<<

SOLVING SECRETS

Edited by M. E. Ohaver

A PROFITABLE LESSON ON THE SOLUTION OF A CRYPTOGRAM IN AN UNKNOWN CIPHER SYSTEM-ALSO READER

O solve a cryptogram in a cipher may be so unapparent as to require known cipher system, it is the application of delicate cryptographic only necessary to apply tests to discover their presence. methods peculiar to that If a comparison is again allowable, this cipher. But suppose the would be like identifying the man by his system is unknown. How, Bertillon measurements, or by his finger• then, would it be possible to determine the prints. method of solution? To demonstrate the practicability of these That the fans are much concerned with tesf s, we have applied some of them in this this question would seem to be indicated article to a cryptogram that was submitted by the volume of mail from readers who to this department without solution or ex• stale that they have successfully used the planation. different methods described in these columns This cryptogram was chosen from many for various ciphers, but who wonder if there similar ones for several reasons. In the first is any way of finding what method to use place, it illustrates all the points brought when the cipher is unknown. up by the various tests. Again, it happens Fortunately, in many cases a cipher will to be in a standard system that we wanted leave an indelible impress on a cryptogram, to present to our readers, anyway. And, allowing the system to be identified, or, at finally, the fact that the inditer was abso• any rate, to be recognized within limits, by lutely certain his specimen could not be de• certain more or less easily recognizable ciphered—an opinion he no longer holds—• characteristics. adds zest to the problem. Sometimes these peculiarities are distin• The cipher was submitted through an• guishable at sight. To illustrate this, con• other reader of FLVNN'S WEEKLY, who ac-' sider the Nihilist numerical cipher, the num• companied it with the following letter. bers of which are within the limits 22 DE.SR SIR: and 110. I inclose a message written in cipher. This To determine a cipher system in this way was made up by a man who has used the is somewhat like identifying a man on'the same for private messages. street by the color of his eyes or hair; by a He claims his cipher cannot be solved. For you, and your department, I accepted the missing right index finger; or by a manner• challenge, and am sending on the message. ism of gait or gesture. SAM'L J. MCNARV. On tlie other hand, the peculiarities of a Cincinnati, Ohio. IS) V 154 FLYNN'S WFFKLY

Things begin to look interesting already. the groups are given herewith in this in• Suppose we have a look at the cryptogram stance.

