it was early spring war. and sultry glowef the .fternoon the very .reezes seemed to shawe the de.icious la.gu.r o. universal nature ae laden oh.various and mingled eerfu.es of the rpse and the jessamine the wio d.ine andtis wilddlower th.y sluwly wa.ted theia drtgrant o.fering to the npen rindow wheoe sat the lnvers tle arden. sun whont .uol uain her alushing dac. and fts gentle .eauny was more liee tee ryeatia of sime yi.d rumon.g oa the .airy rnspi.atiin od d dream than tha actual realitw of earth ttnderly h.w alver gazed ueon her as her cluwteroug ringlets were edved .y a.orous and spoytive zep.yrs and when he permemted the rude intrusion od the sunlight he sprang tu draw the curt.hu aut ste gently stayud him ne ni dear charles she softay sard much wathea y.eld n have a little zun then nn ain at ralprobably to be interpreted as:
It was early spring, warm and sultry glowed the afternoon. The very breezes seemed to share the delicious languor of universal nature (and) laden of various and mingled perfumes of the rose and the jessamine, the woodbine and the wildflower, they slowly wafted their fragrant offering to the open window where sat the lovers. The ardent sun shone full upon her blushing face, and its gentle beauty was more like the pyrexia of some wild romance or the fairy inspiration of a dream than the actual reality of earth. Tenderly her lover gazed upon her as her clustering ringlets were eased by amorous and sportive zephyrs, and when he (perceived) the rude intrusion of the sunlight, he sprang to draw the curtain. But she gently stayed him, "No, no dear Charles", she softly said, "much rather would I have a little sun than no air at all."My respect to the person who solved it, taking up the challenge even as originally intended, with no help from computers, simply sitting down and permuting letters. The solution was first announced on the Poecipher e-group (see link below).
| poe crypto contest: | http://www.bokler.com/eapoe.html |
| slashdot discussion: | http://slashdot.org/articles/00/03/09/0112237.shtml |
| poecipher egroup: | http://www.egroups.com/group/PoeCipher |
| original article of Graham's Magazine, Dec 1841: | http://www.eapoe.org/works/essays/gm41sw03.htm |
The cryptogram consists of 6 variants of the alphabet, upper-case, lower-case and small-caps each both upright and upside down. The letters are very equally distributed, evidently to discourage frequency counts. Only 12 out of 156 letters are missing altogether,
#w,!b,!l,!o,!p,!q,!s,!x,!#o,!#v,!#x and !#z.
(where ! indicates inversion, # smallcaps), all other letters occur at least once and not more than 11 times (e, of all letters, has this maximal count)
Now if we just had that cipher on a piece of paper, with no known context, we wouldn't know where to begin, for all we knew it could be noise or an obfuscated code rather than a cipher with no hope of ever being cracked. But this one was published in a magazine as a challenge to the readers who through many issues had been following Poe's arcticles on cryptography. It is presented as a letter by one W. B. Tyler to Poe, claiming to present a perfect ("insoluble") cipher.
Poe refutes this claim somewhat mockingly and passes the examples provided as an exercise to his readers, explaining he has no time to bother with it. The first example in the letter has been found to be meaningful and solved only in 1991 by Th. Whalen, and it proved to be a simple substitution cipher with the individual words written backwards.
