NSA on 10-17-2012 ursuantto E O 13526 MDR Case # 5477 U GJUU0U GJI1 fTlDU1V 13l B JU1UU GJ B13rnl B 00 $ iJ $ OO iJU1 1Uim $lDU1 $ 'l 2nd Issue 1991 o BaOKBREAKERS' FORUM ON MACHINE AIDS SEA HUNT o o o o o o o o o o o TEXTURE REPORTING o o o o o o o o THE TEN MOST WANTED o o o o o MILLIMETER WAVE DETECTION SYSTEMS I o 1 4 9 i 7 ON BECOMING A WELL-ROUNDED CHINESE LINGUI ST CRYSCOM REQUIREMENTS FOR UNIX-TO-UNIX ' 1r- - r 5 1 DATA ENCRYPTION-DECRYPTION EXCHANGE o ---'I--- DATA FUSION o ON THE LIGHTER SIDE o o o o FINGERTIP INTELLIGENCE o o o o o ' COMMENT ON GISTER o o o o o o o o o ' USING C FOR COMPUTE-INTENSIVE C O D E 1300K REVIEW SEIZING THE ENIGMA o Vera Filby BULLETIN BOARD o o o o o o o o o o oooo o o o ON THE ENIGMA o o o o o o o o o I 1 I3REAKING INTO OUR PAST AN ENIGMA OF ANOTHER KIND o o David Gaddy EDITORIAL LETTER o TO CONTRIBUTE o o o o o o o o o I T1Ug QOCt ll'4KHT COHTAINS CODEWORD MA J3RIAL ---- -20 22 22 V L 86-36 1 2 l 24 I 25 28 30 31 33 36 36 o 37 CLASSI I D BY SA EURSSM 123 2 DECLASSIFY ON Or 9 RatiR9 AgeAEY'S Deter iAatieA ReElblireQ NOT RELEASABLE TO CONTRACTORS DOCID 4010108 CRYPTOLOG Published by PI Techniques and Standards VOL XVIII No 2 2nd Issue 1991 1 PUBLISHER ---' BOARD OF EDITORS 1 EDITOR 963-1103 I Computer Systems 9E i3-1l03 Cryptanalysis I 96 - o 5 2 8 Cryptolinguistics I 963-4382 Information Resources 1 k963-3215 Information Science 1 k963 c 3456l Information Security 72- 23 51 Intelligence Reporting 63-5068 Language 63-ag57 Linguistics 634814 Mathematics 63 55 6 Puzzles 63 r 01 Research and Engineering 61-8362 Science and Technology 63-4958 Special Research Vera R Filby 968 7504a Classification Officer Bardolph Support Clover Support Macintosh Support ' Illustrator I I I I 963-'5463 963 3369 963-1103 961-8362 963-3360 To submit articles and letters please see inside back cover For New Subscription or Change ofAddress or Name MAll name and old and new organizations and building to Distribution CRYPTOLOG PI NORTH or via PLATFORM cryptlg @ bar1c05 via CLOVER cryptlg @ bloomfield Please DO NOT PHONE about your subscription or matters pertaining to distribution Contents ofCRYPTOLOG may not be reproduced or disseminated outside the National Security Agency without the permission ofthe Publisher Inquiries regarding reproduction and dissemination should be directed to the Editor All opinions expressed in CRYPTOLOG are those of the authors They do not represent the official views of the National Security Agency Central Security Service POR OPPIOtAL USf3 ONLJPY P L 86-36 EO 1 4 c EDOCsI D3 4010108 q 9P SEOIlEl'f' UMBRA P L 86-36 i CRYPrOLOG page 1 'f6P SECtti 1'f' UMBRA 2nd Issue 1991 P L 86-36 EO 1 4 c DOCID P L 86-36 EO 1 4 c 4010108 'fOP eM'f UMftItA 2nd Issue 1991 CRYPTOLOG page 2 'fOP SHCftH'f UMBR A P L 86-36 EO 1 4 c DOCID 4010108 'f6P S eM'f l JMftItA 2nd Issue 1991 CRYPTOLOG page 3 ' QP l l CIQ ' lJMHR DOCID 4010108 BOOKBREAKERS' FORUM ON MACHINE AIDS I I 1971 20th ANNIVERSARY 1991 Wednesday Thursday Friday 30 October 1 November 1991 Informal Seminars on Computer Aids to B'ookbreaking I Who should attend Anyone working on codes or code charts Linguists Cryptanalysts Bookbreakers Cryptolinguists Providers of computer support to ode problems Managers of code problems For more information and to get on the mailing list please fill out the coupon below or write your name organization and building on a piece of paper and send it to I IChairn1 1J 1l Bookbreakers' Forum on MachirieAids P15 P L Ops-INORTH Name _ Building- - Organization 2nd Issue 1991 CRYPTOLOG page 4 POR OPPIEURfMJ USE ONlJJPY _ 86-36 DOCID 4010108 'flOP S13EURRH'fl UMBRA SIGINT with Feeling P L 1GBP34 2nd Issue 1991 CRYPTOLOG page 5 'f6t SEeRE'f UMBlL t EO 1 4 c P L 86-36 86-36 DocrD EO 1 4 c P L 86-36 4010108 GBPoP SI3 JR I3 GBP lR IBR l 2nd Issu 991 CRYPTOLOG page 6 ' FOP g JH I3' F UMBRA DocrD 4010108 P L P L 86-36 2nd Issue 1991 CRYPI'OLOG page 7 eRE'f HA bK VY COMIWT Clll FEb8 O fbJPY 86-36 EO 1 4 c DO'C I d6 - 3 64 a 1010 8 P L 86-36 P L 86-36 - I FOYO The staff of W332 will be very happy to brief any audience The number weall is 9636388 secure ask fori I I Find the hiddenrnessagel The first five sbfutians win CRYPTOLOG mugs P1 members andCRYPTOLOG Baar 1ft1embers are not eligible far a pri e j 2nd Issue 1991 CRYPTOLOG page 8 SBeRH'f' EO 1 4 c P L 86-36 I 1I M fD L I fA 9 t n'6Q R LO OliWI EO 1 4 c P L 86-36 DOCID 4010108 L SECRET 86-36 NEW MILLIMETER WAVE DETECTION SYSTEMS L - U One of the st ongest images of the recent I R541 spur further research in this direction One of the most promising proposals in the continuing war in the Middle