How to make a cryptogram

Put in a phrase, get a substitution puzzle with a blank for every letter and an answer key.

Enter your own message — a birthday greeting, a class motto, a treasure-hunt clue — and it becomes a puzzle to be decoded.

Step-by-step

1. Type your phrase.
2. Choose how many letters to reveal as a starting hint.
3. Generate and print, answer key on its own page.

Choosing a phrase

• Long enough to give the solver traction. Under about twenty-five letters there is too little repetition for frequency reasoning, and the puzzle becomes guesswork.
• Ordinary words. Names and invented words cannot be reasoned towards; the solver arrives at them only after everything else is done.
• Keep the apostrophes. They are one of the strongest footholds a solver has.

Setting the difficulty

The number of revealed letters is the main control. Revealing the E and T of a long phrase makes it approachable for a child; revealing nothing makes it a proper puzzle for an adult. Revealing a letter that appears often helps far more than revealing a rare one.

No letter stands for itself

The substitution is a derangement, so no letter ever maps to itself. This is the traditional convention and a real aid: a solver can immediately rule out the possibility that ciphertext H means H.

Why frequency is the way in

A substitution cipher keeps the shape of the language completely intact, and that is what makes it solvable rather than hopeless. In English, E is the most common letter by a clear margin, followed by T, A, O, I and N; so the symbol that appears most in a reasonably long puzzle is very often E, and the first guess is rarely wasted.

The structural clues are stronger still. A one-letter word is A or I and almost nothing else. A letter standing alone after an apostrophe near the end of a word is S or T. THE is the most common three-letter word in the language and its pattern is distinctive once two of its letters are known. Doubled letters are most often LL, SS, EE, OO or TT. None of this requires arithmetic — it is pattern recognition, which is why cryptograms are approachable without any background in ciphers.

What makes a puzzle fair

Those techniques all need material to work on, which is the main constraint when writing one. A very short text does not contain enough letters for frequency to mean anything, and the solver is reduced to guessing — forty or fifty letters is about the floor, and a hundred is comfortable.

The other trap is vocabulary. A proper noun the solver has never seen cannot be deduced, only stumbled into, because the pattern-matching that cracks ordinary words has nothing to match against. The same goes for unusual spellings, dialect and technical jargon. A good cryptogram text is ordinary language of reasonable length; the difficulty should come from the cipher, not from the solver failing to guess a surname.

Frequently asked questions

How long should the phrase be?

At least twenty-five letters. Shorter than that and there is not enough repetition for a solver to reason from, so it becomes guessing.

Can a letter represent itself?

No. The mapping is a derangement, which is the standard convention and helps the solver eliminate possibilities.

Should I include names?

Sparingly. Names cannot be deduced from language patterns, so they tend to be the last thing solved rather than a way in.