry of Jesus with the piercing) the addition in Matthew 27:49 is clearly secondary, but what about the external testimony? Happily, this is one of the Teststellen in the Matthew volume of Text und Textwert (no. 63, volume 2.2). There are some minor variants (addition of ευθεως before εξηλθεν and the order of ‘water and blood).

With the addition in Mt 27:49:
the majuscules ℵ B C L U Γ,
the minuscules 5 26* 48 67 115 127* 160 364 782 871 1010 1011 1057 1300c 1392 1416 1448 1555 1566 1701* 1780* 2117* 2126 2139 2283 2328T 2437* 2585 2586 2622L 2680 2766* 2787,
and NA28 adds some Vulgate mss, and the middle Egyptian, and there is the CPA and Ethiopic.

Without the addition: everyone else (including 15 witnesses that leave out the whole of the verse).
[Incidentally, Text und Textwert did not pick up the majuscule U-030 in support for the addition. It ought to have listed U-030 under a new variant, 3D, with ευθεως and the order ‘blood and water’.]

On external evidence, the addition has definitely a very good shout. Or, to put it in the short-hand principles behind the THGNT, “In light of the external evidence, do we have good reason not to print the reading of the ‘earliest and best manuscripts’?” And indeed, this is one of those high-profile cases where I think that the transcriptional and internal reasons outweigh the external evidence. We should beware of treating any group of manuscripts as so reliable that we ignore what stares us in the face.

However, is there any way we can bolster the argument for the inclusion of the addition? Obviously, if original, the removal of the extra words may solve a problem in the sequence of events in comparison to the other gospels: Jesus did not die because of the spear thrust and neither should the text give any suggestion as such. Therefore, the shorter text provides a less difficult reading.
And then there is Dan Gurtner, in the recent Holmes Festschrift (who does an excellent job of discussing the versional evidence). He is also bold enough to put the suggestion forward that it is perhaps John who is editing the original text of Matthew and places it at a different, more appropriate location in his narrative. However, ultimately this possibility (I don’t think Dan proposes the originality of the longer text of 27:49) raises so many other problems that the simpler conclusion of influence of parallel accounts is preferable over any complex, redactional theory. We may wish the combined cluster of ℵ-01 B-03 C-04 L-019 Γ-036 to be infallible, but it is not. The ‘best and earliest manuscripts’ do not always present us with the ‘best and earliest readings’.

Incidentally, a comparable variant happens at Matthew 27:35, where we have another intrusion inspired by the gospel of John. It concerns the added fulfillment of Psalm 22:19 as found in John 19:24. In the variant we see a similar adaptation of the Johanine language (ἵνα ἡ γραφὴ πληρωθῇ ἡ λέγουσα) towards Matthean style (ἵνα πληρωθῇ τὸ ῥηθὲν διὰ τοῦ προφήτου), just as happened in the longer text of 27:49. The difference is that the external evidence for the addition in 27:35 is less impressive, but it is a good illustration of the same phenomenon as in Matthew 27:49. As far as I can see almost every transmissional strand suffered these harmonisations.

Gurtner, Daniel M. “Water and Blood and Matthew 27:49: A Johannine Reading in the Matthean Passion Narrative?” In Studies on the Text of the New Testament and Early Christianity: Essays in Honor of Michael W. Holmes On the Occasion of His 65th Birthday, edited by Daniel M. Gurtner, Juan Hernandez and Paul Foster (New Testament Tools, Studies and Documents 50. Leiden; Boston: Brill, 2015), 134-50.