Itself? A 8 L s J 5 E 3 N I KB ol jmepprqvhoi j I 3 R 4 Q 3 mzphxzuwvbhnauo - O 10 SI X 2 Ivuohphabkskle U 7 T s Z 4 tlulvbfbwopwtp houvryucxbbatd 31 16 22 zbqpyaat jdeyiap (25.4%) (13.1%) (18.0%) fkkkokrgcutqoak pyi rjpwofbkohw A difference of more than 5% in any t p d K id z.— group from the 40%-30%-2% averages could be taken as evidence that the cipher The several tests about to be given in• is not of the transposition class. Here all volve the number of limes various cipher three groups are outside their respective characters are used. Accordingly, a table limits, rendering such a conclusion even of their frequencies Is herewith appended: more probable. Having thus disposed, apparently, of the A 8- H 7 O lO V S B 8 I 3 P II W S transposition possibility, we will now pro• C 2 J 5 Q 3 X 2 ceed to discover if the cipher is of the sub• D 4 K 8 R 4 Y 4 stitution class; and, if so, whether the char• S I Z 4 E 3 L 5 acters are fixed in their values as in the sim• F 3 M 2 T 5 ple , or variable as in G 2 N I U 7 122 multiple alphabet and other varieties of Had this cryptogram been one of num• substitution ciphers. bers or signs, we could have at once as- In the simple substitution cipher, where siuned that it was of the substitution vari• a given character always represents the ety. Since it is literal, however, it can be same letter, quite frequently the number of one of either the substitution, transposi• different characters in a cryptogram will be tion, or null class, not to mention combina• less than twenty-six, since one or more let• tions of these. ters of the alphabet are often unused even If our specimen is a , in long messages. it will react positively to the vowel-conso• A'peculiarity, however, of ciphers em• nant group test, given by Parker Hitt in ploying characters of variable values, is that his " Manual for the Solution of Military almost always all of the characters will be Ciphers," a work formerly published by present even in short cryptograms. The the Army Service Schools Press, Leaven• present cipher would thus seem to be one of worth, Kansas, but now, unfortunately, out this kind, using, as it does, all twenty-six of print. letters, presumably the whole number of This test is based on the fact that in characters employed by the cipher. average English text the total frequencies Another characteristic of the simple sub• of the vowels AEIOU, and the consonants stitution cipher is that repeated words will LNRST and JKQXZ, will ordinarily not at each recurrence be represented by the vary more than 5% one way or the other same cipher characters. The present cryp• from 40%, 30%, and 2%, respectively, of togram, as shown, contains a number of the total number of letters. The figures two-letter recurrent groups, and one of have been obtained by countless experi• three letters, but none of any greater length, ments. which would be likely if the cipher were A common method of applying the test is of the simple substitution type. to first count the vowels directly from the .Another aid in recognizing the variable cryptogram, not taking the consonant substitute cipher is that ordinarily it af• counts unless the vowel count falls within fords no characters of either extremely high the prescribed 35%-45% limits. For illus• or low frequencies, corresponding respec• trative purposes, however, the counts for all tively to the substitutes for F, T, A, 0, N, SOLVING CIPHER SECRETS 155 et cetera, and J, K, Q, X, Z, et cetera, of substitute cipher will in this case be more the simple substitution cipher; the tendency than 2%, approaching 20% as a limit. In being, on the other hand, for all characters applying these tests due allowance must lie to approach an average frequency of made where substitutes for the word-space, (ioo% 26 letters—3.85%) approxi• punctuation marks, figures, and so on, may mately 4% for each character. have been used. This being so, we are able to offer the fol• Having eliminated the simple substitu• lowing test, based on the fact that the com- tion cipher we will next investigate the pos• hkKd frequencies of the five most used let• sibilities of a variable substitute system. ters, ETAON, comprise approximately 45% At this point it is necesary to mention of all letters in average English text. that such ciphers are legion. Some of them, In the simple substitution cipher the five for example the Vigenere chiffre carri, the most used characters will either represent Cronsfeld, and Saint Cyr ciphers, use a the above five letters, or other letters of fixed series of alphabets, determined by a practically the same frequencies, whose short . total will thus approximate 45% of all the In others, the fixed series of alphabets is characters in the cryptogram. avoided by using a continuous and non-re• In variable substitute ciphers, however, peating key; 01, as in autokey ciphers, by where a given character can represent sev• allowing the letters of the message itself to eral different letters, and a given letter can determine the alphabets. have several substitutes, the combined fre• Again the cipher may be based on di• quencies of the five most used characters graphs, as is the , for exam• will fall below the 45% average, approach• ple, in which each letter can have five sub• ing (5 lettersX 3.85% = 19.25%) approxi• stitutes, and each substitute can represent mately 20% as a limit. any one of five letters, depending on the The following comparative table shows letter with which it is paired. the five, most used letters, ETAON, with fre• However, this multiplicity of classes and quencies taken from the table of 10,000 in types can hardly be more than mentioned FLYNN'S WEEKLY for January 23, 1926, here. For the present we must content and the five most used characters, POABK, ourselves with following up the main stream of the present cryptogram. to only one of its many branches, reserving the others for later exploration. E 12=11 P It One method of finding if our cipher uses T Qso 0 10 A 806 A 8 a fixed series of alphabets is to apply the 0 800 B 8 Kasiski test for recurrent groups, described, N 712 K 8 in detail in FLVNN'S WEEKLY for August 7. 4510 45 Should this test result negatively, some (45%) (36- SVr) other ty|>e of variable substitute would The total of 45 characters is 36.8% of probably be indicated. It is evident, how• 122, the whole number of characters in the ever, from an inspection of the subjoined cryptogram. The, nearer this total ap• table of recurrent groups in the present proaches 20% in any given cryptogram, the cryptogram, that a series of ten alphabets more probable does it become that the ci• has been used. pher is not of the simple substitution type. OL 20— 1 — 28 FB 110—50=60 The difference of 8% in this case is not JM 14— 3=11 WO loS—52=56 HO 58-11=47 PW 107—54=53 as large as might be, but, taken with what PH 33—17=18 WTP 115—55=60 follows, it is still sufficient to eliminate sim• PH 57—17=40 AT 78—60= 0 ple substitution and null ciphers. PH 57—35=22 DZ 121—71=50 VB 48—23=25 102=75=27 The above test may also be extended to UO 32 -28= 4 JF> 120—80=40 the five least used characters, if care is LV 47—30=17 YI 103—83=20 taken to assign a frequency of zero to any OH 113—33=80 characters not used in a given cryptogram. Our ' cryptogram is accordingly tran- The combined frequency in the variable scribed, as shown, in lines of ten letters, 156 FLYNN'S WEEKLY