East was that of reporters broadcasting seemingly effortlessly from rooftops search for battlefield superiority is the use of millimeter-wave guided weapons This in Baghdad or the sands of Kuwait But this kind of mobile reporting is really not so easy wavelength region is very well suited for such a mission Metal targets stand out well from any because the equipment needed is quite heavy the antennas are about 7-8 feet in diameter natural background and the target is not and several technicians are necessary to operate obscured by dust fog rain or the smoke of the equipment This may change quite rapidly burning oil wells The small antenna would fit into a modest missile of only 3-6 inches in in the future however as the push to higher frequencies in the millimeter range convendiameter yet it would provide both sufficient gain and small enough beam width to readily tionally 30-300 GHz will make truly suitcasesize earth terminals possible With antennas as find tank-size targets Together with some modest processing capability such as fire and small as a foot or so in diameter and with forget warheads delivered from long range enough bandwidth to accommodate all the could provide an enormous advantage on the reporters of a future operation a truly large future battlefield numbern such millimeter wave systems will really wire the global village The military is already studying the use of this wavelength region for communication satellites which would use much smaller much less complex ground sites yet provide an enormous increase in channel capacity Can the commercial world be far behind U Another image left with us from DESERT STORM was of smart bombs that homed in on their targets with surprising accuracy Although these weapons were hardly news to even the most casual reader of Aviation Week their actual successful use in combat will surely 2nd Issue 1991 CRYPTOLOG page 9 SSCRE't' IIAf DLEl VIA COMH f C1IAfU'E L f l DOCID 4010108 SEEURRECf' 2nd Issue 1991 CRrt'TULUli page 10 SECRE'f i UAUDLEl VIA COMHFf' CHA UH3L8 ON 6 EO 1 4 c F L 86-36 DOCID 4010108 2nd Issue 1991 CRYPTOLOG page 11 'M KMf L SISCRET VIA COM'Hff CIIAfUUlLS Of LY DOCID 4010108 SE6RE'I' Figure 3 shows a millimeter quasi-optic configuration that will be used in this new design The detector is of a planar design which also incorporates an antenna placed at the focus of the quasi-optic system There are several candidates for the detectors two will be discussed in detail superconducting tunnel junctions and Si Ge superlattices U r I I 1--c 1cm p L 86-36 _f I -RF EO 1 4 c --- 1 V 818 in Planar Antenna inl -t----t--il L Quartz Hyperhemisphere Teflon Lens Figure 3 QuasI-optical system Detectors L I CU The superconducting detector depends on the quantum mechanical tunneling between two Quasi-optics superconductors This effect is best understood from studies of tunnel junctions consisting of U Quasi-optical concepts treat millimeter radiation as one would optical radiation in that two thin superconducting films separated by a several-nanometer-thick tunnel barrier made lenses and other focusing elements are used to concentrate the energy to be detected The main from the oxide of one of the metals or from some other insulating material advantage over standard parabolic reflectors is that the system can be designed so that an U Superconducting receivers for infrared and array of detectors can be in focus and at the millimeter waves depend on the nonlinear same time allow a staring nonscanning system response of the tunneling currents when they to cover a whole sector without any need for are driven by constant and RF voltages In fact mechanical movement To take full advantage it has been said that this current-voltage of this new capability implies that a detector nonlinearity is the strongest known in physics array is available that has dimensions no larger than the wavelength of the radiation In the U Figure 4 shows an experimental I-V curve millimeter wave region this means that the for a tunnel junction The time-dependent detectors must be made by some photolithographical process 2nd Issue 1991 CRYPTOLOG page 12 SECRE't' U diDLEl YUt OOMH1'l' OIhY1USLS O lLY DOCID 4010108 SECRE'I' 60f- - 3C 40f- - involve photon assisted tunneling leading to an enhancement or reduction of quasi-particle current on a voltage-energy scale of the millimeter wave photon U It is the nonlinear tunnel junction I V characteristics that are to be used in this new millimeter wave detector system These junctions are made by photolithographic techniques together with any tuning structures needed to match impedances