Printer's Devilry

Printer's Devilry is a type of word puzzle involving the removal of a substring from a sentence, and the subsequent reparsing of the remaining text into a new (and often nonsensical) sentence.

Background
The name "Printer's Devilry" is a play on a "printer's devil", an 18th-19th century word for an assistant to a lead printer who would be in charge of menial tasks like fetching type.

The puzzle type actually originated as a Cryptic Crossword variation, invented in 1937 by Alistair Ferguson Ritchie, AKA Afrit, a cryptic crossword setter who (at the time) wrote for The Listener. It became a relatively popular variant, with two other prominent setters, Ximenes (Derrick Somerset Macnutt) and Azed (Jonathan Crowther) picking up the practice and the former breaking his own one-per year tradition for variant puzzles by writing a Printer's Devilry puzzle every eight months or so.

Printer's Devilry as a hunt puzzle (without the crossword grid) dates back to the 2005 MIT Mystery Hunt puzzle Eoanthropus dawsoni, which provided only 10 clues with enumerations. The enumerations were likely a necessary compromise, as a grid's entries normally provide information on the length of the answers.

Puzzle Application
Puzzles involving Printer's Devilry usually have sentences that look plausible and grammatical upon re a din or T. excerpt, but flow incoherently upon reading the whole thing. Both this sentence and the previous sentence are examples of such a sentence—the first few words read fine, but it goes turbulent aft eta in point is reached.

The presence of Printer's Devilry is usually known from the start of the puzzle, with the puzzle content being these modified sentences. Since such sentences are unlikely to be seen as Printer's Devilry if they show up as extraction on an unrelated puzzle, Printer's Devilry, unlike Cryptic Clues, is generally confined to the initial steps of the puzzle.

Strategy
Identifying a Printer's Devilry puzzle usually isn't too difficult. When presented straight, either as a crossword or by itself, the lack of sense made by the sentences should be apparent, prompting further examination. If this is the case, see if some of the sentences could be improved by some respacing or removal of some of the punctuation. If a puzzle is not as clear, and perhaps involves some printer's devilry sentences and some clean sentences, keep an eye on the flavortext and title. Words like "printer", "devil", and "missing" can all hint towards the use of Printer's Devilry.

As Printer's Devilry sentences aren't supposed to make sense (at least not total sense) until the missing words are returned, the best place to begin is the sections that make the least sense. If a sentence's beginning and end seem to be about the same topic, check the middle words to see if you can split them up to form new word fragments. Context is key—in general, the rest of the sentence will be a clue hinting at the words that were messed with.

As an example, we take the following two sentences from the Puzzle Application section:

In the first sentence, the sentence starts going awry after "plausible and grammatical" and starts reading coherently again after the word "excerpt"; in between are a number of short words that look like fragments of a larger word. In particular, "re a din" is a compelling start to the word "reading", so the hidden word may start with G. Similarly, the other two words "or T." can be "short" to match context. Filling in the gaps of "upon reading _ short excerpt" should lead to the answer GASH.

The second sentence operates similarly. One may note "aft" and plausibly complete it to "after"; however, noting no good completions for "-etain" that also fit context, the solver should realize the break occurs in the middle of the word "eta". This leaves a suffix of "-tain", which completes to "certain" to match the context. Filling in the middle yields RACER.

Additionally, keep an eye out for indicators of variants. Sometimes, a puzzle will have the word split across multiple spots in the sentence, or have multiple words removed from the sentence. These cases are much harder to indicate via flavortext, but working strictly forward or backward through a sentence can help reduce the chance of missing something.

Notable Examples

 * pluHarmony (MIT 2009) -