by which arrangement all the letters in ly with OP and OH in the cipher, both any alphabet are thrown in the same col• give the key letters CU, which give good umn. While we have thus isolated the char• sequences in columns 3-4, expanding the acters enciphered in each alphabet, we still key to -CUP.AT—. know absolutely nothing of the structure of The rest of the cryptogram can be sim• these alphabets. ilarly deciphered. Thus -TILLE— in line Given sufficient material, such alphabets I is obviously ARTILLERY. And -EM- can be resolved regardless of their com• PLO— in line g is probably some form of plexity. But here we have only a few let• the word EMPLOY. In this way the whole ters from each alphabet; not enough to key, OCCUPATION, is developed. An in• solve'an unknown mixed alphabet by a gen• terlinear translation of the message is sub• eral method. An assumption as to the form joined. of the alphabets, and their relationship, is I 2 3 4 S 6 7 8 9 10 therefore necessary. 0 C C U P A T I 0 N Since ciphers of the 'Vigenere type are 0 L J M E P P R Q V to most generally known, more than likely the A R T I L L E R Y S alphabets are of that variety. If so, a solu• H 0 1 J M Z P H X z 20 tion can readily be effected by guessing at H 0 U L D B E B R 0 a word of the message. U W V B H N A U 0 L 30 u G H T I N T 0 A C The most fertile ground for speculation V U 0 P H A B along these lines is the longest recurrent H K s 40 T I to N A T T H £ V group, WTP, which more than likely sig• K L E T L U L V B F SO nifies one of the most used trigraphs. Ac• E R y B E G I N N I cording to Valerio. these are: THE, AND, B \ 0 0 W T P H 0 u 60 THA, HAT, EDT, ENT, FOR, and so on. N G 0 F T H E B A T Suppose that we assume that WTP= V R Y U C X B B A T 70 THE. T L E A N D S H 0 U In the Vigenere cipher this would require D Z B Q P Y A A T ] 80 L D B E A C T I V E the key letters DML. These are unlikely, D E I A P F K K K since they do not form part of a word. Y <» L Y E M P L 0 Y E D But they can be definitely rejected for the 0 K K G C U T g 0 A too reason that they will not decipher any other A S L 0 N G A s A N series of three letters in columns 5-6-7. K P Y I R J P w 0 F no The Vigenere square, however, can be E N E M Y K E M A I used in many ways not prescribed by its B K 0 H •w T P D G J 120 originator. Suppose, for example, in search• N S 0 N T H E F I E D z ing for a method that will produce a likely L D, sequence of key letters, that WTP and THE be taken from the sides, and the key letters Message: " Artillery should be brought into action at the very beginning of the battle from the body of the square. and should be actively employed as long as an This plan will give the key letters PAT, enemy remains on the field.'' whi; h, upon trial, produce probable se• quences in columns 5-6-7 throughout the The cipher employed in this instance was cryptogram. devised in 1857 the English admiral. Sir Further suppositions may now be made Francis Beaufort, K. C. B. It is common• as to additional letters of the key, as well ly known as the , and sev• as other letters of words of the message. eral editions of it have been published. For example, the first WTP is preceded by It was observed by August Kerckhoffs OP, and the second by OH. These are that Beaufort, in using the central square probably two-letter words, whose initial for key letters instead of cipher letters as letters are identical, as, for instance, in the Vigenere, provided an altogether dif• BE-BY, IF-IN-IS-IT, or OF-ON-OR. ferent cipher from the latter, the alphabets The words OF and ON, tried respective• being in reverse instead of norma) order. SOLVING CIPHER SECRETS 157