Arrays of junctions can easily be fit into the quasi-optic concept - 20 ---'-- CD 01------ '-1-----1 o 2 4 Bias Voltage mV FIgure 4 I-V curves Tor SIS current I t resulting from a voltage V across a tunnel junction can be written dV I t Ie sin tP V G V G' V cos 4 C dt The first term on the right is the lossless Josephson-pair tunnel current which depends on the phase difference I of the wave function of the two superconductors The second term VG ' V is the product of a voltage and a nonlinear coductance called the quasi-particle term The third term represents an interaction between the first two k CU Most infrared and millimeter wave devices can be classified as either Josephson effect or quasi-particle devices depending on which nonlinearity is the important effect The nonlinearity in quasi-particle conductance G V arises from the energy gap of width 2 in the density of excited single particle quasi-particle states In an ideal junction at T O K quasiparticle current can only flow when a bias voltage V 2 fe is applied Because of the singularities in densitie's of states at the gap edges the onset of tunneling current at 2 e has infinite slope in superconductivity theory U Figure 4 shows the discontinuity at 2 mv bias voltage Though the slope is very steep there is always some rounding of the corner at 2 and some leakage current is apparent at lower voltages The other curves in Figure 4 2nd Issue 1991 Mixers U The junction I V due to quasi-particles can be used in several ways to make a broadband receiver system The nonlinearity can be used to make a quasi-particle heterodyne mixer which converts signals from some high RF down to a lower frequency where amplification and signal detection are more convenient When strongly pumped by a local oscillator at WLO the local oscillator frequency the mixer produces a linear response at the intermediate f equency WIF when a small signal is supplied at the signal frequency ws -mwLo wIF CU The full quantum theory is required to describe mixing in Superconductor-InsulatorSupercondueting SIS junctions whose I-V curves are sharp on the voltage scale hwfe When quantum effects are important the theory predicts net mixer gain and very low noise and because of arguments related to the Heisenburg uncertainty principle there is a minimum zeropoint noise power for mixers of hwB SIS mixers have been built at 33 GHz and at 90 GHz which have measured noise temperatures within a factor of two of this quantum limit It should be pointed out here that the requirement of small current flow at voltages below 2AJe sets an upper limit of about Tc 2 on the operating temperatures that can be used for SIS mixers where Tc is the critical temperature CRYPTOLOG ' page 13 SECRET HANDLB '1M COMHi' ' OIIAtiNEJLS OHLY DOCID 4010108 EGRE'J' Operating temperatures that can be used for SIS mixers where Tc is the critical temperature where superconductivity vanishes This cryogenic problem has been eased with the U The quantity in square brackets above is recent availability of NbN junctions with the second difference of the unpumped I-V curve T c 15'K At present there is no appropriate for the three points V Vo and V Vo - hw e high Tc SIS junction technology divided by the first difference computed at V Vo - hw e In the classical limit where the U Thin film SIS tunnel junctions can be current changes slowly on a scale of hwle the integrated in a planar configuration with logdifferential approximation gives the usual result periodic or log-spiral antennas which have for a diode detector central lobes and are self-complementary meaning they have a real impedance of 120 Q These antennas shown in Figure 5 fit nicely into the quasi-optical concept of Figure 3 Figure 5 Wide Band Antennas Self complementary planar antenna Real anterma impedance 100 lm Broadband response Single mode throughput An 2 If the I-V curve is sharp enough that the current rise at 2A occurs within the voltage scale hw e and if the bias voltage Vo is just below 2A e then SI e hw This quantum limit to the responsivity corresponds to one extra tunneling electron for each coupled photon D Since direct detectors do not preserve phase there is no quantum limit to the detector noise analogous to that of the mixer The intrinsic noise of the SIS direct detector is simply the shot noise in the dark current I at the bias point Yo For a postdetection bandwidth B the noise equivalent power NEP in units of W Hz1l2 of a RF-matched detector is I Dg f'efiodic Antenna Direct Detectors D When the signal power is increased the responsivity of an SIS detector falls however the power required