The bad and the ugly

This page describes the things that make some cryptic puzzles bad. Some of the material here is my opinion, but I hope most of it reflects the faults that good cryptic setters try to avoid, even if they don't always succeed - I know how hard it can be from my own attempts at setting puzzles.

Thoughts about Fairness

There is no definite set of rules which establishes whether any given cryptic clue is fair or not. Some element of judgment is always required.

There is a crucial difference between difficulty and fairness in cryptic clues. A clue can be difficult without being unfair, or unfair without being especially difficult for those who know what kinds of unfairness are most common.

The effect of unfairness is often cumulative. A slightly dodgy wordplay doesn't matter nearly as much if the clue has a precise definition and the grid provides a lot of checking. If all three of these factors are problems, solving the puzzle can become a very frustrating experience (or is just abandoned).

The Grid

Although most of the 'fairness debates' about cryptic puzzles are about the clues, the grid is important too. It's also very easy to see whether a puzzle has a decent grid. If it doesn't, then in my experience it's pretty unlikely that the clues are much good, so I rarely bother to look further if the following standards aren't met. This rule works quite well for separating the wheat from the chaff when searching for decent puzzles on the web.

All rules can be bent for special occasions, but for a standard cryptic, there's no excuse for bad grids when there are so many computer programs available to help you find the right words.

Many of the rules about grids come from the limitations of the "hot-metal" printing technology used in newspapers until roughly the 1980s. Allowing setters to design their own grids would have been too expensive, so each newspaper had a set of standard grids. Setters would just choose one and fit words into it. This is still true for many newspaper puzzles today, in spite of the theoretically infinite range of grids offered by new technology. I guess setters don't want to worry about the possible rejection of their puzzle because the crossword editor doesn't like the grid - the standard grids are approved by definition. The Guardian even announced the retirement of one of its least impressive ones once, when it appeared for the last time. As of 2009, the Independent is the only one of the five broadsheet puzzles which is known to allow setters to design their own grids for daily 15x15 puzzles. The Guardian and FT may do so occasionally, but I'm pretty sure the Times and Telegraph never do.

Amount of checking

Part of the challenge of solving a crossword is 'pure' clue solving - working out the answer to a clue without any assistance. Another part is using the answers you have already to solve the harder clues, from checking letters. This is only possible if there are enough checking letters. The best rule for blocked grids is that at least half the letters in each word should be checked, so a 6-letter word might have three checked letters (occasionally 4), and a 5-letter word would usually have three. This is the principle used by the Times. Some other newspaper puzzles have grids where answers with an odd number of letters have less checking - two letters in a 5-letter word, for instance (but not one letter in a 3-letter word). There should never be more than two unchecked letters in succession in any grid entry.


All the words in the grid should be linked together. If you ever see Daily Telegraph puzzle No. 1, you may notice that it's actually nine independent 'mini-crosswords'. Large parts of the grid only linked by one clue are a weakness. An example is a grid which I think has been used at some stage by all the papers. It has quarter-turn symmetry and in each of the 4 corner, 3 8-letter words cross 3 6-letter ones. The corners are joined together by 4 10-letter words which cross over in the middle of the grid. The intersections of the 10-letter words are the only connection between the 4 corners, and its very easy for solvers to find themselves badly stuck in one of these corners.


By tradition, crossword grids have 180-degree (half-turn) rotational symmetry. Quite a few grids in newspapers have 90-degree rotational symmetry as well.

Black patches

Large areas of black squares look unappealing and waste space in the grid. Few newspaper grids have areas of more than 5 black squares joined together.

Word-length distribution

It's easy to find (and design) grids where all answers are 5,7 or 9 letters long and there are no words at all of other lengths. In the absence of thematic reasons for using grids like this, ones with a good range of answer lengths are preferable.

The clues

Good and bad definitions

  Definitions should be precise, and should match the part of speech being defined. Categories like "Fish" or "bird" are rather weak definitions - some extra description to narrow down the range of possibilities gives a clue much more credibility.

"In an orchestra" is not a fair definition for TROMBONE or TROMBONIST. "Someone in an orchestra" would be OK (the conductor is addressing these players when he asks the trombones to tone it down a bit), as would "member of an orchestra". However, even Ximenean high-priest Azed allows an exception, for "verb phrases as clues to nouns that could stand as their subjects". This means that "wags its tail and is man's best friend" is OK for DOG, but not "furry and four-legged".

Commonly associated words can cause trouble - a clue discussed in rec.puzzles.crosswords a while back used 'monk' as a definition of GREGORIAN. It took someone with sharper wits than me to point out that although many of them sing Gregorian chant, there is no Gregorian order, so 'monk' and 'Gregorian' cannot mean the same thing.

'False generalization' is another issue for discussion. Under Ximenean rules, you can use 'dog' to define ALSATIAN, but not the other way round, unless you add something like 'perhaps' or 'for instance'. Since he took over in 2002, Times Crossword Editor Richard Browne has relaxed this rule in at least some cases, though not all his setters take advantage of the opportunity. On the purely practical basis that the number of 3-letter possibilities for "What is an Alsatian?" is many times smaller than the number of 8-letter possibilities for "Name a type of dog", and the same applies to other things like US states and types of fish, I find it hard to believe that solvers are really inconvenienced by the change, once they know about it.

Substitution test: as a simple rule of thumb, if you cannot invent a sentence where the answer to the clue can be replaced by the definition (without changing the meaning of the sentence or producing nonsense), then the definition is probably a dud. (Exception for Azed's verb phrases.)

