More than you ever want to know about grids

Last week we had a slight delay on a report when the Times website had no grid for a puzzle.  To get round this if it happens again (or the more likely ‘wrong grid’ problem), I’ve obtained a set of Times grids from Anax and constructed an index which the bloggers can use to identify the right grid (assuming the clues are numbered correctly…).  This means that without having to collect all the grids from observation (and combining information I already knew), I can tell you that:

  • There are 64 grids in current use.  The oldest ones date back to fairly early in the editorship of Edmund Akenhead (Times xwd ed 1965-83), who got rid of most or all of a rather eccentric collection including some absolute shockers, and put together a set of 25 grids.  These were culled and added to by later editors Brian Greer and Mike Laws, so only 13 of Akenhead’s 25 are still in use.
  • All follow these rules: at least half the letters checked in every word, no more than two unchecked letters in succession, double unches never at the beginning or end of words, and no part of the grid can ever be isolated from the rest by filling in a single white square.
  • As you probably know already, all Times crossword grids have a clue numbered 1 Across. In all but two of them, it starts in the top left square. They all have a 1 Across because none have answers in even rows and odd columns, or even rows and even columns.  Of the 64, 49 have answers in odd rows and odd columns, 8 have answers in odd rows and even columns, and 7 have ‘double unch’ patterns where both even and odd rows and/or even and odd columns are used.
  • In the Akenhead set, there were shortages of some answer lengths such as 13 letters.  Among the 64 grids there are now 40 11-letter words, 30 12-letter words, 36 13s, 38 14s, and ‘only’ 35 15s.  I guess the low number of 12s is because a 12-letter answer implies three black squares – the grids generally keep the number of black squares to a minimum.
  • “Big black shapes” are rare – only about 13 of the grids use any areas of 7 or more blocks making a solid shape, and none of these shapes includes a 2×2 square of blocks.
  • Numbers of answers: 5 grids have 32 answers, 2 have 31, 25 have 30, 5 have 29, and 27 have 28.
  • 17 of the 64 grids have quarter-turn rotational symmetry as well as half-turn.
  • If you list the across clue numbers for each grid, each set of clue numbers is different.

26 comments on “More than you ever want to know about grids”

  1. In terms of answer length, the spread is a little over 7 different lengths per grid.
    On the stingy side, three grids offer only four different answer lengths – one has sixteen 7-letter answers, another has sixteen 8-letter answers.
    One grid boasts eleven different answer lengths, from 3 to 13 letters.
    1. I guessed right for one of these – the sixteen 8s one is Akenhead’s one with blocks making a letter E for Edmund – presumably kept for sentimental reasons. I’m pretty sure its also the only one with any fully checked answers – its two four-letter words.
  2. Pete, just noticed in your message…

    “…constructed an index which the bloggers can use…”

    Can’t find a link to this, or is it soon-to-be-posted?

  3. I’m not sure why 12 letters implies three black squares. Can’t you have 12 letters than block-space (part of a word perpendicular to the 12-letter word)-block?
    1. You can’t get away with much round here! You’re right – it doesn’t imply 3 blocks. And grid 20 is an example of the method you suggest.
  4. I find the above posting fascinating as this kind of research is dear to my heart.

    But what I have studied is limited to crosswords published in Indian newspapers (but remember most of these have their origin in the UK).

    Indian newspapers are as careless as their UK counterparts in publishing crosswords and they too bungle in all sorts of ways including mismatched grids. On these days I have had no problem in determining the right grid and summoning it for my use.

    And for my own composing work I use six original grids created in such a way that the maximum word-length in the first grid is 10 letters, second 11 and so on till sixth 15.

    Of course one could construct a grid given clue numbers and word-lengths, I suppose.

    I must admit that my interest in crossword grids stemmed from reading such seminal works like Ximenes on the Art of the Crossword, Anatomy of the Crossword by D. St. P. Barnard and Teach Yourself Books: Crosswords by Alec Robins.

    – Rishi

  5. In my grid too the 12-letter word is all letters and rests along the perimeter.


  6. I always worried that my crossword enthusiasm was bordering on the anorak, but now I know it is. Not only was I fascinated by the stats provided here but I downloaded the grids and had a look at them. Is it too late for me?

    By the way, would the definition of “anorak” as “somebody perhaps a little too interested in a certain subject” be valid in a crossword? – See! there I go again!

