— by Hot and Trazom in the June 2008 Enigma. (Somewhat updated in the subsequent months.)
During the years we have been editing cryptic crosswords for The Enigma, we have developed certain editorial standards — many of which are shared more-or-less universally within the United States cryptic community — as well as certain aesthetic preferences, which may be slightly more individual. It took us until now, however, to actually write them up. We offer them here as a guide for submitting cryptics for publication in The Enigma.
Before getting into the specific dos and don'ts, we would like to correct the misconception that all Enigma cryptics must be difficult. While we are happy to publish the toughest cryptics in the US, we know that many members appreciate more straightforward puzzles. Ideally, we would like to serve the whole range of solvers.
The two most common types of cryptics are a 12×12 or 13×13 bar diagram with a theme, and a 15×15 black-square diagram without. Other sizes are possible, depending on your goal; black-square diagrams can sometimes be themed, for instance, and beginning constructors may want to start with a smaller grid. In fact, small (8×8 or 8×9) themed bar diagrams have turned out to be popular. Nevertheless, the standard types represent the traditional expectations of solvers, and anything bigger is difficult to fit in The Enigma.
Words should ideally be at least four letters long, unless the puzzle is smaller than usual or has three-letter thematic entries. Never use two-letter words. (Note that this applies only to grid entries, not clue answers. A two-letter or even one-letter clue answer that will be transformed by a thematic gimmick into a full-length grid entry is perfectly fine.)
An unchecked letter, or “unch” in cryptic slang, is a letter that appears in only one word, not two. Unches should never be consecutive within a word.
In a bar diagram, approximately one-third of the letters in any word should be unches, never more; in a black-square diagram, approximately one-half, never more.
Ideally, every word should include at least one unch. The reason is that, unlike in a standard crossword, you want the solver to have to solve every clue; if a word appears automatically in the grid by virtue of the solver getting all the crossing words, then an opportunity is lost for them to marvel at your wit and ingenuity.
Diagrams should be symmetric whenever possible. The definition of “possible” in this context is dependent on the constraints of the theme and the technical proficiency of the constructor.
In general, entries should have “dictionary nature,” meaning that they could theoretically appear in a sufficiently large dictionary, or a reasonably justifiable word list. “Third cousin” and “orange juice” for example, have dictionary nature; “my cousin”, and “orange stew” do not.
Every clue should have a plausible surface reading. The best clues have a surface reading so strong that it is hard to see the underlying structure of the clue. At the same time, the cryptic reading of the clue must be accurate. In particular, the word or words connecting definition and wordplay, if any, should be readily understandable as indicating equivalence, or should point in the direction from wordplay to definition (not vice versa).
Both surface and cryptic readings of a clue must be grammatically correct. There is a tug of war between surface and cryptic readings. Ideally, both sides should win, but in a square-dealing clue correctness trumps appearance.
Every definition, whether of a clue answer or of any word used in the wordplay part of the clue, must be correct in both its meaning and its part of speech. 11C is the final arbiter. Words that are in NI2 or NI3 but not 11C can be clue answers, but try to avoid using them in the wordplay; also, don’t use an NI-only sense of a word that appears in 11C.
Avoid definitions by example. “Puzzle” can be a definition for “rebus”, for example, but do not use “rebus” to define “puzzle”. Use “rebus, for one” or something similar.
In general, wordplay should offer the solver a route to the answer that is independent of the definition. So, for example, double-definition clues should involve two etymologically distinct words that happen to be spelled the same (and are listed separately in 11C), not two different senses of the same word. Similarly, charade clues should not simply break compounds down into their component parts (OUT + REACHED is a poor charade, OUTRE + ACHED a good one).
Indirect anagrams (anagrams of synonyms) are not allowed. However, some constructors include abbreviations and other non-ambiguous ingredients in anagram fodder.
Any 11C abbreviation is acceptable as part of the wordplay. It is not necessary to indicate that it is an abbreviation.
Phonetic clues should be completely phonetic; partly phonetic clues are inelegant. Wordplay indicators must work grammatically. For example, “last rites” is not an acceptable indicator for S, because a correct construction would require “last of rites”. In most cases, a compound word may be used heteronymically as part of a wordplay indicator, such as “redhead” for R.
Try to avoid clues for which two answers are possible because of symmetry. “Vampire returned key,” for instance, could be either BAT or TAB.
Try to avoid repetition within a puzzle: a solution word used elsewhere as part of a clue, for example, or the same indicator used twice.
We allow misleading punctuation and capitalization within a clue, though we have aesthetic misgivings about the latter.
We do not end clues with a period.
We do not require an exclamation point or question mark at the end of an ”& lit” clue.
We use a question mark for punny definitions.
For “first base” we use (9, 2 wds.) in bar diagrams, and (5, 4) in black-square diagrams. Likewise, far-fetched is (10, hyph.) in a bar diagram, (3-7) in a black square diagram.
Words that are 11C or non-MW are untagged. Words that are MW but not 11C are so tagged (NI2 or NI3). The tag follows the clue, in [square brackets].
Obscure words are acceptable, and need not be tagged. On the other hand, if there is to be obscurity in the wordplay, it should be 11C obscurity.
For archaic or regional words, including an indication such as “formerly” or “in Edinburgh” is a kindness to solvers, but is not required.
Please use American terminology: “wordplay” rather than “subsidiary indication,” and “diagram entry” rather than “light”.
There is no reason to apologize for or identify obscure words, nor to offer a count of capitalized words.
Exceptions to many of these rules are negotiable, particularly for a first-time contributor, or in the case of an aesthetic disagreement rather than something that is out-and-out wrong. However, the final product must be fairly within the limits of solvers’ expectations. In the case of disagreement, the decision of the editor of The Enigma is final.
A few of our constructors work within the parameters of British square-dealing setters, which differ in some respects from U.S. traditions. That is acceptable, but the puzzle will of course still be edited.
If possible, please send us the puzzle via e-mail (email@example.com). You may include the text in the e-mail itself, and/or attach a .txt, .doc, .rtf or .pdf file, and/or send the diagram in a standard graphics format, and/or a spreadsheet.
Members are permitted to fax submissions. If you need to do that, contact the editors by e-mail first.
Solutions should be in a separate e-mail message, or a different page.
We will work with you to reach a publishable puzzle. In rare cases, the test solvers' feedback leads to last-minute changes that we normally don't have time to run by the constructor. For that, we apologize in advance.