to saturate the device is much greater than that of a single junction mixer 2nd Issue 1991 CRYPTOLOG page 14 gA DLE SBORB'f' VII COMfP'IT CH-A IP'IEL8 ONLY DOCID 4010108 SBEURRE SI of 3600 amperes per watt was within a factor of 2 of the quantum limited value e hw at 36 GHz These results have been repeated at much higher frequencies recently The NEP measured at 36 GHz was 2 6 x 10-16 W Hz 1l2 which is essentially the performance of the best millimeter wave astronomical instruments U Little progress in the sensitivity of SIS direct detectors has been made since the first experiments However the limit set by shot noise in the junction current is not fundamental The fundamental thermally activated leakage current is of order I-exp -MkT which is many orders of magnitude smaller than the extrinsic leakage current that is observed in all SIS junctions Progress has been slow because relatively few applications of SIS junctions depend in a critical way on leakage current This situation is now changing The promise of X-ray detection concepts based on quasi-particle trapping and subsequent tunneling has greatly increased recent activity Several well funded groups in the US and in Europe have announced their intentions of producing junctions with leakage currents reduced by orders of magnitude 1co Ai U1 A a b c c Figure 6 SIS Junction an needed for coupling This is also shown in Figure 6 along with the equivalent circuit D The full power of microstrip technology can be applied to the coupling of SIS direct detectors for millimeter wavelengths The loss Integration of Direct Detectors and Antennas in superconducting films is negligible for films at temperatures well below Tc therefore D The SIS direct detector is a fast very broadband RF detector and is the most sensitive lossless broad band couplers band and low pass filters and even multiplexors can easily be detector at millimeter wavelengths for implemented The possibility also exists to temperatures 10 K As a planar lithographed deposit several direct detectors on a single structure it is compatible with planar antenna and to couple them via multiplexors lithographed antennas coupling structures filters etc Such elements can be fabricated into and band pass filters A frequency multiplexed spatial array of detectors can be produced in a focal plane arrays by conventional optical way that can meet all the objectives of a search lithography This includes the selfsystem complementary structures discussed above such as log-periodic and log-spiral which have no Novel radiometer for millimeter waves characteristic length dimension and therefore have real terminal impedances and extremely D A novel radiometer configuration suggested wide bandwidths The impedance of the SIS by Prof Richards uses one SIS junction pumped detector junction however has both real and with a local oscillator as a heterodyne down arts so that a matchin ssue 1991 CRYPTOLOG page 15 SECRET IWfDLE 'IfA OOMH'lCP OH dHH3LD OnLY DOCID 4010108 SECREi' ret 90 r--- - C T' r TTTT'- rrrTTm---r-rTTrrm 80 connect the mixer to the detector The overall system is being assembled for a proof of concept 70 FIELDABLE SYSTEMS 60 50 40 30 20 t g 10 0 -10 -20 uJ -30 -40 -so -60 -70 -80 -90 L --l L- _ l -l -l-l U --L JJ- 1 Figure 7 S y noise vs e evation ang e U With the very low noise wideband receivers available from SIS technology 1N 4 K there are several options as to how an actual fieldable system architecture would look For example at W band 75-110 GHz one can see from the data in Figure 2 that present mixer noise alone leads to a temperature of 9000 0 K 15 db 90000 K 'If the system is constrained to look at the earth at T 300o K then the increase in bandwidth for the same sensitivity would be 9000 300 30 converter followed by a second SIS junction which is used as a millimeter wave photon detector From one viewpoint the mixer is a preamplifier for the photon detector It amplifies the photon arrival rate by the product of its power gain and the frequency downconversion ratio The SIS photon detector is used at the relatively low IF frequency where its NEP is better than at the RF frequency From another viewpoint this system is a simplified version of a heterodyne radiometer that avoids the need for the IF amplifier because of the conversion gain of the SIS mixer and the excellent performance of the SIS direct detector This configuration appears to have higher sensitivity than the SIS direct detector while retaining its high speed and operating temperature It appears to be simple enough that planar integrated arrays or detectors are possible Prof