Amy Ostrander is an undergraduate student from Brandeis University who has taken the very sensible step of broadening her horizons beyond that excellent institution and is visiting Edinburgh for a year to take linguistics courses here. She just pointed out to me something very cute about prescriptivism, Latinophilia, and the so-called "split infinitive".
The familiar practice of putting modifier constituents between to and a plain form verb in the infinitive clause construction (as in to really love someone) is calling "splitting the infinitive" because it is thought that to love is a word. People apparently see the modifier as separating the two parts of what would be in Latin a single word: amare ("to love") is one word in Latin, and certainly no adverb is permitted to occur inside it. That suggests a principle saying, when something is expressed by a single word in Latin but by two in English, it is bad grammar to separate those two English words with a modifier. What Amy pointed out was this. Suppose we accepted the principle (absurd and perhaps unchampioned though it is). We would face a problem. It cannot be maintained, even by prescriptivists whose pronouncements imply that they might defend it. The reason is that (among thousands of other examples) a word like amo in Latin also translates as two words in English: I love. Everyone, prescriptivists included, agrees that it is grammatical to split them, as in I really love you. So under that principle, what makes it OK to split the two parts of amo in an English sentence (I really love) but not OK to split the two parts of amare (to really love)? Nothing. The putative prescriptivists are being inconsistent. So never mind the fact that the principle is absurd; things are worse than that, because no one can really believe it or obey it. Thanks to Amy, I now see that it is an even more utterly stupid idea than I thought it was before.
Anatoly Vorobey points out to me by email, quite correctly, that the the argument from Latin is a bit of a myth: as the generally excellent Wikipedia article points out (particularly here), there aren't really any known prescriptivist grammarians who seriously advocated the because-Latin-therefore-English argument. However, there are some who seemed to think that to + love has a kind of logical unity that the one-word Latin translation confirms. And Amy's observation tells just as well against that weaker viewpoint.
Of course, although it was Amy's remark that first drew this to my attention, hardly anything is truly new: you can see essentially the same point being made in a couple of comments that Nathan Myers has made in the past here on Language Log, one here and the other one here.