But apart from the imaginative force of novelty, why a lake of ice (Cocytus) instead of the more familiar lake of fire (Phlegethon)? Wondering where Dante got this idea, I made a very cursory search of the scholarship and didn’t find nearly as much as I expected to on the subject. Perhaps I don’t know where to look. One model (as seem obvious enough) could be Virgil, who hints at both burning and frigid regions of the underworld (cf. frigidus annus, i.e., winter, in Æneid, Book VI, at l. 311). Milton, by the way, picks up the same thread for his geography of Hell in Paradise Lost (cf. “a frozen continent” where “cold performs the effect of fire”, Book II, ll. 587, 595).
I’ve read that some Islamic stories depict a frozen underworld, as do some other Asian mythologies (e.g., Chinese, Tibetan, Hindu). And certainly this is true of the legends of the frozen north, as for example, the Old Norse Niflheim. But I can’t see any of these being a model for Dante. So, apart from Virgil, where else might Dante have gotten this startling image of sinners up to their necks in ice, Satan beating a cold wind from his enormous wings, and the rest? Where is the ice in Christianity? Given the overtly Christian skeleton of Dante’s poem, this seems a pretty orthodox backbone to me.
I was thinking about this recently, as I have from time to time, and a new idea struck me. Could it be a case of simple word-play, from the similarity of the Italian words, inverno “winter” and inferno “hell” ? Poetic conceits are often born in just this way; at least, they were for me when I wrote more poetry, once upon a long time ago. In fact, it was while reading another poet  that this idea sprang into my head. In Dante’s case, the two words differ by only a single letter — in fact, in both cases the letter is a labiodental fricative consonant, one voiced, the other voiceless. Words don’t get much closer than that. It’s even possible that Renaissance folk-etymologists conceived of a connection between the words (though this is nothing more than a random thought on my part). Does anyone know whether this theory (inferno / inverno) has ever been suggested in Dante circles? (No pun intended! :)
And could it be that simple? It’s certainly the kind of idea that would occur to (and appeal to) the philologically minded. After all, Dante did write one of the earliest serious works on the relationship between languages, De vulgari eloquentia (“Eloquence in the Vernacular”), in which he makes the case for the possibility of eloquent expression in the vernacular [i.e., Italian], as opposed to Latin. This incomplete work is one of the earliest “modern” explorations of grammar, philology, and historical linguistics (albeit at a primitive stage).
Dante clearly seems to have had the requisite appreciation for and facility with words that could have led him to build an extended metaphor on the association of inverno with inferno, whether through (a) serious-minded presumptions of common etymology, (b) playful and poetic legerdemain, or (c) subconscious imagery implanted by near-homophony. It might even be read as a “low philological jest” worthy of Tolkien . Whether right or wrong, whether the whole story or only a small part of it, I think it’s a striking idea. What do you think?
 Dante uses both words in the poem: inferno, in addition to the title, occurs throughout; inverno pops up twice, in cantos XXI (“bolle l’inverno la tenace pece”) and XXXII (“d’inverno la Danoia in Osteric”). The first is especially interesting, juxtaposing as it does the senses of winter and boiling (implying extreme heat rather than cold). For other occurrences of both words over the course of the entire Divine Comedy, see Edward Allen Fay’s Concordance of the Divina Commedia, New York: Haskell House, 1969 (a new edition of the original, published by The Dante Society in 1888).
 I was reading Neruda’s Los Versos del Capitán in a bilingual edition when, for whatever reason, invierno fairly jumped off the page at me. Perhaps it was Neruda’s alchemical word-magic, particularly of nature and the passage of the seasons, but the similarity of Spanish infierno, invierno struck me, and I thought immediately of the same implied pairing in Dante’s Italian. In French, they aren’t quite so similar, retaining more of the phonetic flavor of the original Latin (hiver, enfer, from hībernus “wintry”, infĕrus “below, underneath”).
 My good friend Merlin DeTardo sums up the connections (what few there are) between Tolkien and Dante in an entry for the J.R.R. Tolkien Encyclopedia: Scholarship and Critical Assessment, Ed. Michael D.C. Drout, New York: Routledge, 2006, pp. 116–7.