Richards may apply for a patent for this novel receiver design that uses the strengths of each part of a tunneling junction operations U Preliminary steps have been taken to test this receiver concept The antenna coupled SIS mixer and the SIS direct detector have been fabricated and separately tested The new element which has been successfully tested is an efficient low pass microstrip filter required to EO 1 4 c P L 86-36 2nd Issue 1991 CRYPTOLOG page 16 SECRE'I' lIAlfBLE VIA eOMft tp CIlAtHU3LS et LY EO 1 4 c P L 86-36 DOCID 4010108 'FOPSI3CR'E'f U One of my lifetime goals is to feel comfortable with the Chinese language and to understand Chinese culture and society The trouble is that Chinese is difficult to learn and nobody knows that better than a student ofChinese even after 30 years With that in mind I'd like to present my views on what becoming a well-rounded Chinese linguist means in the SIGINT environment and to offer some suggestions on how to become one U In these days of high technology people in all walks of life necessarily have become increasingly specialized Doctors specialize in various kinds of ailments lawyers specialize in various kinds of law mechanics specialize in different makes of car At NSA we are no less specialized We have specialists in computers traffic analysis collection and language analysis just to name a few fields Furthermore our specialization extends beyond these major categories for the SIGINT linguist the result can be using language that is very narrowly focused SAerstend to be so specialized that we are encouraged to diversify in order to broaden our experience Butin the language career field one can specialize in graphic or voice I did not support the distinction because I believe that a majorgoal ofthe language certification process should be to identifY all-round good linguists encompassing both graphic and voice but I will concede that it works well for individual career advancement identifying linguists whose profeSSIonalism is somewhat more focused If you have been around since the Chinese PQE first began you will remember that in the early days aspirants were required to demonstrate expertise in graphic AND aural ChiC nese 0 000 To have passed one of the early PQEs was a major accomplishment and those who did genuinely could be considered to be good all-round linguists not only in a general sense but also as SIGINT linguists Furthermore they reasonably could be expected to be capable of handling any language task with a minimum of training They certainly met the standard ofthe day which went something like Professional Chinese linguists should be able to handle any language task which theymight be called upon in the middle of the night to perform without assistance and produce P L 86-36 finished material upon which an end-product report reliably could be based ON THE TESTING PROCESS 0 eeO The PQE testing process has evolved from identifying all-round professional linguists to 2nd Issue 1991 CRYPI'OLOG page 17 'f'8P SElBB8'1' l l E JPYIt EO 1 4 c BQl IJWI' G1L NNElI JS 9Nl1JPY 8 6- 3 6 1 4 c DOCID L 40 td1oB e classroom walls I like to think of linguists who have that desire as purists to some extent we learn because it's there I don't mean to imply that we consider ourselves scholars although I'm sure there are some among us who fit into that ategory I' U What I'm really driving at is that learning L FOUO The point is that the certification process in my opinion identifies and fosters well-rounded linguists only in the sense that they are wellrounded with regard to related-fields and SIGINT disciplines not in the sense that their specific language expertise is broadly based THE WELL-ROUNDED LINGUIST U Now my idea of an well-rounded linguist doesn't mean that you have to be highly proficient in every aspect of the language although that's certainly a noble goal to shoot for I do believe however that well-rounded Chinese linguists should have certain basic skills and those skills are the same for virtually any language either native or acquired Namely they are the abilities to read write and speak the language to a sufficient degree that they can function well in-country Chinese has to become an avocation and wanting to become better at it is just natural However as natural as that desire may be it also takes a fair amount of work 'fS CCO Since I'm just a student of Chinese and not a true scholar I want to share a few of my ideas which upon reflection helped me in the hope that others might find them useful Written Chinese U Let's start with written Chinese I can't over-emphasize how important