Solving Cipher Secrets

SOLVING CIPHER SECRETS

Edited by M. E. Ghaver

FOR THE FIRST TIME HERE, THE SOLUTION OF A NUMBER OF CRYPTO•

GRAMS IS EXPLAINED-ALSO A TIP ABOUT THE RADIO CONTEST CIPHER

FTENTIMES a cryptogram In this article for the first time we will may be so short that—de• actually consider the solution of a number pending, of course, on the of cryptograms. The cipher selected for relative complexity of the this purpose is one of the numerous variants cipher — its solution be• of the famous Vigenere alphabetic square, comes a difficult matter, being that given by John Wilkins—after• if not altogether an impossibility. ward Bishop of Chester—on pages 72 to 76, In such cases decipherment can often be inclusive, of his " Mercury, or the Secret materially simplified if a number of cryp• and Swift Messenger," an early work on tograms in the same key are available. cryptography published in London in 1641, With one exception, noted below, all the This form of the cipher uses the same methods so far discussed in this department type of alphabet as its famous original, and have depended upon the analysis of a sin• is identical in its results, but holds one ad• gle cryptogram. vantage, at least, over it, in that instead of In some instances single cryptogram requiring a ready-made table of the whole methods may be used with a number of number of alphabets, it employs a special messages, the added effectiveness being due table, formed of just those alphabets se• to the increased bulk of material. On the lected by the key word. other hand, some of these multiple message For example, if the key word TRY be methods are peculiarly adapted to a number agreed upon, the table will consist of three of cryptograms, not being applicable to the alphabets, one beginning with each letter of resolution of a single example. the key. The alphabets used by Wilkins A certain insight into multiple crypto• consist of but twenty-four letters, / and V gram methods has already been afforded being omitted. For in the English alphabet readers of FLYNN'S WEEKLY in the issue of of that time the letters / and / were used July 3, where a method was given applicable interchangeably, as were also U and V. to the solution of a number of transposition Here, however, the full twenty-six letter cryptograms in the same key. alphabet is employed.

(Message alphabet) A B C D E F G 11 I J K L M N 0 F Q R S T U V W X Y Z T TUVWXYZABCDEFGHI JKLMNOPQRS R RSTUVWXYZABCDEFGHI J KLMNOPQ Y YZABCDEFGHIJKLMNOPQRSTUVWX

(Cipher alphabets) 794 SOLVING CIPHER SECRETS 795

Now, to encipher any message, as the quired—after each cipher letter, taking care short example given at (b), write a letter to preserve the columnar arrangement, the of the key above each letter of the message, cryptogram itself constituting the first col• taking both in their order, and repeating umn. The letters forming any word of the key as the length of the message may the message will then appear in a single col• require, as shown at (a): umn, under the key letter by which it was enciphered. (a) Key: TRYT RY TR YTRY (Key) (b) Message; MEET ME AT ONCE (c) Cipher: FVCM DC TK MGTC Z YXWVUT SRQ P... (d) Regrouped: FVCMD CTKMG TC F GHl J KLMNOPQ... X Y Z .4 B C D /; F G H I . . . The key letter above each text letter ,iow X VZABCDA; FGHI... M NOPQRSr UVWX... indicates the alphabet in which it is to be D EFGHIJK LA/. .. enciphered. For example, the first letter of VWXYZABCDA'. .. Y Z B C D E . . . the message, M, is to be enciphered in the R S V U V W X . , . HIJKLMNOP... T alphabet, its substitute in that alphabet G n I j K L M .V O . . . being F, which is accordingly placed below VWXYZABCD. . . X YZABCD£'F . , . M in line (c). Similarly, the second letter, E, of the Of course, if a whole line of the crypto message becomes V in the R alphabet. The gram is enciphered in the same alphabet, third letter, also an E, becomes C in the or if normal word divisions are observed, y alphabet; and so on. The cryptogram decipherment by the above method is still for MEET ME AT ONCE, with the key easier. TRY, is thus FVCM DC TK MGTC, as Now that the reader is acquainted with shown at (c). the system, it is well to produce the cryp• As described by Wilkins, the normal word tograms he is expected to decipher. To in• divisions are observed in the cryptogram. crease the interest, these cryptograms have In the present day, however, the customary been made the captured correspondence of procedure would be to regroup the letters, a supposed hand of kidnapers. preferably by fives, as shown at (d). In these cryptograms, the key letter has In passing, it must be mentioned that been changed with each text letter; also, ex• Wilkins also suggests the use of mixed cepting that a twenty-six letter alphabet alphabets in this cipher instead of the has been used, and normal word divisions straight A-to-Z arrangement, and the have not been ohsdrved, the cipher is other• change of key letter with each word or line wise exactly like that described by Wilkins. instead of with each letter. The resolution Here ape our cryptograms, six of them, of mixed alphabets will he taken up later. numbered for reference.

