I I I I I 18Plll• II -IMIUEIDWtlllk 13 0 i' tll001lJISf IDWV l lBIDlill lU l l OOBtill lU f OOul e'iilWUl Bl ij Ul JlblDI t J J ij tl0is EO 3 3b 3 PL 86-36 50 USC 3605 A Aft fAf I I l Frederick w Walton Jr 1 ·TOO MANY GARBLES ••••••••••• • •• • ••••• Harry G Rosenbluh ••••••••• 3 · RE·PSYCHLING THE CODE CLERK • • • •• • •• Shirley Barton ••••• •• 4 - are DPOllMIDT S tDtltIS saa•·a•• II I ep lle• R Declassified and Approved for Release by NSA on 02-26-2021 pursuant to E O 13526 MDR-109388 EO 3 3b 3 PL 86-36 50 or 8R0RUl1 HHPBt - lf E - • - ' 1 FRIDERICK W WALTON JR A913 In recent years the Warsaw Pact has reco11- I nii ed the need forl - July 75 • CRYPTOLOG Page l TOR SECRET l'lfFA 1 use 3605 TAP lilil'iBI• WIIBIM EO 3 3b 3 PL 86-36 50 USC 3605 - July 75 •-CR P'l'OLOG Page 2 TOP SFCRW IJlfPD t iOJilllBlti•l laL T Y G S SOO MAN fARBLEft A Moral Tale for Cryptanalyst• By Harry G Roaenblu Some years ago a cryptanalyst working in the Latin American Section rejoiced A link that had fomerly carried only plaintext had suddenly popped up with a short cipher inessage They probably wouldn't use anything tricky 0 said he so I should be able to break the system on this one message He nearly did For he sake of the story let's pretend that this is the text of the one message our cryptanalyst had to work with U I PVHLV FSLGC TDQSD QXGMK I degarbled and translated the •essage as follows p t N0 50TROS NO QUEREMOS LOS 6fRODUClf S f B u O OWBJDOS A UJ'l PR EIIOS CJTAOOS EN sm d lJ P aJO S1 ee TE JIGJZM 6 A'IJ • STOP SI LA FiBJICA a e NI u a s REB tJA 'SOS PJ CIOS 1N POJO INFORMEN01 PRONTO JA f _£KDQH AMAEP LQGVH We don't want t lul products offeNJd at ths prl cee quoted in lJOUl' teZ gram of APIY g_LCQ CICSS EZOICH RCHMC yeete rday If the faotory l-or JB r-e ths pricfle a bit i nfoffll us at 01 041 ZHRVE QPAYM OZZXA HGCIO HHSQV f VCRV The four-letter hits MAEP and PLQG and the repeated trigraphs DQV and KOQ with the dis tance between them divisible by 3 gave the first clue In a relatively short time the analyst handed me a worksheet with the decrypted text and cipher alphabets UDQVI PVHLV FSLGC TDQSD QXGMK NOSO'l ROSNO QUERE NOSLO SQBOD JABDQ VTPLQ GXDQH AMAEP LQGVH UClJOS OBREB IDOSA LO'l'PR EBIOS APIYK DQLCQ CICSS EZOICH ROM CfTAD OSENS VTELD GS NA CBAn ZHR IE QPAYM OZZXA HGCIO HHSQV SSTOP SILAP ZBSIC AREBZ JADSO HNZSA 1 n' t that an awful lot of garbles for such I was recalling the passage in MC-I about a S-10 garble rate being acceptable and 21 out of 130 is more than 16 Ny friend muttered so11ething about a possibly inexperienced code clerk ••• poor transmission ••• you can't always depend upon percentages ••• 16 isn't that nich more than 10 • But I wasn't satisfied Suppose instead that the 111essage is all right but the recovery isn I t quite correct Not knowing what to suppose I started from the degarbled text and the cipher message and reconstructed the three alphabets that the encrypter 1111st have had in front of him I ended up with this chart Plain A 8 C D E F G H l J IC L M N O P Q R Cipher 1 H I l K L ti N O P Q R S T U V X 1 Z Cipher 2 0 P Q R S T U V JC Y Z A B C D E F G Cipher 3 I Z A B C D E P G H I J K L #I N O P Plain 5 TU V X Y l Cipher 1 A B C D E F G Cipher 2 H I l K L N N Cipher 3 Q R S T I V X Those tricky characters had simply slid an alphabet without a Wagainst itself usidg the word HOY today under plain A as setting Other three-letter words were used for setting subsequent messages they could easily be spotted by writing the alphabets in encrypt order Moral l'l-y to r-spJ'Oduce ths Of' Ptornatsl ia'l s thB ai plun- CUll'k 'IUlfld to '1MJ'llpt thcl 11 a short inessage HNZSA PDQCC NVQHP CDVGK LOIA£ PDQCC NVQMP CDVGK LCMAE SPSDC IOSVN POBOI NPORN ENOTP PVCRV RONTO Cipher Plain l Plain 2 Plain 3 ABCDEFGH I J ICLMNOP R TUVWXYZABCDEPGHI K Cipher Plain l Plain 2 Plain 3 STUVWXYZ LNNOPQRSTUVWXIZABC CDEFGHIJKLMNOPQRST LNNOPQRS DEPGHIJK fflBBBtZg6 UVliXIZAB July 7S CRYP'l'OL0G Page 3 CCNWDE I'li' •L MttUlf 5 III XPSPE RI I 't O 3 3b 3 EO 3 3b 6 · L 86-36 50 use 3605 EO 3 3b 3 PL 86-36 50 USC 3605 ma CFillliS iiftllR1t _ 'RE-PSYCHLING THE CODE CLfRK • iy Shirley larton Fl3 4llf 'lll e • 1 J Throughout the 1960 1 s the VietnaJDese Communists were straightforward in their cryptographic habit a I rn Id u 1 to 'hA fail' v securel I I However the VC cooe clerkS noi unnKe i netr Japanese counterparts in WWII as discussed by Mr Callimahos had a peculiarity that gave the cryptanalyst that helpful boost In 1968 the VC beRan a aradual chanae from I While there were indeed Qther features exploited by those engaged in the long-term indepth analysis of these c0111UUnications this particular one proved espe ially interesting to those of us engaged in the initial analysis of I July 75 • CRYPTOLOG • Page 4 ror sscem ua 11a t r I Non - Responsive e t