  7. I have a question, please. Given a set of clue numbers and solution enumerations, is the grid for the puzzle unique?
    1. I think it is as long as the grid follows the kind of rules that Times grids follow. Richard Browne implied by e-mail that you could determine each of the Times grids given this information.
  8. Many years ago (20?) I refreshed the Indy grids because of the uneven distribution of lengths, there being a particular lack of 11s (there seems to be a natural aversion to both 3-letter answers and a black line 4 squares long). For the last few years we at the Indy have been allowed to produce our own grids, the vast majority have which have been entirely sensible.(Everyman and the Church Times — doubtless among others — have always allowed this licence). Fixed grids go back to the days of metal blocks ,and there is little need for such limitations these days with computer technology — though fixed grids can make editorial control easier. That said, I believe that The Times has an excellent set of grids, though personally I’d like to see a few more that don’t have a totally white-based perimeter. And of course some papers still have some grids which are much less good than others! Don Manley
    1. I certainly agree with the principle of allowing setters to design their own grids. Can’t say for certain, but I think it’s a requirement that Times puzzles are supplied in Crossword Compiler format (I’ll happily stand corrected), in which case the editor could quickly check for e.g. an amount of cross-checking that falls below a minimum requirement – although I’d add that Crosswordman is better for highlighting overunches.
      It could be argued that certain grid designs become part of the furniture, which is fair enough, but with 64 grids at The Times I very much doubt that solvers would bridle at – or perhaps even notice – new ones.
      Funnily enough, from a setter’s point of view I can see a potential downside to freeform grid-making. Those of us with a modest database of good clues awaiting use are frequently surprised at how quickly only 2 or 3, placed in a set grid, start to restrict possibilities in remaining lights. Would it perhaps be inviting “burn-out” – using too many good ideas at once and thus depleting stock – to allow freedom of design?
      Yes, I’d welcome the freedom – but I have this niggling suspicion I’d end up using too many good clues too quickly!
  9. On the contrary — you are in control to use the flexibility and you can fiddle grids to avoid duplication. There is however a problem with grids that are not totally white-framed. One tends to have a predominance of 6s unles one either blocks out a lot more or is prepared to have n unches in a (2n+1)-lettered word. Don
    1. I’d have no objection to freely-invented grids for the Times puzzle, except to ask that it’s not done until getting the xwd onto the club website is just a matter of identifying the files containing the print version of a puzzle and having a computer program convert them to the ones required for the web site version. It’s the lack of this facility that caused the mistake that led to this discussion!

      In the meantime, if one or two Times setters got together and designed a few extra grids to solve particular problems, I suspect the xwd ed would quite happily add them to the set.

  10. You seem to know the editor’s mind better than one of his setter’s, PB! It’s not a discussion that I wish to pursue with him at the moment though
  11. When it comes to fleshing out a crossword skeleton with words, what aids do setters turn to? Especially for the starters, those lengthy slots where one would like to use phrases.

    TEA or similar software?

    But then I am not sure if these apps return “phrases only”. I think the phrases appear amidst long words and so one might have to scroll down a lot before picking up a phrase.

    More than twenty years ago we had The Crossword Phrase Dictionary, compiled and devised by R. J. Edwards. A composer can just riffle through the pages of a configurationwise section and pick up a likely candidate.

    I yearn for a phrase dictionary with updated dB.

    The more recent Oxford Crossword Dictionary and similar “keys” have phrases but again they are buried among words.

    1. Software – Crossword Compiler! I also have a very old trial version of what was Bryson and is now Crosswordman software, and while I can only submit Times puzzles using CC the other one is useful in many ways, e.g. it can sort word lists according to how “difficult” the letters are; especially good for making those coded crosswords.

      CC’s search function draws on a number of word lists – basic English, compounds & phrases, literature, movies and several more – plus, of course, a “default” one which generally is all that’s needed for a standard puzzle.

      The danger of referring to the phrase list for long answers is that many – possibly most – are not dictionary phrases even though they sound very familiar. If I ever resort to this list I keep a dictionary to hand so I can check, because I’ve been caught out in the past. A nice feature, by the way, is that you can filter the lists so they only display answers that will allow intersecting answers in the grid to be filled. It’s not totally reliable as this filter refers to the default word list, so there may still be possibilities by looking elsewhere; but it certainly helps.

      I doubt that setters have an active preference for phrases. What we tend to look for is visible wordplay or, at least, not a selection of appalling letters. In a way, phrases come with an in-built warning sign; an enumeration such as (5-2,8) yields so few possibilities that experienced solvers will probably guess the right answer just by looking at the enumeration and one or two given letters in the grid.

  12. Are Times crossword setters supplied with a copy of all the 64 grids so they may use any of them for their monthly quota? If so, how is wider distribution of grids ensured?

    Or, are certain grids used only by certain compilers?

    1. It’s fair to assume all setters get the full set of 64 grids, as the editor tries to maintain as much variety as possible.
      Setters won’t have grids assigned to them, but they may well have preferences. For me, it’s more a case of there being some grids I don’t like simply because I feel they’re not challenging – remember a few years ago The Guardian finally dumped a real shocker of a grid because it was far too easy for the setter to fill, resulting in ridiculously and unfairly tough crosswords.
      Times setters are expected to leave a reasonable gap between puzzles on the same grid.

Comments are closed.