I'his is illustrated in the subjoined P alpha- Having thus proved that the supposedly bets of these two ciphers. irnpossible was possible after all, we sent

Text- A BCDEFGHI J KLMNOPQRSTUVWXY/, Vigenere: PQRSTUVWXYZABCiJEFGHI JKLMNO Beaujort: PONMLKJIHGFEipCBAZYXWVUTSRQ ' The cipher is thus identical in results our correspondent, Mr. McNary, the key with the United States army cipher disk, word, stating—all this occurred some time illustrated in Webster's New International ago—that before long a method of solving Dictionary. a similar cipher would be given in FLYNN'S Kerckhoffs also observed that the Beau• WEEKLY," and making so bold as to predict fort alphabets were similar in structure to that his friend, who had prepared the pres• the reciprocal substitution alphabets of the ent cryptogram, would not only be con• Porta cipher, an early system yet to be vinced that such ciphers could be solved, described here. The Beaufort P alphabet, but also that he would easily be able to in the Porta arrangement, is as follows: solve them himself. „ VWXYZABCDEFGH The article referred to was printed in the UTSRQPONMLKJI February 20, 1926, issue of this magazine. Beaufort facilitates the use of his table, And, true to our expectations, a few days shown herewith, by increasing the number later we were in receipt of the solutions of rows and columns to twenty-seven. All to the two Vigenere ciphers in that issue, four sides of the square are thus exactly along with the following letter: alike, which permits the user to enter any side he pleases, follow the row or column DEAR SIR: to the key letter, and leave the square at t am the parly who sent you through Mr. either of the adjacent sides. McNary last September a sample of a cipher a b 0 d e f g h i 2 k 1 m n 0 P q r s t u V V/ X y z a "b 0 d e f g b i 2 k 1 m n 0 P q r s t u V w X y z a b 0 a e JL g h i k 1 in n 0 P q r s t u V w X y z a b' c d e f S b i 2 k 1 m n 0 P q r s t u V w X y z a b c d e f 6 h i 2 k 1 m n 0 P q r s t u V w X y z a b c d e f S b i 2 k 1 m n 0 P q r t u V w X y z a b c d e f 8 b i 2 k 1 m n 0 P q r s t u V w X y z a b c d e f g b i k 1 m n 0 P q r s t u V w X y z a b c d e f & h i k 1 m n 0 P q r s t u w X y z a b c d e f g h i k 1 m n 0 P q r s t u V w X y z a b 0 d e f g b i 2 k 1 m n 0 P q r s t u V w X y z a b c d e f g b i i k 1 m n 0 P q r s t u V w X y z a b 0 d e f g b i 2 k 1 m n 0 P q r s t u V w X y z a b 0 d e f S h i j k 1 m n 0 P q r s t u V w X y z a b G d e f g h i d k 1 m n 0 P q= r s t u V w X y z a b 0 d e f g h i j k 1 m n 0 P q r s t u w X y z a b 0 d e J- g h i J k 1 m n 0 P r s t u w X y z a b c d e f g h i J k 1 m n 0 P q r s t u V w X y z a b c d e f g h i J k 1 m n 0 P q r s t u V V/ X y z a b c d e f g h i 2 k 1 m n 0 P q r s t u V w X y z a b 0 d e f g h i j k 1 m n 0 P q r 3 t u V w y z a b 0 d e f g h i 2 k 1 m n 0 P q r s t u V w y z a b 0 d e f g b i 2 k 1 m n 0 P q r s t u •w y z a b 0 d e f g h i J k 1 m n 0 P q r B t u V w y z a b 0 d e f g h i k 1 ra n 0 P q r s t u V w X y Z a b 0 d e f S b i k 1 m n 0 P q r s t U V w X y z a b 0 d e f g b i k 1 m n 0 P q r s t a V w X y z a b 0 d e f g b i 2 k 1 m n 0 P q r s t u V w X y z a 158 FLYNN'S WEEKLY