/J 20 -y so ss (1) JOPTE CGENS NTFDF PPGZD THBTM WXCCD JXPU. (2) OSYIV CHXIV NUQFZ XPOFP TFUCI SGTPH M. (3) LOETS MUTYE KHGII VVKTFN GPNLW. (4) YSBPX SNPMS HQQMN DVVYR QIHIU OJPTH GTPH. (•;) CLEGG DVSVK MHRUY FAICI PJIKl ELNKM KVVW. (6) TVYJJ SAHVV AZRVE VFTFT -SFRRR KVM.

But both the other devices could easily he Now, since this is supposed to be a series solved by " running down the alphabet." of cryptograms in unknown cipher, it can• Suppose, for example, that the key letter not be assumed that they are in the same he changed with each word, thus: key. Possibly a number of different ciphers have been employed. Again they may be (a) Key: TTTT RR YY TTTT in the same system, hut with different keys. (b) Message : MEET ME AT ONCE (c) Cipher: FXXM DV YR HGVX Consequently, before any two or more of (d) Regrouped: FXXMD VYRHG VX the above cryptograms can he combined • To solve this, it is only necessary to write for solution, it is necessary to know that the alphabet—all twenty-six letters if re• they are in identical ciphers. 796 FLYNN'S WEEKLY

If the reader desires he can try each gram. Thus, the group PT occurs at the cryptogram individually by the transposi• third letter of cryptogram No. i; at the tion test given in the September 4 issue of twentieth letter of No. 2; and at the twenty^ FLYNN'S WEEKLY. These tests will elimi• eighth letter of No. 4. The reader may nate the transposition cipher, and allow us easily check them. to consider the possibility of substitution Were there enough recurrent groups in ciphers. each cryptogram, they could he tested by In this latter class, the substitutes may the Kasiski method in the usual manner, consist of one, two, three, or more charac• See FLYNN'S WEEKLY for August 7, 1926. ters. The present cryptograms consist of Here all such groups happen to he acci• 34, 31, 25, 34, 35, and 28 letters, respec• dentals, which are of no value by the above tively. Some of these numbers are not method. evenly divisible by 2, 3, 4 . . . and con• The Kasiski principle, however, is not sequently we can assume—unless substi• limited in its application merely to a single tutes of mixed lengths have been used— cryptogram. It can be used just as well that at least some of the cryptograms are with any number of cryptograms, the re• of the straight letter-for-letter substitution current groups of which can be treated ex• type. actly as if they occurred in a single speci• Messages enciphered in the same key will men. To illustrate this, suppose that we often have similar predominating characters examine the groups found in both No. i or groups of characters. For instance, that and No. 2. cyptograms Nos. 5 and 6 both contain a Groups (2) Intervals large number of V's might be pointed out (il Factors PT 20 as significant. 3 17 17 TF 12 21 9 3-9 Predominant groups, however, will re• FP IS 19 4 2-4 ceive the bulk of attention here. And to XP 32 16 16 2-4-8-16 save our readers a few hours of merely rou• tine work, a complete table of all the re• It will he seen that the intervals current groups in all six of the present cryp• figured here exactly as if the recurrent tograms is herewith appended. groups occurred at their respective numeri• cal places in one and the same cryptogram, (1) (2) (3) (4) (s) (6)_ PT 20 28 — instead of two. The largest predominating 3 — — SN 10 — — 6 — — factor is 4, which suggests a fixed period TF 12 21 18 — — 18 cipher using four alphabets. By this sup• FP IS 19 — — — position PT and TF become accidental re• TH 21 — — — 29 current groups. At least, we may progress XP —16 — — — — with that assumption. IV — 4-9 — — — — CI — 24 — — 19 — The supposed natural or periodic groups, GTPH — 27 — 31 — — FP and XP, however, might have resulted HG 12 30 — from using different keys of the same length, DV — — 16 —6 — — but with certain characters ~ in common. W — — — 17 32-33-34 9 VY — — — 18 — 2 Just what fxirtions of these keys must here KM — — — — 10-29 — be identical, if they are not entirely so, can KV — 31 26 be found by transcribing the two crypto• FT — — — 17-19 — — — — — grams in groups of four, the supposed The above differs from the ordinary table period. FP and XP are underlined to facil• of recurrent groups in that it lists the itate the examination.

