Listener 4269, Journey to the Centre: A Setter’s Blog by Ilver
Posted by Listen With Others on 23 December 2013
I was intrigued by whether I could set a crossword where every cell in the grid was relevant for the thematic element of the solution. Coupled with this, I have always been impressed by crosswords with a high element of shading, such as the brilliant Keep by Phi, and wanted to try to emulate a similar level.
With this in mind I started to review various grid-based puzzles to see whether any might give ideas as to how to construct such a grid. Clearly the constraints were going to be a challenge. The one that got me on the right track was Minesweeper – the simple game where cells have numbers associated with them indicating the number of adjacent cells which contain mines. This gave the germ of the idea that the letter in each cell could represent a number indicating how many adjacent cells needed shading.
I wanted a grid with a decent amount of shading – I was also a little bothered that if there was too little I may not end up with a unique solution. I have always been a fan of Doctor Who and knew that the 50th anniversary was approaching. The nine-letter title seemed like a good opportunity to fill the grid. I put them into a 13×13 grid and thought they just about looked OK, despite the flaw in the W and a lower case R. On the whole it was not bad. I filled in the numbers in the grid (at this stage using 0-8) and then checked that this gave a unique solution, which thankfully it did (at least after a few attempts).
I am enough of a Doctor Who fan to know that TARDIS is short for Time and Relative Dimension in Space. I opened Chambers not really expecting it to be there, but there it was – although it did have Dimensions rather than Dimension. I checked the BBC site which had the singular, so I thought that I should be able to get away with the cryptic representation T REL DIM in EM for space. I was keen to include this as then the L could also represent the fiftieth anniversary in a blocked off central square.
I now had a grid with shading, the numbers associated with each cell and 9 letters for the final step, and so 9 number/letter relationships defined. Clearly there was no way I could grid-fill in any conventional manner and would need to give myself as much flexibility as possible. I vaguely recalled an earlier Listener I had done with a similar device and thought this method of entry might help. I have since checked and indeed it was a Listener, so thank you Law for A Three Pipe Problem – another brilliant and very challenging puzzle which I would recommend anyone to try.
Up until now I had been working on the basis that each letter of the alphabet would represent one of the numbers 0-8, based on the Minesweeper example. As I tried to work on this basis I quickly realized this would be impossible – I needed more flexibility. I switched to 0-4 as being the least numbers that I could get away with. I rechecked that this still gave a unique solution which it did.
I then started to be a bit more methodical about the grid fill and counted the frequency of numbers in the grid and mapped this against the generic frequency of letters in words. I came up with a preliminary allocation of certain letters to each number, making sure that the more frequent ones had high frequency letters including vowels, S, T etc, with the lesser used ones picking up some of the more infrequently used letters. I avoided letters which were too obscure (Q etc.). I also kept quite a lot of letters up my sleeve to allocate as the grid fill progressed.
I thought that I would have four journeys, each starting at a corner of the grid and finishing by the middle square. I resolved that the paths should not cross themselves but obviously should cross others as this would give the unching requirement. These journeys were supposed to represent trips in the TARDIS, although the Doctor would certainly have something to say about this much crossing of his own time line! Starting at each of the four corners seemed an aesthetic thing to do and I also guessed that the corners were going to cause the most problems so starting here was also logical.
On to the grid fill. I started in the top left with a word fitting the numbers and thought I may as well head straight towards the top right to deal with one of the corners and see how it worked out. Actually it was not so bad fitting these two paths together and the top part of the grid started to come together.
All the time, I had to look at the unching requirements and so started to weave in a third path to make sure of this and then marched that path off towards another corner to address another one of those tricky areas.
I am not sure there is a lot more to say about the grid fill other than thank goodness I was doing it on an ipad so could easily copy the grid, rub out, start a part again etc. I was also very careful with each new entry to try to avoid difficult letters and also to avoid adjacent letters which might create issues for a subsequent path. It was really just a long slog, but always felt doable and, sure enough, after several weeks of painstaking progress it was finally complete. Looking back at my notes, this part of the setting process took about a month. I am guessing maybe 1-2 hours a day on average, so ballpark 30-40 hours to complete the grid fill.
I have to admit cluing is not my favourite part of a puzzle; grid design is what I enjoy (this is why working as Rasputin with two other setters is a joy as I can slack off a bit at this point). I had in mind by now that the letter/number indication would have to be through the clues, and appearances of letters seemed the obvious way to go – thank goodness I had switched from eight to four in terms of numbers. This meant that 26 clues would be needed for the alphabet and the rest for a message giving a steer towards this. I counted up the clues and went for COUNT A LETTER IN EACH OF THE OTHER CLUES. I wrote the clues with the extra letter device. Also at this stage I wanted the shading instruction to be something to be deduced too, so included extra words in certain clues giving the instruction, but this was later removed following editorial feedback.
Off the puzzle went for test solves. Many thanks as ever to the test solvers who have the worst task of trying to solve a tough puzzle, which also has mistakes in it! But with some polishing it was ready for submission.
The feedback (after a long wait) was that they did not really like the extra word instruction, but critically felt that the puzzle with 59 clues and a long preamble would not fit in the space in the newspaper and also I would need one extra clue (LOB) to avoid an unchecked sequence of three letters in HOMEFELT. Looking at the space needed, I would roughly need to halve the length of the preamble and fit every single clue onto one line. I also needed to change the message to COUNT A LETTER IN EACH OF REMAINING CLUES. Needless to say this was possible with some work, although the preamble did perhaps become a little opaque on the shading instruction and the clues became a bit uninventive in terms of complexity.
The puzzle went to the second vetter, who approved it but could not see the thematic relevance of all the crossing paths. Quite right — in my zeal to reduce the preamble, all reference to a journey and tangential links to the theme had gone. The puzzle was also still called HOW? which was bound to be a complete give away given the buzz around the anniversary.
I went through all the titles for Doctor Who episodes to get some inspiration and there was Journey to the Centre of the Tardis. How lucky was that? Change the title to Journey to the Centre, squeeze one reference to journey into the preamble and it just about all made sense again. The editors thought about publishing it one week early or late as a time travel joke and to preserve the numerical slot but in the end decided to delay the numerical.
Thanks again to my test solvers for their excellent input and comments and to the editors for their patience.