which I was certain was absolutely safe, and as Strong "vidence of a multiple alphabet which I had employed for years with the cipher. confidence of a backwoodsman. And now? Discov> xy of the key letters PAT, by ap• You have taught me a most valuable lesson plying 1HE to WTP, and determination of through your department, and I thank you the period by trying PAT at critical inter• profusely. vals of five and six letters, is but a matter I tind it again profitable to consider also of minutes. the other fellow's opinions. Yours very truly, Indeed, it may be stated without exag• - FRED WALTHABD. geration that a cryptogram like this can be Cincinnati, Ohio, Feb. 23, 1926. deciphered almost as quickly without the In answer to this we may say briefly key as with it. that we are always glad to be of service Thus have we outlined, as far as possible to any of our readers. in a single short article, a plan of analysis At this point we must remark that the that will enable the reader to determine the present analysis has been made complete type of an unknown cipher, and its method as it is only to familiarize the reader with of solution. a general procedure for a wide variety of And accordingly we are saying absolutely unknown ciphers. nothing about the following two crypto• To the experienced eye, the predominance grams. of seldom used letters eliminates the trans• The solver only needs to place them in position cipher at sight. his cryptographic test tubes, use the proper The lone three-letter recurrent group, cipher reagents, and observe the resultant with its sixty-letter interval, stands out to reactions. the retentive memory at the first reading These will tell the story.

CIPHER No. I. • MYAEW HNOTD ILEAN LARIE lEHMT FSVKO EAR LA ETNNE RNADX MNAKB PASRT ADOEL PEOXT ETDHN DLIEA DRUEJ OSEPH TEVRO NNUIT DSSEA

CIPHER No. 2. What are you doing to execute the instructions sent you to HCDLLVW XMWQIG KM GOEI DMVtl JN VAS DGUGUHDMITD. If success will be more certain vou can substi• tute EJTFKMPG OPGEEVT KQFARLF TAG HEEPZZU BBWYPHDN'OMOMNQQG by which you may effect O TPQGEXYK above that part HJ OPG KWMCT patrolled by the ZMGRIK GGJUL CW EWBNDLXL.

CIPHERS FROM OUR READERS Some novel ideas are offered in the fol• mentioned, be readily solved with the aid lowing selection of ciphers submitted by of the present article. readers of FLYNN'S WEEKLY. No. 3 hails from Silverton, Oregon. Look Several of these are based on the Inter• , it over, fans! national Moree telegraph alphabet. Just -which ones these are we will leave our read• CIPHER No. 3 (Rudolph O. Casperson). O I GMVOLQPWGDIMI ers to discover for themselves. However, a KTBWLPVKWPVZXJB copy of the Morse alphabet is appended Z W Z Q W V R T V V K F L R E herewith for the convenience of those not Now take a whack at No. 4, of Toronto, familiar with it. Canada, vintage. Mr. Bell writes that his 0 V . A .- H .. .. cipher " is liable to fool quite a number of B I .. P .-- w C J • — - Q —. - X - fans." How about it? D -.. K -.- R .-. Y - E . L S ... z - CIPHER No. 4 (W. R. Bell). F M -- T - A girl to see to was B slay i am No T i aRe G --. N -. U . ,- Me oO A on PYK So oN aLoe KL MEN rOT Such of these ciphers as are not based =os z WhEn ARE You e zKC went to WoRk gone is one Co? To THE dEP Ot T obri n on this alphabet should, unless otherwise gBA CKH iSp riso Ner SEN gumS s oRe. SOLVING CIPHER SECRETS 159