1234 1234 1234 1234 1234 1234 1234 1234 12

No. I JOPT ECGE NSNT FDFP PGZD THBT MWXC CDJX PU No. 2 OSYI VCHX IVNU QFZX POFP TFUC ISGT PHM groups repeated in all six cryptograms, and By this arrangement the group FP will not merely those of each individual crypto- be found to occur at the third and fourth SOLVING CIPHER SECRETS 797

letters of the period; and XP at the fourth the greater the amount of material in any and first. Consequently, unless these groups cipher, the less difficult is it to solve. are also purely accidental, it may be as• For example, cryptogram No. i when sumed, tentatively at any rate, that both its thirty-four Characters are divided among these cryptograms are in four alphabet ci• its four alphabets, has only eight or nine phers, whose first, third, and fourth alpha• characters in each alphabet. And No. 2 bets, at least, are identical. has only seven or eight characters per alpha• Were these cryptograms longer, or with bet. To resolve even a simple alphabet more groups in common, the results by this with so few characters is sometimes a diffi• method might he more convincing. Never• cult matter. theless, sufficient have been shown—espe• If the similar alphabets of these crypto• cially since these cryptograms are from the grams—and any others in the same key— same source, the kidnapers, you know—to are combined, however, larger frequency make it highly probable that both crypto• tables of the several alphabets will he avail• grams have the same key. able for analysis. Just as the presence of identical recur• To illustrate, the frequency table for rent groups at regular intervals in a num• alphabet No. i of cryptograms Nos. i and ber of cryptograms can thus he taken as 2 would here consist of tlie first characters evidence of a common key, so the absence in every group of four; these being the of such groups may he considered as indi• letters /, E, N, F, and so on, of No. i, cating different keys. and 0, V, I, Q, and so on, of No. 2. Tables Thus, No. I and No. 3 have only one for the other alphabets would be similarly group, TF, in common. This, occurring at formed. the twelfth letter of No. i and the eight• The several alphabets so isolated can he eenth letter of No. 2, gives an interval of resolved by any desired method: as, for in• 6, suggesting a period of 2, 3, or 6. stance, by assigning the value E,T, A . . . We already have stronger evidence, how• to one of the most used characters, as de• ever, that No. I has a period of 4. No. 3 scribed in FLYNN'S WEEKLY for August 7, would thus seem to he in a different key 1926; or by detecting the amount of dis• than No. i. What the length of its key placement, or shifting, of "a known cipher is can be discovered by comparing it by alphabet, as shown in the issue of October 2. the method already shown with other cryp• The actual work of sorting the present tograms of the series. six cryptograms according to their keys, Having thus explained how it is possible and of solving the different alphabets, to find whether or not two or more crypto• should afford the reader an hour or so of grams of this kind are in the same key, it' fascinating entertainment. is left for the reader to determine for him• And don't forget, either, that these cryp• self how many keys have been employed in tograms, deciphered, will reveal the story of the present instance, and what cryptograms the kidnapers. have been enciphered in each. Who will he the first to discover their These points determined, such crypto• plot? grams as are in the same key can he com• Send in your solutions; and look for the bined for solution. For, as a general rule, translations in next Solving Cipher Secrets.

