The 1881/94 Scrivener Textus Receptus is not the King James in Greek

While Scrivener was aiming to provide the Greek upon which the KJV was based, he did not always do so accurately.  The most frequently cited example is Ephesians 6:24 where Scrivener omits the terminal Amen, presumably because the 1611 KJV omitted this word, although the Oxford/Blaney 1769 edition and the so-called Pure Cambridge Edition (1909) both include it.  On this point Scrivener's TR agrees with the NA28 and the 1611 against contemporary KJVs.   

There are other places where it is unclear whether the King James translators may have followed a different text.  The reason for this lack of clarity is that the King James is not a woodenly literal translation and because the process of translation is sometimes ambiguous.  

Revelation 15:4 provides two examples: (1) contemporary King James editions have "thou" and "art" in italics and (2) the English word, "holy," could refer to at least two different Greek words.  Scrivener provides no Greek work for "thou art" and provides only one of the at least two different Greek words for "holy."  On this point, Scrivener seems to have assumed that the KJ translators followed Beza's printed text.

The Complutensian polyglot was, however, also available to the KJ translators.  The CP inserts εἶ (thou art) and has a different (from Scrivener's choice) Greek synonym for holy.  

In the second case, it's unclear whether the King James translators would have even cared about this variant.  However, if one assumes that in the case of "thou art," they followed the CP, then it might make sense to assume that they secretly had in mind the CP synonym for holy rather than the Bezaean synonym for holy.

Ultimately, this attempt to perfectly divine the precise Greek behind the King James Version is an impossible task.  Scrivener did a good but imperfect job of it.  On the other hand, the KJ translators were not given the task of producing a Greek text, and - as far as we know - they did not do so.  Because of Archbishop Bancroft's rules for the translators, we cannot know which spelling of place/person names was "right" in the minds of the KJ translators because they were told to stick with traditional spellings.

Because the KJ translators did not woodenly follow Greek word order, we cannot read their minds to see if they thought the original Greek order was one way or another.

Because Greek article usage is significantly different from English article usage, we cannot always be sure about which articles the KJ translators read in the Greek.

Because the KJ translators sometimes translated with a dynamic equivalence instead of a formal equivalence, we cannot always determine which of several variants they were following.

This is not an exhaustive list of issues.

Theodore Beza at Revelation 15:3

 At Revelation 15:3, Beza has a textual note in his 1598 text, which seems wildly inaccurate, at best: 

(Zoomed into the relevant portion)

(source)

Specifically, Beza asserts:

Sanctorum, τῶν ἁγίων. Vulgata seculorum, τῶν αἰώνων, contra Graecorum omnium codicum nostrorum fidem. Arethas autem habet τῶν ἐθνῶν, gentium.

This translates to:

"Of the saints, τῶν ἁγίων (of the saints). Vulgate of the ages, τῶν αἰώνων (of the ages), against the testimony (lit. faith) of all our Greek manuscripts (lit. codices). However, Arethas has τῶν ἐθνῶν (of the nations), of the nations."

In point of fact, there don't seem to any Greek manuscripts that support Beza's reading, the vast majority agree with the Vulgate, and the reading he attributes to Arethas (presumably the Greek commentator) is the minority (possibly the earliest) Greek reading.

I haven't done the work to sort out whether the τῶν αἰώνων (of the ages) or τῶν ἐθνῶν (of the nations) is the original, but I am troubled by Beza's bold and extremely wrong assertion.

Beza seems to be following Erasmus who, from his first edition, had the "saints" reading:

(source)

There is no explanation for this in Erasmus' annotations at Revelation 15:

(source).

Erasmus' source was a copy of Andrew of Caesarea's Commentary on Revelation, but Andrew's text follows the "nations" reading:

(source)

So, it seems that Erasmus may be the one responsible for this error (unless he had a copy of Andrew's commentary that itself contained a textual variant, of course - there are multiple manuscripts of the commentary in existence).  Another option is Erasmus back translating incorrectly from a Latin abbreviated form of seculorum (scm or perhaps sclm), thinking it was meant as an abbreviation of sanctum (scm), or some similar error of sight.  It's very hard to speculate, without knowing which precise manuscript Erasmus used here.  There is some interesting debate on that point, although I believe that there is a current scholarly consensus view on that question.

Incidentally, here are Hoskier's notes on this issue:

(source)