No. 5, submitted bj' a Brooklyn, New lished several years ago in the Waverly York, fan, is also about as good as they M agazine. come. And if you get it you're good, too! Each number of the present cipher signi• fies two different letters. The legitimate CIPHER No. 5 (Frank N. Dodd). translation, using one set of letters, forms a 9084s 26715 39485 71950 72954 79653 82870 64850 15069 30081 complete sentence. The parallel translation, 73546 77289 09276 37851 59749 using the other set, consists of all-dictionary 028,46 60107 36985 47213 words, which do not, however, form a sen• Tlie next cipher, a clever one from Tomp- tence. kinsville, New York, can be read at sight if CIPHER No. 7 (John Q. Boyer). 4-5 io-s-10 9-1-10-4 2-5-8 6-5-9 1-9 you know the system. Try to discover it. 5 3-1-10 3-1-10 9-T-9-1 2-5-4 5-10 8-1-10-4 10-1-1-8 8-5-9-1 CIPHER No, 6 (Carlton Beil). and gyp open drip an go pyt og gaps in jump Our readers who are always inquiring into a apt is not spot as pay t opyn do page about books and articles on gap in ye pyt ug spin qyp gage pig jyp go t will be glad to learn that Mr. Boyer is the any sea a set ye pyg jap a jogg. author of an article, in part about ciphers, No. 7, submitted by Mr. John Q. Boyer, in the " Key to Puzzledom," the text hook Baltimore, Maryland, a former president of of the above mentioned league. This hook the National Puzzlers' League, and a new• also contains a special article on the con• comer to these pages, requires a few words struction and solution of the National Puz• of explanation. zlers' type of cryptogram, examples of Mr. Boyer has attempted a '• double ci• which are printed in each issue of The pher," that is, one having two sets of words Enigma, the official organ of the organiza• for a solution, somewhat similar to one pub• tion.

SOLUTIONS TO PREVIOUS CIPHERS How did you make out with the Saint As wc can scarcely imagine a lime when Cyr ciphers in FLYNN'S WEEKLY for Au• there did not exist a necessity, or at least a desire, of transmitting information from one gust 7.? individual to another in such a manner as In No. I the recurrent group ZBR, at the to elude general comprehension, so we may /St and j/st letters with an interval of jo, well suppose the practice of writing in cipher and GHQ, at the 27th and 75th letters with to be of great antiquity. the interval 5/, are alone sufficient, since Now for the rest of the August 7 ciphers. they both have j as a common divisor, to No. 3 (William Bellamy) was put in cipher indicate three alphabets. by using the Vigenere table in the February The 91 letters of the cryptogram, divided 20, 1926, issue of FLYNN'S WEEKLY. The among these three alphabets, gave about 30 Beaufort table in this issue will also serve letters for each alphabet, ample to deter• the purpose. mine the key 3vord, GUN, by the method This cipher is of the autokey variety, given, thus unlocking the message: " The based on the Vigenere system. Mr. Bel• enemy losses have been heavy and they lamy appends his own explanation. have nearly exhausted their supplies of am• " Any letter of the message," writes Mr. munition and provisions." Bellamy, " when its next previous letter is No. 2, although lacking In recurrent known, will be found in the line which be• groups of more than two letters, had plenty gins with that letter, and in the column of the latter, fixing 6 as the period, or num• headed by the letter to he deciphered. It ber of alphabets. The cryptogram, consist• is necessary that some next previous letter ing of 232 letters, thus gave nearly 40 let• of the message should be known. In this ters for each alphabet, making it fairly easy instance it is the letter Q, supposed to be to get the key word, FRANCE. The text secretly communicated as beginning the in this instance was a sentence from Edgar message." Allan Poe: The message, wholly made up of short 160 FLYNN'S WEEKLY words intentionally, is a bit of advice from by Jules Verne—" is appended for illustra• one Nihilist to another: tion.

