A central principle of cryptic crosswords, at least in the United States, is that the two parts of a clue should be truly unrelated to one another. A clue is supposed to provide two independent paths to a solution, and the solver isn’t well served when those paths overlap in their etymology or word usage. Nor is this just a practical matter of solving—there’s an aesthetic angle involved as well. A charade clue that splits CAR PARK, say, into CAR and PARK doesn’t provide much pleasure for a wordplay aficionado, by contrast with CARP and ARK.

But like all conventions, this one can be challenged, and there are some interestingly ambiguous borderline cases. Most obviously, clues that involve puns or whimsy—of the sort that our predecessor, Frank Lewis, used to write frequently and that we write occasionally—often depend on using the same or related meanings of at least part of their answer. For example, a favorite Frank Lewis clue referenced the Detroit Lions in the course of clueing PROFESSIONAL PRIDE; that one used “professional” the same way in both definition and wordplay, but the pun on “pride” was enough to make it a keeper. Our notes to one another on clues in progress will sometimes say “it’s not very double” (in either an accusatory or apologetic tone), and then there’s discussion about whether the breach of convention is worthwhile.

An even more interesting, and treacherous, issue arises in reverse: when the two parts of a clue are actually independent of one another, but don’t look at a glance as though they are. We’ve come to refer to this as “the Ottoman problem.” In Puzzle #3246, we clued OTTOMAN as follows:    Holy Roman Emperor: man associated with an empire (7)

If you check the dictionary for the etymology of “Ottoman,” you’ll find that it’s unconnected to either “Otto” or “man.” The empire is named for Osman I, and the footstool in turn for the empire. So the clue is perfectly fine—or so we thought before we heard objections from solvers. That’s when we learned that “Look in the dictionary and you’ll see that we were right!” can be a pretty weak defense. Not all solvers necessarily want to check the dictionary to see whether a clue violates the etymology taboo, and there’s really no reason why they should have to.

Since then, we’ve learned to avoid writing even legitimate clues that run the risk of looking bogus to the casual solver. (The question arises most often in connection with charades and double-definition clues, which depend on keeping etymological strands distinct.) And in doing so, we’ve made some interesting discoveries. For example, when we needed to clue CUBICLE for Puzzle #3282, we found that the word does not seem, etymologically, to be a “little cube”—but the false connection seemed so plausible that we still didn’t want to use it. The clue we wound up with was:    Treacherous clue about writing implement in a workspace (7)

Like so many things, this is a judgment call, but it’s not hard to make in practice. The upshot is more or less the cruciverbal correlative of Stephen Colbert’s “truthiness” metric. It’s not enough for a clue to be right—it also has to feel right.

This week’s cluing challenge: OTTOMAN. To comment (and see other readers’ comments), please click on this post’s title and scroll to the bottom of the resulting screen. And now, four links:

• The current puzzle

• Our puzzle-solving guidelines | PDF • Our e-books (solve past puzzles on your iOS device—many hints provided by the software!)

• A Nation puzzle solver’s blog where every one of our clues is explained in detail. This is also where you can post quibbles, questions, kudos or complaints about the current puzzle, as well as ask for hints.