The suspicion has been raised whether Tyler might not be Poe himself posing as one of his readers. I think this is very probable. You see, there were no typewriters in 1841. So unless Tyler had his own printing press, we should naturally assume his letter had been handwritten. Now, if you needed 6 variants of the alphabet, would you choose smallcaps? I wouldn't. I'd probably choose upper and lower case, unadorned and underlined and 'overlined' or something similarly practicable, rather than engaging in the penning of reversed characters and painstaking distinction of character size. Yet Tyler is quoted as follows:
"I have employed the Roman-capital, small letter, and small capital, with their several inversions..."Are we to assume Poe has silently adapted the character sets for printing and changed the text of the letter accordingly without any editorial note? Incidentially, underlining would have been preferable to inverting letters even in printing, as in some instances it becomes nearly impossible to distinguish whether a letter is inverted or not. I have typed the cryptogram and compared my version to one posted on slashdot. Between the two versions there were 11 discrepancies, 8 due to disputable reversal of smallcap H, O, I, N or Z and 3 due to confusion of n or p with inverse u or d. With this difficulty in mind, consider that Tyler goes on to state:
"...giving me the command of 130 characters, or an average of five to each letter."while 6*26=156; this obvious inconsistency is simply screaming for your attention. It has been observed that the opening paragraph by Poe contains exactly 130 letters:
"The annexed letter from a gentleman whose abilities we very highly respect, was received, unfortunately, at too late a period to appear in our November number:"However, I find it unlikely that this paragraph represent a substitution key, because it is lacking the letters j,k,q,z. But let's see, eg. line 4: (hope you see this in fixedwidth)
|
theannexedletterfromagentl ABCDEFGHIJKLMNOPQRSTUVWXYZ |
|
emanwhoseabilitiesweveryhi abcdefghijklmnopqrstuvwxyz |
|
ghlyrespectwasreceivedunfo #abcdefghijklmnopqrstuvwxyz |
|
rtunatelyattoolateaperiodt !ABCDEFGHIJKLMNOPQRSTUVWXYZ |
|
oappearinournovembernumber !abcdefghijklmnopqrstuvwxyz |
|
!S!TA!V!an!#a!Z#t#c!GDYR!C ... a pt r oi ? t v l eatr u |
The question is, which of the alphabets is the odd one out? Either of the smallcaps ones because of the risk of confusion? (which wouldn't help much though, as you'd still have to decide whether a letter is of this class or not) More involved possibilities exist, of course, like ignoring the 26 most easily confounded letters, or picking the next letter of each alphabet in turn, or two messages intertwined, each modifying the key for the other message's next letter (unlikely because there would not be enough entropy left for the redundancy, as proclaimed and observed), or some wicked effect of the leftover alphabet on the substitution key, but all this seems rather too far-fetched for someone who has no computer at hand and in a context where already a one-to-many key is accused of being too taxing.
Remaining to be addressed is the question of spaces. I think they are either real or to be ignored. Of course the space marks or the letter-counts of the 'words' could have some wicked effect as well, e.g. as simple as "drop every first (or nth!) letter of a word" Since Poe drops word separation even in his simple ciphers, I see no reason he should mark them in his most obscure.
Poe writes to one Bolton regarding the cipher:
"It is unnecessary to trouble yourself with the cipher printed in our Dec. number -- it is insoluble for the reason that it is merely type in pi or something near it. Being absent from the office for a short time, I did not see a proof and the compositors have made a complete medley. It has not even a remote resemblance to the MS."(what does he mean by "type in pi"? should it read "pi(=3.14159..) in type"? obviously the required meaning is "noise") This is from a letter to an individual who had proven his ability at solving riddles, so while the article may well be contrived to humble the hubris of some of his readers (and the letter a parody of such readers), this sounds like a genuine warning to somebody respected. We must assume the cryptogram was indeed intended to be meaningful, whether we think Tyler is a real person or an alias of Poe's, already because of the effort put into composing and typesetting it, (and because it would be altogether too cheap for Poe to publish white noise?) but the printers have jumbled it so much Poe considered it effectively undecipherable as printed. The problem is that Poe is dissuading Bolton from "the cipher printed in our Dec. number" without specifying which one, while the first of these turned out to be decipherable enough, with only minor misprints. Still, it is conceivable that Poe was so preoccupied by the second cipher that he didn't mention the first one either by negligence or because he thought it unworthy of his or Bolton's attention.
In conclusion, it seems likely Poe was posing as somebody else to be able to violate his own principle of ease of decryption to propose a final exercise to his readers. In his own name, he deprecates the cipher for being too tiresome:
"Lord Bacon very properly defines three essentials in secret correspondence. It is required, first, that the cipher be such as to elude suspicion of being a cipher; secondly, that its alphabet be so simple of formation as to demand but little time in the construction of an epistle; thirdly, that it shall be absolutely insoluble without the key - we may add, fourthly, that, with the key, it be promptly and certainly decipherable."Unless the MS of the cryptogram can be recovered we are left with a jumbled version that Poe himself considered unsolvable. But Poe didn't think of the possibility of brute force CPU-power. Therefore a systematic attack of the cipher might still stand a chance of success, but it should anticipate a lot of misprints, ideally permute all remotely similar characters. So this is how I would do it: Drop the spaces and start permuting, looking for patterns like 'the', 'and', and, in a boustophedonic spirit, 'eht' and 'dna', in keys where such patterns occur repeatedly, start scanning for other dictionary words, with a high tolerance of typos and interposed garbage. The search space will still be enormous, especially if you allow for arbitrary 'dummy letters'; the search should start with keys that ignore one of the alphabets. This is assuming the message is in English, of course, and not engineered in a Perekesque way (no e's and things), which would be unfair, as the claim was to present a secure cipher, and secure would mean you could encode any message you like [Right, strictly this would rule out any substitution-based key because of frequency analysis of phrases, c.f. the Enigma code which was compromised because the Nazis had to end every message on "Heil Hitler", but let us say, secure if not used repeatedly]