As Dr. White mentioned in his radio program yesterday, the Authorized version (KJV) has "love of God," in 1 John 3:16, where most other versions do not, and where (apparently) few Greek texts have the words. The most popular versions of the KJV place the words "of God" in italics, which is usually used to indicate that the words have been added without explicit support in a sufficient number of Greek manuscripts.
Dr. White raised the interesting question: how did it get in there? I need to investigate the Textus Receptus. My electronic copy has the words, but sets them off (as though not in the "original" textus receptus). Scrivener's transcription of the Textus Receptus identifies the words as not appearing in the Textus Receptus, but as being found in "B" (i.e. Bezae, which appears to refer to a printed edition of the Greek New Testament by Calvin's successor made in 1565).
I have not done an exhaustive search. Nevertheless, I found the following general attestations:
Latin / Greek
Pope Sixtus V version of the Vulgate, as approved by Clement VIII (link) has the "dei."
Complutensian Polyglot has "ton theon" in Greek and "dei" in Latin. (not readily linkable, thus partial text shown below)
Rheims (from the Vulgate) (link) has "of God."
Wycliffe (from Vulgate) (link) has "of God."
Lamsa (from Aramaic) (link) has "his."
Murdock (from Aramaic) (link) has "his."
All other major English versions (aside from the Authorized version) and Latin versions seem to lack the "of God." Other English versions that have it include the "Literal Translation of the Holy Bible," the so-called "Modern King James Version," and Webster's Bible - all of which appear to be sourced at least partly in the KJV.
Other versions that have it are the Biblia Gdanska (if my reading of the Polish is correct) "Przez tośmy poznali miłość Bożą, iż on duszę swoję za nas położył; i myśmy powinni kłaść duszę za braci. " also has it. Although this version appears to have been revised in the 19th century, I would be suprised if it were simply harmonized with the KJV, although it does appear to have been based on "the Textus Receptus." It should be noted that there was a long-lasting Latin influence in Poland (note that they use Roman letters unlike most of their other Slavic brethren), and perhaps we can blame the Vulgate for this appearance in the Polish Bible. It does not appear in the Old Slavonic, as far as I can tell.
The Spanish version known as Las Sagradas Escrituras Version Antiguaalso also has it, in italics, and I think we would discover that the LSEVA is more or less simply a translation into Spanish of the KJV.
It seems highly unlikely that the KJV translation team had the Aramaic version of 1 John available. The KJV translators may have had a copy of the Complutensian Polyglot - I'm not sure. If they did, I'd be ready to blame the insertion on that source, which they would have trusted to at least some degree.
Regardless, it appears possible that ultimately the word was derived from the Latin of the Vulgate, rather than from any Greek manuscript. My suspicion in this regard is enhanced by my belief that the Vulgate played a large role in the acceptance of the Johanine Comma in chapter 5 of 1 John, which was naturally translated by the same translation team as chapter 3.
My best guess then is acceptance of the testimony of the contempory Vulgate (as with Wycliff, Rheims, and possibly the Complutensian Greek, via a reverse translation). There's another possibility, which is conjectural emandation. The word translated "love" has an article. As certain modern Greek scholars are wont to point out, the proper translation of the article is one of the most widespread issues that modern scholarship has with the KJV. It may be that the KJV translators believed that because "love" had a definite article (την αγαπην), the words "of God" were conveyed via the article, through antecedent reference to verse 1.
Another way that the emandation could occur is simply as an aid to the reader. Although the KJV is mostly a literal translation, there are a few instances where what would today be called "dynamic equivalents" are used. Perhaps this is such a case.
I suppose it is even possible that Beza and one or more "Vulgate" translators (though apparently, according to Dr. White's comments on the radio program yesterday, not Jerome himself) made a similar emendation.
Ultimately, the sense is the same (I believe) whether we accept the words "of God" as genunine or not. I suspect that most of the major textual critics agree, which is why this particular variant is not discussed (as far as I can see) in most of the major critical editions of the Greek testament. For example, neither my NA27 nor my UBS4 has any mention of this variant, and Scrivener.
P.S. While we are speculating - there is also another way that "theos" could have crept into the text. The word immediately after "love" is a conjunction that begins with an Omicron. In the uncial manuscripts (especially those on payprus) it would be possible to mistake an Omicron for for a Theta (they differ by a single horizontal line). Furthermore, a horizontal line in the paper (whether due to the lined characteristics of papyri, or lines written to assist in writing straight, or simply a stray mark) over a Theta is an abbreviation that was widely used to denote the word for "God." There were no commas in the original, so it would be remotely possibly that a sleepy scribe could mistake hoti for the abreviation for theos plus the indefinite pronoun ti (τι).
UPDATE: My original post made reference to B as referring to Bezae, which I took to refer to "Codex Bezae," an early manuscript - and one carefully transcribed for publication by Scrivener. On further consideration, though, I think Scrivener more likely meant the 1565 printed Greek Testament published by Theodore Beza.