Manuscript evidence of Revelation 16:5

The current ECM website data for Revelation 16:5 provides transcription data for around 100 witnesses to Revelation 16:5.  Some of those are witnesses are "corrector" witnesses.  

A comparison of the transcription reveals both a general harmony with the ECM/NA28 as well as numerous departures from the ECM/NA28.  Most of the departures seem to be minor.  

Example 1: Phrase Insertion

About 10% of the witnesses insert "του επι" to the first phrase of the verse, "και ηκουσα του αγγελου του επι των υδατων λεγοντος" and a big percentage note a punctuation mark: "λεγοντος·".  The punctuation mark in this example is an ἄνω τελεία (ano teleia) also known as άνω στιγμή (ano stigmi) or simply "Greek semicolon."  This punctuation mark has no viability because such marks were a later development used for aiding reading, as evidenced by their absence in the oldest manuscripts. 

The addition of the "του επι" would change the phrase from "and I heard the angel of the waters saying" to "and I heard the angel upon the waters saying".  I am not familiar with anyone who argues that this is the original text.  The majority and earliest manuscripts don't have this reading. The Vulgate does not have this reading.  I have not checked patristic witnesses about this myself.  

Hoskier (link) identifies seven manuscripts for this variant as well as Arethas for a similar variant:

The TR, like the NA28, follows the shorter text.  

Example 2: Verb Change

A single manuscript (2048) apparently has ηκουσε instead of ηκουσα to form the phrase "και ηκουσε του αγγελου των υδατων λεγοντος."  The difference in meaning "he heard" instead of "I heard."  2048 is an 11th century manuscript of Revelation.  Hoskier identifies the variant as corresponding to a single manuscript:

The Vulgate does not have this reading.  I have not checked patristic witnesses about this myself.  It might be very hard to distinguish between uses of "he heard" as the commentator's own introduction (i.e. to refer to John rather than the commentator), although in theory there could be commentary that would make the issue clear.

The TR, like the NA28, follows the overwhelming majority of texts.

Example 3: Article Change

It seems there was an error in the copying of 2286, as a result the firsthand text seems to have written and then erased "τω" rather than "του" before αγγελου.  Similarly, the firsthand text of 01 seems to be "των" but was corrected to "του" .  

Hoskier notes the error and correction of 01 thus:

The article "των" with "αγγελου" would be irregular.  Although it is found in one of the earliest manuscripts, considering that it is corrected there and considering that the text of 2286 may simply witness a similar clerical slip, this variant does not seem to be viable.  It's not directly translatable in English or Latin.  

Once again, the TR, like the NA28, follows the overwhelming majority of texts.

Example 4: Phrase Omission and Participle Change

At least three manuscripts (2026, 2057, 2495) completely omit the phrase, "και ηκουσα του αγγελου των υδατων". Two of those replace "λεγοντος" (a genitive participle) with "λεγων" (a nominative participle).  As to the former issue, Hoskier notes:

As to the latter issue, Hoskier notes:

These third and fourth variants of  "λεγωντος" or "λεγοντως" for "λεγοντος" did not show up in the transcription list I have.  Nevertheless, in my list, manuscript 2845 has "λεγοντα" for "λεγοντος", which Hoskier does not seem to note.  Ultimately, each of the manuscripts that depart from the overwhelming majority seems to be a 15th century manuscript.  The Vulgate uses the accusative for "saying" (which I think should serve the same purpose as the Greek genitive here) and English while it doesn't distinguish participles in the same way, seems functionally the same.  I didn't check the patristic witnesses myself, although it can be seen that the Armenian seemingly has a different take.

Yet again, the TR, like the NA28, follows the overwhelming majority of texts.

Example 5: Word Order

Several manuscripts change the order of the words of this first phrase.  For example, 2329 and 2886 place "λεγοντος" before "των υδατων", while 2847 places "των υδατων" before "του αγγελου."  Manuscript 2847 also replaces "ηκουσα" with "οικουσα".  2847 is a 16th century manuscript and 2886 is a 15th century manuscript, while 2329 is a 10th century manuscript.  

Hoskier does not seem to note the latter issue, but regarding the former issue observes:

Once again, the earliest and vast majority agree, and the TR and NA28 follow the earliest and vast majority of texts.  I don't think the word order issue would result in a translation difference, and it would seem unlikely to be clearly identifiable in the patristic evidence, which I haven't checked.

