It’s both a bounty and a hindrance to cryptics that there aren’t hard and fast rules for what to expect. There’s good stuff going on in just about every cryptic, but your own expectations often run afoul of any given constructor’s interpretation of the “rules.” Differing reference points easily expand into an inability to understand what is being said. And while there’s a lot of good solving guides to be found on the net, guides that wade into the finer points of construction are harder to locate. (Stella Zawistowski has begun doing this and that’s a great resource that we look forward to seeing more of.)
With that, welcome to Tap the Sign. This is a place where we will discuss a lot of what we would call the cryptic “don’ts” … but much like the school bus in the Simpsons episode it references, it’s a low-enforcement zone. We’ll point out why we consider some wordplay to be inelegant. For the right clue, any “don’t” can be overlooked.1 However, when something starts to show up a lot in a puzzle, that’s when you have to tap the sign.
Today our focus is on Bits & Pieces clues, where you take some letters from the front or back or middle or what-have-you of fodder words. These indicators are meant to get the clue writer out of a common, and mostly interesting, predicament: Their wordplay is a letter or two off from something awesome. A perfect anagram is just missing an F; these two common words become this larger word if I can just add an R and an S. These are great uses for B&P constructions.
First off, unless a puzzle has a significant number of short (3–6) letter words, you probably want to limit your bits & pieces usage to about 40% of clues. Above that, the solver is likely to clock how samey your wordplay feels.
Within a bits & pieces clue, don’t use different kinds of bits and pieces indicators.Taking the first letter from one word and the last letter from another just feels weird. So does using the outer letters and middle letters together, or really any combination thereof. If you’re in this position, the savvy move is to find to a longer word that closes the wordplay gap rather than taking an axe to a bunch of different words in different ways.
The only partial exception to mixed indicators is letter substitution clues, where you will absolutely have to clue two different letters and perhaps there is a good surface reason to use two different kinds of indicators. But even here, remember that there are unforced abbreviations in common parlance. There are ways to get single letter without chopping up words and phrases.
And of course don’t use any indicator more than once a puzzle. There are robust options for taking firsts and lasts; middles and edges should be used more sparingly, being rarer and tougher for your solver. Let your indicators set the tone for your clues rather than being held hostage by them.
One more thing that’s harder to quantify: Make sure your bits and pieces indicators clearly and cleanly indicate what you want the solver to do. Take “some.” What is ”some nerve?” Is it the N? The NE? The RV? The American standard for such a thing is to resolve ambiguity de minimis and starting from the beginning, meaning that the above is N and N alone. If you want more than that, you have to indicate it; see also “bit of.” (Note that this is different from hiddens where, by virtue of having given the solver the definition for the whole string of requested letters, ambiguity of this sort is allowed.)
This gets even trickier for middle letter indicators. Certainly the “heart” of an even-numbered letter word is the middle two letters. But what if you want the middle four of eight? Is that still the heart? How many times can you apply an outer letters removal indicator to get down to some subset of the letters in the middle? (If “uncovered” removes the first and last letter, “twice uncovered” may be needed to iterate this.) The general answer to these questions is to make your instructions as explicit as possible so the solver won’t feel cheated — if you need ELLE, perhaps that’s “quartet from the center of excellence.” That gets into the matter of different reference points mentioned up top: If a solver comes to think your puzzle can’t be trusted to resolve its ambiguities, they’ll leave yours to find one that’s closer to their expectations.
Always happy to discuss further in the comments.
1 This is distinct from a “can’t,” which are rare indeed — you can’t, for instance, violate the part-of-speech agreement between the definition and answer.