learning Chinese characters and reading are The more characters you know the easier it is to learn new characters and new character compounds thereby increasing yourreading and oral vocabulary Remembering a word or phrase is easier if you can visualize the characters for it So firstly you should take as many formal reading courses as you can 'ffl CCO So how does one go about becoming an but don't stop there Subscribe to-at least well-rounded Chinese linguist There is no single scroung rigina11anguagemagazines and or roadmap on how to do it but if you haven't alnewspapers and read read read Take the time ready guessed I have some personal opinions now and again to mark characters which you based upon my own experience and those of my don't know in at least one short article look them colleagues up in a dictionary make and study flashcards I'm a staunch advocate of flashcards Translate U First and foremost of course is education a short passage occasionally to keep your skills That doesn't necessarily mean having a college sharp especially if you don't get that kind of degree but it certainly means having a solid practice on the job As your vocabulary grows background in formal Chinese language training the reading gets easier and more enjoyable for Many of us acquired this training through intenyou are relying less on the dictionary Also sive instruction at the Yale University Institute of frequent reading in Chinese helps keep you Far Eastern Languages and later the former abreast of what's happening in Chinese society American Embassy School of Chinese Language and Area Studies in Taichung Taiwan U As to learning cursive script take a basic course learn to write some of the cursive forms U Secondly and perhaps equally important is and whenever you have some time practice using that one just has to love the language and culture them until they become second nature you don't There has to be a desire to go beyond the minihave to become extremely proficient to make it mum necessary to perform the job and this I useful When you doodle doodle on your note believe is the key to success in learning beyond pad in Chinese using what you have learned 1st Issue 1991 o CRYPTOLOG o page 18 'fIep St CRH'i' IIM'F BM lffA OOMH Onl'JdtfJfELS OPHrY 4010108 'l'8PSBOBilil'l' Think of word compounds and write the characters a few times just to improve your handwriting and for retention If you can't recall a character stop and look it up Read as much handwritten material as you can get your hands on until you feel fairly proficient at it Once you are proficient though you'll get rusty if you don't use it often you will be able to pick it up fairly quickly when the need arises come familiar enough with the STC book so that you can at least tum to or get close to the correct page for a given radical when you want to look up a character Even if you don't learn the exact STC groups for lots of characters simply knowing how to find them rapidly may help you either on the PQE in solving garbles in STC traffic and other operational situations where STC is required Aural Comprehension CU For aural comprehension listen to recorded materials from the rariio or videotape the Chinese language news from Taiwan off the PBS television station in Washington if you can Take every opportunity to speak Chinese in restaurants or wherever else you find someone with whom to talk-being ever mindful of security considerations of course If you have a good ear take a tour in transcription to improve your aural comprehension and expand your general knowledge MUSINGS U On a more general note pick up a ChineseEnglish dictionary from time to time and just browse through it You'll be surprised sometimes to find the characters for word compounds you know how to say but not necessarily write not to mention revisiting compounds long since learned and forgotten Moreover you can learn some interesting things which you might not otherwise learn from newspapers and other media U Now I know that family friends outside interests and commitments all compete for the same free time needed for language study such as I have described Nevertheless I've found that if you're serious you can find some time in a busy schedule even if it's not daily I sometimes think of what my calligraphy teacher in Taiwan once said about practicing writing Chinese characters with a brush pen When I reached a certain point he told me that I no longer needed his instruction and that if I would just take 15 minutes to practice each day I would do just fine D Sixteen years later I often wish I had taken his advice Instead I let something