Summary (so far)

As you have hopefully seen so far, even in just this first phrase of the verse (i.e. Rev. 16:5a), there are only seven Greek words, but there are more than seven variant readings in the five examples I've provided above.  Moreover, Hoskier identifies additional variants such as the omission of the initial "and" (in the Sahiddic and some of the Boharic) and the addition of "holy" (ἁγίου) before angel, as well as the omission of "the angel."   Nevertheless, in each case the vast majority of the Greek copies agree with one another, and in each case the TR and the NA28 agree.  In fact, essentially  

Rev 16:5 (STEP bible)(NA28)

(NA28) Καὶ ἤκουσα τοῦ ἀγγέλου τῶν ὑδάτων λέγοντος ...

(Nestle) και ηκουσα του αγγελου των υδατων λεγοντος ...

(Ant) και ηκουσα του αγγελου των υδατων λεγοντος ...

(Byz) και ηκουσα του αγγελου των υδατων λεγοντος ...

(Elz) και ηκουσα του αγγελου των υδατων λεγοντος ...

(SBLG) καὶ ἤκουσα τοῦ ἀγγέλου τῶν ὑδάτων λέγοντος ...

(SRGNT) Καὶ ἤκουσα τοῦ ἀγγέλου τῶν ὑδάτων λέγοντος ...

(THGNT) καὶ ἤκουσα τοῦ ἀγγέλου τῶν ὑδάτων λέγοντος ...

(Tisch) καὶ ἤκουσα τοῦ ἀγγέλου τῶν ὑδάτων λέγοντος ...

(TNT) καὶ ἤκουσα τοῦ ἀγγέλου τῶν ὑδάτων λέγοντος ...

(TR) και ηκουσα του αγγελου των υδατων λεγοντος ...

(WHNU) και ηκουσα του αγγελου των υδατων λεγοντος ...

Even the Complutensian Polyglot agrees:

In short, despite the presence of lots of variants, the first half of the verse is relatively uncontroversial.

Interesting Variants in the Second Half

In order to reduce the amount of space, let me summarize the variants in the second half (i.e. "... Δίκαιος εἶ ὁ ὢν καὶ ὁ ἦν καὶ ὁ ὅσιος ὅτι ταῦτα ἔκρινας"), limiting myself to the variants with Greek text support in the 100 witnesses for which I have transcripts.  Before I do that, I think it worth noting that literally all 100 witnesses have no variation regarding the word "δικαιος" (righteous).  

1. Manuscript 61 had ο and then δι before δικαιος but ultimately was corrected to avoid such insertions.
2. Manuscript 2344 adds Lord in nomina sacra form ("κ̅ε̅·") after righteous (Manuscripts 2049 and 296, which are from the 16th century and copied from a printed text likewise have the inserted word).
3. Manuscripts 104, 141, 2042, 2495, 2847, and 2919 omit εἶ.
4. Manuscripts 131 and 2042 add "ην" before "ὁ ὢν"
5. Manuscript 2495 adds "ων" before "ὁ ὢν"
6. Manuscript 2847 omits the "ὁ" before "ὢν" 
7. Manuscript 2847 omits the "καὶ " after "ὁ ὢν"
8. About 27 witnesses (including the earliest witness) use the word "ὃς" rather than "ὁ" before "ἦν"
9. About 23 witnesses (including the earliest witness) include the καὶ before ὅσιος .
10. About 49 witnesses (about half and including the earliest manuscript) omit the ὁ before ὅσιος
11. Manuscript 2432 has a reading of ωσιος rather than ὅσιος although it is corrected.
12. Manuscript 2847 has "και ω ων ωσιος" instead of "καὶ ὁ ὅσιος"
13. Manuscript 2919 has "ο ο αμγοσιος" instead of "καὶ ὁ ὅσιος"
14. Manuscript 2495 omits "καὶ ὁ ὅσιος"
15. Manuscript 2026 inserts "εν τοις εργοις σου" (in your works) after "καὶ ὅσιος"
16. Manuscript 469 inserts "και ο αγιος" after "καὶ ὁ ὅσιος"
17. Manuscript 2196 omits "ὅτι ταῦτα ἔκρινας" 
18. Manuscript 2845 has "παντα" for "ταῦτα"
19. Manuscript 792 has "δικαια" for "ταῦτα"

As you can see, all but three of these variants are cases where there is an overwhelming majority reading.  Moreover, as you can see, even with just these copies, there are 13 words and 19 variants.