Good and bad wordplays

Afrit's injunction

This is a quote of a quote - I'm copying from Don Manley's book. The original is in Afrit's Armchair Crosswords, which has recently been re-published. In the Introduction, he mentions the (unwritten) Book of the Crossword. His injunction says:
We must expect the composer to play tricks, but we shall insist that he play fair. The Book of the Crossword lays this injunction upon him: 'You need not mean what you say, but you must say what you mean.' This is a superior way of saying that he can't have it both ways. He may attempt to mislead by employing a form of words which can be taken in more than one way, and it is your fault if you take it the wrong way, but it is his fault if you can't logically take it the right way. The solver, for his part, is enjoined to read the clues in an anti-Pickwickian sense. This also requires explanation. To take a remark in a Pickwickian sense is not to take it too literally; therefore to read a clue in an anti-Pickwickian sense is to close the mind to the acquired metaphorical meaning of the words and to concentrate upon their bald literal significance. if you do so, you may find you are being presented with an anagram of the solution, or that the solution is hidden in the clue, or a bit of jugglery with its components is being done.
I can't improve on this as a statement about what good cryptic clues are all about. It's here in the section on wordplays because the wordplay is usually the part where you have to concentrate hardest on the "bald literal significance". When that significance is at variance with the solution, the clue is unfair.

Instructions must be in the right order

I've seen far too many "container and contents" wordplays which read "A about B" when they mean that A is inside B. I've also seen "A wearing B" when B is inside A. This is normally done when this word-order makes a more convincing surface. Unless there are suitable meanings of about and wearing that I don't know about, such clues are unfair.


A hangover from the old idea that any change in punctuation is allowed in a cryptic clue. Nowadays, such practices are best indicated by the phrase "punctuation may mislead" in a rubric. By "word-twisting", I mean clues like "Beaten Act? Indeed!" for DEFEATED, where the word "indeed" has to be translated to "in deed". As "indeed" is one of the most clichéd of these, it causes about as much trouble as an unindicated anagram. The problem is: where does it all end? Try "Drink in Kirby's church" = KIRSCH = 'Kir' by s, ch = Church. Is this clever or going too far? I think it's the latter, and I just invented it...

Indicators must be present for anagrams, hidden words, etc.

Unindicated anagrams are common in some puzzles, and regular solvers are so inclined to try anagrams of suitable-looking words or word groups that unindicated anagrams rarely prevent a puzzle from being solved once the solver gets over the initial shock. Like the routine stealing of five yards at soccer throw-ins, they are commonplace but still cheating.

Surface padding

Another quote - from Azed this time
Every clue shoud contain a definition or equivalent of the answer plus a cryptic treatment of its component parts, and nothing else. Every word in the clue must have a function as part of the whole, and there should be no superfluous verbiage.
Even Azed himself breaks this rule slightly I think, by using words that link the def and wordplay. However, his link-words always "have a function" because the cryptic reading with the link-word still makes sense. To help produce a more convincing surface meaning, some setters add "superfluous verbiage" words into clues, especially nonsensical link-words such as "with" - my own particular pet-hate. Any clue which does this is strictly speaking unfair, though regular solvers will gradually learn to ignore the verbiage.


Some wordplays rely on bits of general knowledge - e.g. that there was an Archbishop of Canterbury called Temple. These should always be checked for accuracy. No-one is immune from mistakes - a Times puzzle in the same week as I'm writing this included a clue implying that jumbo jets could break the sound barrier, and the same paper once thought that Otago was a New Zealand town rather than an administrative area.


Wordplays should not rely on obscure facts or word meanings.  That's easy to say! - what counts as too obscure will always be arguable, so all a setter or editor can do is use common sense and consideration of the intended audience. My best advice to any setter on this point is to make sure that over the grid and set of clues in the puzzle, a range of areas of knowledge is used. Three or four clues all based on books and all in the same corner will get some solvers justifiably annoyed. (I suspect good crossword setters and editors watch out for exactly this kind of problem.)


Clues that rely (e.g.) on "fah" and "far" sounding the same may irritate solvers for whom they do not sound the same - Scots for instance in this case. Personally I'm happy with homophones like this as they work in the pronunciation recorded in the usual reference dictionaries, but that does just happen to match my own speech quite closely.

The answers

Yes, these can be unfair! For daily paper cryptics, there should be a reference dictionary in which all the answers which aren't proper nouns can be found. This shouldn't be a multi-volume one, or one requiring both hands to lift (and I don't mean the CD-rom version). Most of the words in a daily paper cryptic should be in a dictionary no larger then the Concise Oxford. A few words from larger dictionaries like Chambers are OK, but should be clued with wordplay that makes the answer easy to be sure of - so a cryptic definition is not a good idea.

Many British newspaper puzzles seem to work on the principle that people solving the Saturday puzzle are more likely to have a dictionary available - these puzzles may have more 'dictionary words' than weekday ones.

Proper nouns and phrases are less easy to judge for fairness. There are reference books for many kinds of proper noun - see the booklist for some examples - but judgment should be used as well. Phrases should be well known as phrases - one which are just plausible-looking sequences of words are not on. For example, DEEP PURPLE, PURPLE PATCH and PURPLE HEART are OK, but PURPLE SHIRT is not.

Next Puzzle sources
Previous Solving tips
Contents YAGCC Contents Page
Puzzles Other material about cryptic puzzles on this site

Editor: Peter Biddlecombe
E-mail  Homepage  Last change: