Specialized Crosswords

I don’t intend to write at length on this subject, because I know its fans, though enthusiastic, are a small minority; but the book would be incomplete without some mention of it. It was Torquemada who started such crosswords, and in his Torquemada Puzzle Book he divides those he used into four classes:

Examinations, i.e. puzzles in which all or some of the clues refer to a specified subject or author.
Inserted clue puzzles, where the clue numbers are inserted in a narrative which the answers complete.
Hidden clue puzzles, where the clues consist of single words or consecutive words in pieces of prose or verse, and the solver must find them by trial and error.
Missing clue puzzles, where certain words have no clues, but the solver is told that they fall under some particular heading.

He occasionally used other types, one of which happens to be my favourite specialized Torquemada puzzle, and I have chosen it to appear in the Appendix of this book. It might be called a distant relation of the fourth class. All these four types have been extensively used by later composers, especially in the Listener, most of whose crosswords have always been specialized. Hundreds of new types have been invented by its many composers, of which I shall mention here those which have fascinated me most.

Two were among the many creations of Afrit, “Printer’s Devilry”, nowadays the favourite specialized type of many Ximenes solvers, and ‘Playfair’, in which a few answers are concealed by the use of the Playfair code. This last type is a great favourite with some, but there are others who can’t cope with it: I therefore produce one once a year only. “Printer’s Devilry” I have chosen for the Appendix as an example of my specialized puzzles: its nature is fully explained there,

One type that I always used to enjoy in the Listener was the “Theme”, in which some of the solution could only be reached when an unnamed theme had been discovered by deduction.

This type was, I think, invented by Proton (Mr A. McIntyre), and I use something like it myself occasionally under the title “Spot the Theme’. This type especially well fulfils my chief demand from a specialized crossword, that its solution should involve the dropping of an interesting penny at some stage: the setter’s main difficulty is to see that it drops neither too quickly nor too slowly for the solver’s satisfaction.

Another type, which I have never been brave enough to imitate, though the herculean task of solving it was always fascinating, was the “Knight’s Move” of Cocos (Mr D. H. 8. Cox).

In this the down words made knight’s moves, as in chess, instead of going straight downwards: this made things very complicated, but, with sound clues, as they always were, it was a most satisfying struggle.

Of my own inventions I will mention three: “Misprints”, where a single letter is misleadingly misprinted either in each clue or in half the clues and half the answers; “Right and Left”, where with one exception (to give a start) each clue is really two clues, side by side, leading to two answers of the same length — the diagram is divided into two similar halves, and the solver must decide into which side each answer will fit; and thirdly “Theme and Variations’, which belongs really to Torquemada’s fourth type, clues not being given to a small number of themewords, e.g. the names of the Cinque Ports, nor to their pairs of variations, i.e. words connected in some way with them. This, again, is the “penny-dropping” type of puzzle, which I like best.

There are some types that I dislike solving myself and therefore don’t use. Foremost among these is Torquemada’s second class, the Narrative: in solving these I have always found irritating the search through the narrative for the number of a word that I think I may be able to get from letters already fitted into the diagram. More patient solvers probably don’t mind this process, but it is not for me. I am none too fond of the Examination type: it depends, for me, too much on knowledge and research rather than on ingenuity. Torquemada’s third class I use in a somewhat altered form, which I believe to have been invented by an experienced solver of mine, Mr T. W. Melluish, who used it, I think, for the first time in his classical crosswords which used to appear in Greece and Rome. In this form of that class a letter-mixture of the word required is hidden as well as a definition: this makes it easier for the solver to get a start. It is a weakness of this type that they tend to be very hard to start but too easy to finish, once one has got a start: the last quarter of the answers may tumble out in a rush. But they are great fun to compose, giving opportunities for fatuous humour of the nonsensical kind that I like; and I know they are popular with some solvers, so I serve one up about once a year.

The example of Afrit’s specialized crosswords that I have chosen for the Appendix comes under none of these headings.

By fiendish ingenuity he produced, as those of you who like a tough struggle will find, a puzzle in which every clue has two completely legitimate answers, covered by the whole of its length, so that the crossword is in effect two separate crosswords.

I once had a try at this, and it eventually appeared under the title “Double Entendre”, after I had derived many ideas from consultation by post with Afrit himself. Solvers asked for more, but it took such an unconscionable time to compose that I doubt if they will ever get it. I used the same idea for one quarter of a diagram, to provide a leg-pull, on the last occasion when April 1st fell on a Sunday. The pairs of answers all fitted into each other with the exception of one space, which made one set of answers obligatory for solving the puzzle. The leg-pull succeeded more violently than I expected, and there were only about forty correct solutions. What shall I do when April 1st is again a Sunday? Something dastardly, if I’m spared, but not the same as last time!