Looking at the NA28 (which agrees with the Complutensian here), the Stephanus 1550 TR, and Beza's TR, we see the following differences:

• NA28/Complutensian ... δίκαιος εἶ, ὁ ὢν καὶ ὁ ἦν, ὁ ὅσιος, ὅτι ταῦτα ἔκρινας,
• TR1550  ... Δίκαιος Κύριε, εἶ ὁ ὢν καὶ ὁ ἦν καὶ ὁ ὅσιος ὅτι ταῦτα ἔκρινας
• Beza 1582/98: ...  Δίκαιος, Κύριε, εἶ Ὁ ὢν, καὶ Ὁ ἦν, καὶ ὁ ἐσόμενος, ὅτι ταῦτα ἔκρινας.

In short, all three correspond to the overwhelming majority of manuscripts for variants 1, 3-7 and 15-19, i.e. the minority position is rejected in all those cases.  Similarly, none of the three follow the minority variants 11-14.  That leaves us to consider variants 2 and 8-10.

The insertion of "Lord" (variant 2)

This insertion has weak evidence in the manuscripts and versions.  Specifically, Hoskier notes:

In other words, Hoskier identifies the word as being present in two Greek manuscripts, the Aethiopic, and part of the Boharic.  

The oldest and the vast majority of manuscripts support the non-inclusion of the word.  The reason for the presence of the word in the TR editions (both Stephanus and Beza) is the inclusion without comment by Erasmus from his very first edition.

The presence of the extraneous "Lord" in a small number of manuscripts is most easily explained by parallel corruption from verse 7, which has the word, or from an insertion intended to clarify that the angel was talking to the Lord, not to John.  The alternative explanation that scribes tended to insert the word "Lord" into the text, while correctly premised, does not seem to be the best explanation here, as the usual trigger for such an insertion would be the presence of the name, Jesus, or the title, Christ.  The insertion does not alter the meaning of the text, because the angel is speaking to the Lord.

The word "ὃς" rather than "ὁ" before "ἦν" (variant 8)

None of the three editions we are looking at follow this well attested but minority reading.  The probable reason for rejecting this reading is that although it makes the Greek smoother, it does so at the expense of an intentionally irregular use of Greek here, to signal that the Greek participle is being used as a name/title of God.  The reading here is so similar to the readings found at 1:4, 1:8, 4:8, and 11:17 that it is hard to imagine that a different wording was intentionally used by John here.   

The omission of "καὶ" after "ἦν" (variant 9)

The NA28 and the Complutensian Polyglot, following the majority of witnesses against the minority and the earliest witness, omit the "καὶ" here.  It's a difficult variant to consider in isolation.  Possibly the Complutensian editors just followed the manuscripts they had.  For the NA28 editors, the choice had to be informed by the related variant of  ὁ before ὅσιος, which we will consider next.  In short, "καὶ" makes sense if the word "holy" is supposed to be coordinate with "righteous" (Δίκαιος).  Thus, the insertion of this word can be used as an attempt to correct a perceived omission. 

The omission of the "ὁ" before "ὅσιος" (variant 10)

If the word "holy" were coordinate with "righteous," one would not normally expect the article ὁ to be present, because it would not be needed.  Correspondingly, the article is omitted in quite a lot of the exemplars, seemingly as a false correction to the text.  Interestingly, though, I cannot tell at a glance whether the ὁ is more often omitted when the καὶ is present or absent.

While we may be inclined to agree with Beza that the combination of "καὶ ὁ" that he found in the Stephanus TR is problematic, the better solution is to understand that the "καὶ" was erroneously added.  With that understanding, the point of the "ὁ" is to serve a similar purpose as it did with the previous names, "ὁ ὢν" and "ὁ ἦν," namely to signal that the word "holy" is being used in the Vocative as a name/title of God.

The substitution of "καὶ ὁ ἐσόμενος" for "καὶ ὁ ὅσιος"

The reader will notice that this variant does not have a number.  That's because this variant does not correspond to any of the texts in the list I have.  It likewise does not correspond to any text found by Hoskier.  Some people claim that Beza had a manuscript with such a reading.  If so, it would be a singular reading.

If such a manuscript existed, it seems likely that it was similarly an attempt to deal with the difficult construction "καὶ ὁ ὅσιος" found in the exemplar.  The scribe presumably saw that "ὁ ὅσιος" was irregular and thought that the "καὶ" would be more naturally followed by a third tense participle of the verb "to be" (εἰμί), namely the future middle participle, nominative masculine singular (ἐσόμενος).  

There is no versional evidence that directly supports this substitution, and the patristic evidence brought forward to substantiate this substitution is shaky at best (see the analysis here).

Conclusion

For Revelation 16:5, despite the numerous variants witnessed by the manuscripts, it turns out that there are very few variants that require us to do more than realize that they represent a tiny minority.  While variants 8-10 (from the above list) are interesting and require some thought, they are resolvable.