## ‘Numerical Playfair’ by Zag

Posted by Encota on 9 Jun 2017

It’s that week in the quarter when the Listener Numerical appears. I suspect that the combination of the words ‘numerical’ and ‘Playfair’ may well have deterred *some* regular solvers, as they both do seem to be in the Marmite category for some Listener regulars. However, this was a great puzzle and I hope most people gave it a try…

As a precursor I skimmed the grid and marked every cell that *couldn’t* be a zero – it *might* be important later, I thought. All 2-digit entries, every starting cell and those implied by reversals left only a very few cells that could be 0 (and even most of those soon disappeared once I got started).

Then this was actually quite a gentle puzzle, once a few items were spotted.

- The top right square of the Playfair (PF) can by definition only be 1, 2 or 3. 7d is a square and so ends in the usual 0,1,4,9,6,5 choices. And 9a once converted via the Playfair square must end in the top right square too. So the end of 9a Entry (9aE) is 1. First cell filled! And this significantly limits the options for 6d.
- As 3d is the reverse of 2a, then their centre digits must be the same, so 3d must be of the form xxy, i.e. its first two digits are the same. Where 7a meets 3d, and with 7a defined as a divisor of 3d gives a very limited set of 7a/3d pairs in the form xxy. 3d can only be one of 558, 882, 992 and 996.
- The options for 5d as an anagram of 6d are again very limited, especially when it is clear that the 1 must be its central digit.
- After a while it becomes clear that 5d can only be either 316 or 613. As the 3 and the 6 are both in the centre of a row or column of the PF square, then neither 3 nor 6 can be in a corner of the PF square. This allows two options for the Playfair square to be created, one based on each value at 5d – remembering that in rows two and three of the PF square then there can only be three possible numbers in each cell. Using these partially completed squares then the options for 2d can be fully reduced.
- The last couple go in at 1a and 1d. One option can be eliminated as it results in 16 entered at 1a, which isn’t possible as it is already in at 7d. And the final one uses one of the cells that could be 0, at the end of 1d.

Double-check that all Answers and Entries align and I’m sorted. I hope! Feeling good about Numericals – I must soon have a go at the couple in the last Crossnumber Quarterly issue that I haven’t started yet (just in case the feeling wears off!).

One thought: I wonder how easy it is to use Sympathy software as a solver-assist tool for numericals? I have been creating a lot of custom Sympathy dictionaries (.tsd) recently for various puzzles I’ve been writing, but I have never delved much into using Sympathy for Numericals except as a general purpose editor. Could I get the Playfair gimmick to work on this puzzle for example? If anyone reading this has already done similar then I’d love to hear about it!

Cheers all,

Tim / Encota

P.S. Can you provide the answer to the following deep and meaningless question: **“What number do you get if you take all digits not used in the grid and place them together in descending order?”**

## Alastair Cuthbertson said

Well well. It seems that Piccadilly was a winner which is apt since he set the last numerical to involve a Playfair code square L3199 in 1993. I know I should get out more!