or other interfere with my practice sessions so my calligraphy never reached the point I had hoped it would Recently it dawned on me that if in all these years which I have been associated with Chinese language I had learned just one new character per day even with interruptions from other activities I would know all the characters I would ever need to know w OOO Will the effort to go an extra mile to learn through self-study ever payoff or will it simply be an academic exercise I believe it does payoff It goes without saying that you will derive self-satisfaction as your skill with the language expands but I believe you will find that by increasing your overall language capability your efficiency on the job will improve as well You'll spend less time consulting dictionaries increase your ability to scan for significant items and be able to handle a wider variety of subject matter CPS OeO Moreover you never know what opportunity may present itself in the future if you have the proper skills 1 ---'IYou will look back and thank your lucky stars that you expended extra effort over the years to round out your language and cultural training and experience and take pride in the accomplishments derived from your efforts to become a well-rounded Chineselinguist L - o 1st Issue 1991 o CRYPTOLOG o page 19 ' PSEOBilil'l' IIMfBbB V t't OOMI t'f 0lIMtHl'Jb8 O H JPY EO 1 4 c P L 86-36 OCIO 4010108 ------------seEURRt l'f CRYSCOM Requirements for Data Encryption-Decryption Exchange 13iCRYSCOMExec p o 2nd Issue 1991 o CRYPTOLOG page 20 SJ l G ItJd'ffiLH TflA EUR9MIN't' eHMfNI3LS eNfN L 86- 3 6 P L 86-36 EO 1 4 c P L 86-36 DOCID 4010108 2nd Issue 1991 CRYPTOLOG page 21 P L 86-36 EO 1 4 c BEORH't' If ANBl FJ tJI C IDJ T C lai y 4010108 Si1GH'I'SPOKi3 On the Liehter Side RANK HATH ITS PRIVILEGES 1 4 c L 86-36 Contr ibuted byl ---- Computer and Information Sciences Institute Special Interest Group on If you use data from multiple sources or support those who do to track identify or evaluate Fdmay benefit from exchangingideas and software targets you are doing I MONTHLY MEETINGS P L Wednesday 25 September 1991 0930 9A135 86-36 Subject Geographic Information Systems Speaker I J 51 Thursday 31 October 1991 0930 9A135 Subject New Multi-media technologies for information dissemination SpeakerlL - ---Jlp05 PO OffICI cL ISE eUlY 2nd Issue 1991 CRYPTOLOG page 22 SEGRETSPOD P L 4010108 86-36 D In 1987 W2 set out to expand access to information in order to increase productivity and to improve the quality of its products We were hoping at the same time to make the production process more enjoyable for the analysts by giving them graphical user interface GUI and personal computer networking technology Both are now an integral part of our systems and applications FOUO We have made progress towards the goal of reaching out with our fingertips for intelligence information via computer keyboards we call this fingertip intelligence This means making computers the indispensable tool that people instinctively reach for when they need information or want to share it The process cal1s for o a graphical user interface via windows software that provides a what-you-see-is-whatyou-get known as whiz-e-wig from the initials view on the monitor o running multiple applications word processing spreadsheet graphic editing etc in multiple windows with some background processing o scanning in graphics and images to use in publications or in vuegraphs and slides or to pass on to others over the network reported created in the Word for Windows word processor And we don't have to close anything down to switch applications FOUO That all these applications can talk to one another has proven the value of our set-up For example the new data on target performance we have just entered on the spreadsheet is automatically updated in the report sitting in the word processor Keeping the documentation in sync that is not having to enter newer data manually everywhere it appears save time and assures accuracy This is what analysts like U This process is not limited to the confines of r--------------------- W2 but through networking it can reach the other side of the world We are now working to integrate intel1igence information and to autoIllateprocedures to manipulate assemble and distribute it Wear alsQJrying to improve elec P L 86- 3 6 tronic mail o running analytic processes on minis and mainframes remotely o using an ever-growing number of applications The possi ilities are astounding CD We are taking advantage of the better and faster hardware and software that the computer industry is developing