If ever we are put in , you may find Key: 20-7-18 20-7-18 20-7-. . . Message- TH 1 SCO DE . . . that you can hear me rap. But we must not . Cipher: ZA Q WW JX . . . talk in . For if wc do, be sure they will hear, and jot down aii our taps, and thus in No. 3 (Charles P. VVinsor), the free sub• time iearn to read them. But if we had kept scription cipher in for mum, we might later have been able to use FLYNN'S WEEKLY our code in case of great need. July 3,is based upon the use of the following two numerical alphabets. Mr. Winsor does Cipher No. 4 (F. Baldwin) carried the not claim to have originated this system, message: " Get him at once or he will cause having taken it from P. I. E. Valerio, a trouble for us." The cipher was of the null French writer on cryptography, who pro• variety, with the significant letters indicated posed it as an economical method of tele• by the numerical key 1-2-3-4-5, thus: graphic communication.

1234 51 2 3 - • • A 1 027 J 10 170 S iQ S13 Gee ant at brim bad at . . . B 2 054 K II 207 T 20 540 G E T H IM A T . . . C 3 081 L 12 324 U 21 567 D 4 ro8 M 13 351 V 22 594 The next one. No. 5 (Geo. A. Lauh), was E 5 135 N 14 378 W 23 621 based on the following simple substitution F 6 T62 O 15 403 X 24 648 G 7 189 P 16 432 Y 25 675 alphabet formed on the key word HAR• H 8 216 Q 17 459 Z 26 702 DEN, the name of the officer to whom the I 9 243 R 18 486 message was addressed. Mr. Winsor's message was: " If General HARDENBC7GI JK Hooker's army remains inactive you can L M O P Q S T U V W X y Z leave two brigades to watch him and with• Here H=L, L==:H; A=M, M=A, et draw with the three others. But should he cetera, the alphabet being similar to that not appear to he moving northward, I think used in the Porta cipher mentioned in this you had better withdraw this side of the article. The message conveyed to Captain mountain to-morrow night." Harden was: " Troops landing at Waianea In enciphering, the letters are taken in marching towards Kolckole pass." The pairs, as shown at (a). The values of the italicized words are Hawaiian geographical first letters of these pairs are now substi• terms. tuted from the i-to-26 alphabet, and those In cipher No. 6 (Joseph Murray), based of the second letters froih the 027-10-702 on the subjoined alphabet, the two figures alphabet, as shown at (h) and (c). representing a given letter are separated by The two numbers thus representing a a dash in the cryptogram. Thus 4-52-33- given pair of letters are next added together, 04-41—really is 45 (T), 23 (H), 30 (I), as in line (d), and finally regrouped by 44 (S). The alphabet: sixes, as at (e), completing the process.

A 10 E 20 I ,30 O 40 U 50 (a) IF GE NE RA LH 00 . - . B 11 F 21 J 31 P 41 V 51 (b) 9 7 14 18 12 15 ... C 12 G 22 K 32 Q 42 W 52- (c) 162 135 13s 027 216 40s ... D 13 H 23 L 33 R 43 X S3 M 34 S 44 Y 54 (d) 171 142 149 045 228 420 . . . N 35 T 45 Z 3S (e) 171142 149045 228420 . . .

The key to No. 7 (Austin Minette), a We expect to take up methods of solving modified Gronsfeld, was 20-7-18. This ci• ciphers based on digraphs in later issues, pher is similar to the Gronsfeld with the ex• and regret that we have not the space to ception that the key numbers are not limited offer any suggestions here. We regret also to a single digit, and the count is backward that exigencies of space necessitate the hold• in the alphabet instead of forward. A por• ing over of our list of May 22 solvers until tion of Mr. Minette's message: " This code the next installment of Solving Cipher is taken from the story, ' The Giant Raft,' Secrets. 10 F W