WITH OUR READERS There are some points about ciphers from the best advantage. Some ciphers will be our readers that we must take this opportu• printed along with special articles dealing nity to mention. with similar systems. First of all, do not he disappointed if Another matter of importance is that the cipher you submit is not used at once. your ciphers or inquiries, to receive atten• Many excellent ciphers are being held until tion, must always he accompanied with your the time when they can he presented to correct name and address. If you prefer, 798 FLYNN'S WEEKLY

your name will not be printed in the maga• given the female cipher fans anything par• zine. But unsigned, or anonymous, com• ticularly adapted to their intuitions. Here is a cryptic message which was smug• munications will not be considered. gled out of prison by a woman to let her In this connection a number of interest• friends know what to send her in aiding her ing unsigned ciphers have been received. to escape. One of these was based on the wigwag Let's see what the ladies can make of it! alphabet. Another, from Southington, MALCOLM DEAN MILLER, M.D. Akron, Ohio. Connecticut, used false word divisions. These, and others like them, we are sure, CIPHER No. 7 (Malcplm Dean Miller, M.D.): Metol 9 grains would interest our readers. We regret that Hydroquinone 38 grains we cannot publish them. Sodium sulphite, dry...„ 60s grains Finally, we have been in receipt of a Sodium carbonate, dry 800 grains number of insistent requests that we pro• Water 20 ounces vide an adequate means of placing those " Imogene " interested in cryptography in contact with The next cipher, though easy, is, never• each other. theless, not without interest. Having After a careful consideration of all the guessed at a word of the message, try to ideas advanced as to how this could best be discover the plan of the alphabet. It is done, we believe that the most feasible plan, to a famous statement of a celebrated au• for the present at least, would he to list thor that Mr. Duree (Los Angeles, Cali• in these columns the names and addresses of fornia) refers in his message. those who desire to correspond on the sub• Our correspondent says this cipher is ject of cryptography. sometimes used by bootleggers hack in the How does this idea strike you? If you effete East.

