How to solve cryptograms: substitution cipher techniques

A cryptogram is a short text — usually a quotation — encrypted with a simple substitution cipher: every letter of the alphabet is consistently swapped for a different letter, so every E in the original might appear as Q, every T as M, and so on. There is no key. The solver reconstructs the substitution from the statistics of English: which letters are common, which words are short, and which patterns of repeated letters are possible.

One-letter words are A or I, and I is more likely at the start of a sentence. A letter after an apostrophe is S (plural or possessive), T (as in DON'T, CAN'T), D (I'D), M (I'M) or LL (WE'LL). Two letters after an apostrophe are usually RE, VE or LL. These placements cost nothing and each one reveals a symbol that appears throughout the quote.

Then count. The most frequent symbol in a quote of any length is very probably E; the next few are T, A, O, I and N. A symbol that appears many times but never at the end of a word is unlikely to be E (which ends many words) and more likely to be T, A or I.

Three-letter words are THE, AND, FOR, ARE, BUT, NOT, YOU, ALL, WAS, HER, HIS. A three-letter word whose last symbol is already E is almost always THE, which gives T and H at once. Two-letter words are OF, TO, IN, IT, IS, BE, AS, AT, SO, WE, HE, BY, OR, ON, DO, IF, ME, MY. Doubled letters in the cipher are doubled letters in the plaintext, and the doubled letters of English are, in order, LL, EE, SS, OO, TT, FF, RR, NN, PP, CC — a word pattern like _OO_ is very often GOOD, LOOK, BOOK, TOOK, FOOD, ROOM.

Longer words give up their endings. A four-symbol run ending a long word is often -TION, -MENT, -NESS or -ABLE; a three-symbol run is often -ING, -ION, -OUS, -ENT. The letters those endings reveal (I, N, G, T, O) are among the commonest, so they unlock everything.

Commit a letter as soon as it is probable, not only when it is certain. A cryptogram is cheap to undo and expensive to leave blank: each placed letter fills every occurrence, and half-revealed words are far easier to complete than blank ones. If a placement produces an impossible word (three consonants with no vowel, or a Q without a U), reverse it and try the next candidate. Solvers who insist on certainty solve slowly; solvers who commit and check solve fast.

In MindGlyph, assigning a letter propagates it through the whole quote at once, and the puzzles are genuine quotations, so the last few words can often be guessed from the sense of the sentence rather than the cipher.

Take the cipher fragment Q GMR KZW GWM. The one-letter word Q is A or I. W appears twice and is the commonest symbol, so W is probably E. KZW is then _ _E, which is almost always THE, giving K=T and Z=H. GWM is now _E_ and shares its first symbol with GMR; with Q as I, the line reads I G_R THE GE_, and SAW / SEA fit both at once: G=S, M=A, R=W. The quote is I SAW THE SEA. Every guess was checked against every other occurrence of the same symbol within seconds — that cascade is the whole method.