And it is a very interesting context, 46AND2. I received a private message from a forum participant about this issue just last week. Here's a copy of my reply:
Yes, the copyists were clearly confused by the ambiguities of gender involving Nymphas and so various Greek mss. reflect that ambiguity on how they handle the pronoun which follows. (Of course, we can have exactly the same kind of ambiguity in English sentences containing ambiguous names like Terry, Kerry, Francis, and Kim. If you don't happen to know Terry et al personally, you may be prone to speak very carefully so as to not imply the wrong gender. Indeed, because of the gender neutral movement underway in modern English, we are seeing "their" standing in place of "his" and "her" where neutrality is the goal. This has posed additional complications for Bible translators.)
I recall one Greek mss. which uses the wording "Nymphas and the church at THEIR house." Obviously, not all Greek mss. of the New Testament have the exact same wording. And sometimes translating from one language to another can almost force one to make various grammatical decisions (such as those surrounding gender) which were NOT necessary in the source language.
By the way, one such ambiguity of a LEXICAL type which I had to address yesterday involved the Hebrew word in the Noah account that has traditionally been translated in many of the older Bibles as "mountains". But one problem with that is that the SAME Hebrew word can refer to BOTH hills and mountains. So in that case, the single Hebrew word can be "mapped" into TWO different words in English which convey a relative size comparison which the Hebrew original doesn't address. That is, the source word simply means an elevated topographical feature, both those which are relatively small (which in English we tend to call "hills") and relatively larger (which in English we tend to call "mountains".) So what does the translator do when the English vocabulary choices encourage a choice---between "hill" and "mountain" in relation to the height of Noah's flood waters---that is not sufficiently specified in the original Hebrew text? One looks for as many contextual clues as possible as well as help from secondary sources like rabbinical commentators and ancient translations (such as the Greek Septuagint translation, the LXX, when dealing with the Hebrew Masoretic Text of the Old Testament.) Indeed, the techniques one uses are as complex and varied as the particular situation and many weighty tomes have addressed these topics.
By the way, I recall one project where I rendered Colossians 4:15 via "and Nymphas and the church that meets at Nymphas' house." This was among the most "neutral" ways to resolve the issue because one doesn't want to ADD concepts which do NOT appear in the original language text even while avoiding excessive awkwardness in the wording. Indeed, while reading a King James Bible you may have noticed various words italicized. That font style tends to indicate an additional word added by the translator to address issues of this sort, so that a knowledgeable reader does not naively assume that the Hebrew or Greek source word that typically produced that rendering was present to produce a 1-to-1 translation mapping. Of course, in my Nymphas passage rendering above, I did NOT add any non-mapped word. I simply took the Greek pronoun which REFERS to "Nymphas" and replaced it with Nymphas itself. (That is, rather than use a he/she/it pronoun referring to the antecedent, I supplied the antecedent directly.) It is probably the least controversial solution because there's nothing for anyone to fight over. Everybody agrees that Nymphas' house is where that church group met.
By the way, the italicize font style that many King James Bible use to indicate "inserted" words that don't map 1-to-1 have disappeared from many translations. Why? Because it is impossible to use them consistently and meaningfully (see KJV!) because language translation is NOT a mathematical one-to-one equivalency process--- and it creates a lot of misunderstandings when one reinforces a popular myth like that. Mappings from the source semantic domain to the target semantic domain (e.g. Hebrew Old Testament to English Bible translation) always involves a complex mixture of 1:1, 1:N, and N:1! And the vast majority of Bible readers have a very poor understanding of what "literal" even means. Yes, we see that nonsense on these forums regularly when people try to argue for "literal interpretations" and "the obvious, plain, and natural reading of the text." Whenever you see someone use those kinds of terms emphatically, you can figure they are scripture and translation novices.