CIPHER No. 8 (Mumon Duree): 36 48 44 l8 14 52 36 26 4 46 24 52 40 iS 44 44 8 36 14 38 44 46 40 52 18 52 30 30 52 26 22 24 44 42 24 18 14 38 44 38 12 28 52 26 28 36 26 46 36 16 16 12 iS 44 30 4 36 26 40 44 6 36 24 12 16 would like to have your name on such a Mr. Goldman, Cleveland, Ohio, origina• list, write us at once to that effect. Whether tor of No. 9, saj's that his system can be or not this plan is put into operation will learned in an hour's time. He has attempt• depend upon the enthusiasm with Avhich it ed to make an undecipherable system. Will is received by our readers. it get by the enterprising readers of Now that these affairs are all cleared FLYNN'S WEEKLY? away, we can place some readers' ciphers CIPHER No. 9 (B. Goldman) : before you. And good ones they are, too! EJ GEBUF XEPIQLEFIC GEBIQEC In No. 7 photographic fans will recognize SQUZIJMEH SEP NEBILEF VO the formula of a well known developer. BEM VEMITEPIKUBAKIF DEPICEF, Innocent enough, apparently, it has yet .WIS EJ TEFF SEGIBES ZEPIV been bent to the purposes of cryptic com• GEBUF OVIS PEMIF PUFEQ PEM NEF BEMIC EJ XEJIKK GEBUF munication. This is different from anything SEP SEQUZ BEHIBEJIM. we have yet printed. Read Dr. Miller's let• ter, and try to find the meaning of the The solutions to all the ciphers in this cipher: issue will he given in next Solving Cipher Secrets. Don't fail to see it. In the mean• DEAR SIR : Although your department is most interest• time, try your skill, and send us your solu• ing, it has occurred to me that you have never tions. ABOUT THE RADIO CONTEST CIPHER A complete explanation, with special Radio Cipher, which wa s No. 3 in the method of solution, of Mr. C. A. Castle's October 2 issue of FLYNI r's WEEKLY, will SOLVING CIPHER SECRETS 799 be given in the next installment of this concerned, but they are the first six letters department. of the key. They are the key to tlie next six letters of the cipher, which in turn are the key No method of solving this type of cipher to the following six. The message may be de• has yet been published in these columns. ciphered by using a table, or by writing the In fact, to discover how his cipher could cipher under itself, six places to the right, he solved was one of the reasons that subtracting the numerical values of the key from those of the letters of the cipher (add• prompted Mr. Castle to offer the seventy- ing 26 where necessary), and changing the re• five-dollar radio set. sulting differences into letters of the message. Now we really anticipate some solutions To illustrate:

F R J P A C K N 0 H Z F —T D VV . . , 6 iS 10 16 1 3 II 14 IS 8 26 6 20 4 23 . . . 6 i8 10 16 1 5 II 14 15 . . .

S 22 s 18 25 3 9 16 8 . . . E V E R Y C I P H . . ,

Message: Every cipher formed on this plan carries its own key and is its own key. There is nothing to remember except the general principle and the numerical order of the letters of the alphabet. to Mr. Castle's cipher. And if the expected No. 5 (Paul Napier), like Nos. i and list of solvers actually materializes it will 2, used the digraph table, without, how• be printed in an early issue, naming the ever, including the dash or word space. fortunate winner. Look for it. Accordingly, his alphabets used but twenty- Here are the solutions to the remaining six characters, the right-hand column and ciphers of the October 2 issue: the bottom row of the table being disre• Cipher No. i. Key: L—B. Message: garded. Mr. Napier's message is based on We expect to hold up the messenger a statement in BuUivant's hook, mentioned Wednesday noon. Be ready to act as ad• in the preceding article. Its solution is vised. presented herewith. Cipher No. 2. Key: 0—Z. Message: We leave to-night for San Francisco. Meet Key: K—S. Message: This cipher is based us at Jim's place, and bring the money on a system of digraphs. It was used by a well known European country. As far as I with you. know, it is still used. No. 4 (R. M. Packard) was an in• genious autokey cipher, using the straight In Mr. Levine's, No. 6, we have a re• x-to-26, A-to-Z alphabet. Mr. Packard's markably simple and efficacious method of own explanation of his system is ap• combining two messages in the same cryp• pended: togram. This cipher is based on the fol• This is a variation of the Gronsfeld cipher. lowing two alphabets:

A B c D E F G H I J K L M I 3 3 4 S 6 7 8 9 10 II 12 27 38 29 30 31 32 33 34 35 36 37 38 39

N 0 P Q R s T U V w X Y Z 14 IS 16 17 18 19 20 21 2: 23 24 25 26 40 41 42 43 44 45 46 47 48 49 , 50 51 52

using letters for key and cipher, and extend• To illustrate the method of encipherment, ing your Gronsfeld table to the full alpha• take the first letters of the two present mes• bet, but differing from it by one letter in every place, and giving results equivalent to sages, C (29) of the first message, COME adding the numerical values of the letters of HERE, and S (19) of the second message, the key to those of the letters of the mes• STAY AW.AY. The substitute for these sage, the numerical value of each letter being two letters is 24 5, one of which numbers .its order in the alphabet. The whole cipher is the key. The first (24) is half the sum of 29 and 19, and six letters are nulls as far as the message is the other (5) the difference between this 800 FLYNN'S WEEKLY

24 and either 29 or 19. Mr. Levine ex• text letter is represented in cipher by the presses this relation algebraically, thus: letter directly below it:

2:-l-y = 2g CRYPT x — y=ig A B D E F 2X = 48 G H I J K 3: =24 L M N O Q y= s S U V W X Z . , In deciphering, the numbers must be taken Text: ABCDEFGHIJKLM in pairs. To get the first message, COME Cipher: CAGLSZRBHMUYD HERE, add the numbers of each pair. The Text: NOPQRSTUVWXYZ sums will indicate the letters in the 27-to- Cipher :INVPEJOWTFKQX 52 alphabet. The second message, STAY Using this alphabet, our crypt becomes: AWAY, is deciphered by taking the differ• Wicked iceman .pleads guilty; admits hav• ences between the numbers of each pair, ing stolen costly nickel-plated bucket while these indicating letters in the i-to-26 alpha• honest owners busily played casino near by. bet. The word crypt is here adapted from puz-

Cryptogram: 24 3 30^2 10^2 20 ig 28 3 . . . zledom, where it is applied to the kind of cipher just given, and which we know as Sums; 2g 41 39 31 Message No. i: C O M E . . . a normally spaced, literal, simple substitu• Differences: ig 20 i 23 tion cipher. Most crypts also employ an Message No. 2: S T A Y . . . unusual wording of the message to make decipherment more difficult. A system of this kind offers the advan• In our own example, for instance, you tage of sending a real and a supposed mes• will note that no short words, which are sage in the same cryptogram. Should the usually easy to identify, have been em• cryptogram be intercepted, suspicion might ployed. Further, no word uses the same be diverted by explaining the supposed letter twice. Words of any and all lengths message. The real message might in this can be used in the same crypt. We have way escape detection. Incidentally, frac• used all six-letter words for purely fanciful tions could be avoided in this cipher by reasons. Our alphabet was based on a using only the even numbers from 2 to 104. key word. Any other kind of literal alpha• Ingenious cipher fans will find a thou• bet is permissible in a crypt. sand and one occasions in which a crypto• A total of twenty-seven solutions make gram of this sort may be useful. Prying up our list of August 7 solvers. The most eyes are likely to accept the supposed mes• complete set of solutions was sent in by sage as the ultimate one and neglect fur• Mr. Winsor, who walked off with all but ther attempts to read it. Without hav• No. 3. The arrangement indicates the ing achieved an indecipherable cipher, its order in which solutions were submitted, user may have accomplished similar ends. and has no other signification. No. 7 was in the simple substitution al• Philip Bosso, New York, N. Y. (i). phabet given in the next column, having Charles C. Fulton, Omaha, Nebraska (1-2). been formed on the key word, CRYPT, in Richard A. Miller, Ridgway, Illinois (i- accordance with the following method: 2-6-7). In preparing an alphabet of this kind, Paul A. Napier, Louisville, Kentucky (1-7). first write down the selected key word. Charles P. Winsor, Boston, Massachusetts Then below it, in lines of the same length, (1-2-4-3-6-7). M. Walker, Akron, Ohio (1-2-4-6-7). place the remaining letters of the alphabet, Arthur Bellamy, Boston, Massachusetts forming a number of columns equal to the (1-2-6-7). number of letters in the particular key A. P. Schmutz, Philadelphia, Pennsylvania word. (1-2-7). Now, take the columns so formed down• The September 4 solvers' list, if com-. ward, and from left to right, as shown in piled in time, will be printed in next Solv• the subjoined cipher alphabet, where each ing Cipher Secrets. 10 F W