and also the more appealing features so that now people are using computers not because they have to but because they want to For example we have been successful in retrieving data from the VAX then running an analytic program and outputting the results using Corel Draw graphics and then importing it all into PowerPoint presentation software or into an Excel spreadsheet and inserting the results in a FOUO It is new technology that has made this possible-the faster machines withmore storage and computational capabilities Now we are poised to bring all the pietes together on our DNIX-bas ' yations as part of the W2 projec t will incorporate the features described above a boon to the analysts But no one division or office can implement this vision alone we are working diligently with other people in the Agency and outside too CD If you'd like to pay us a visit caUI l W24 on 968-8963 or me on 968-4236 We'll schedule a demo for you in our dynamic and everchanging environment CRYPTOLOG page 23 FGR GFFI JIA I J USB ONl iJPY 2nd Issue 1991 o EO 1 4 c 4010108 Comment on P L 86-36 GISTER 86-36 U The spreadsheets that can perform these tasks are Lotus 1-2-3 Release 2 2 for IBM machines Lotus 1-2-3 Release 3 1 for IBM and Unix machines Wingz and Excel for Macintosh machines There are ways to enter the GISTER text files from the DOS machines to the Macintosh machines to use these spreadsheets I would be glad to give a demonstration of the technique to anyone interested U I am happy thatl linguist has written such a useful and impressive program on his own which could have cost the Agency a tidy sum were it contracted out P U It is also nice to have a publication like CRYPTOLOG handy that allows feedback on the articles written I would never have known about Jim's work because I am not involved in language problems and my technique mentioned above if useful to anyone would not have been published Update on GISTER P L 86-36 EO 1 4 c 2nd Issue 1991 CRYPTOLOG page 24 JRti'F IIMH3L V A JOMUT'F JIIMR'T LS OUhY P L 86-36 86-36 DOCID 4010108 F 3 ' - - - - - - - - - P L 86-36 c offers advantages besides allowing a programmer the use ofthe magic phase object oriented It encourages writing code that is easier to read and easier to maintain This article describes a short experiment to measure the run-time penalty paid when using C specifically to write readable code c allows the programmer to define data types and operations on these data types This can be used to allow the programmer's code to more directly represent the algorithm I have been looking at some linear discriminant analysis software and decided to use it as a test The first step of this software creates a covariance matrix from known data The experiment described here compares the run times of this first step using Microsoft C version 5 1 as the standard C This result was compared to a c version compiled using Turbo C version 1 0 and Zortech c version 2 1 An early experiment was done using Rogue Wave's matrix classes for C This software was discarded because the increase in run time was incredible and also because the package would not allow the operation matrix vectorT vector The Rogue Wave package is written in the same style as the NIH classes This style has classes inherit properties of more basic classes which inherit properties of still more basic classes This approach has advantages but it causes the software to run slower because of a lot of nested subroutine calls Turbo C and Zortech C both refuse to inline code with loops and most of the matrix code has loops I wrote a simple matrix class which is used in the results below The test code reads 500 data vectors from a file each 90 elements long These vectors are used one at a time to build the covariance matrix Most of the time in the entire test program is consumed by the operation covariance_matrix covariance matrix data_vectorT data_vector Two C versions were created One uses code that directly follows this expression the other uses a subroutine Except for this one line of code the two C version are the same The straight C version uses a subroutine The run times were as follows times are in minutes seconds Type of code used Microsoft C Turbo C code follows expression code uses subroutine N A 0 18 1 22 0 18 Zortech C 1 06 0 18 2nd Issue 1991 o CRYPTOLOG o page 25 P6R 6PPfCtAL USE 6NLY DOCID 4010108 The times are good only to the nearest second or two The table shows that to achieve the run time of the straight C version the time-intensive calculation needs to be in a separate subroutine The results are the kind that make everybody happy If you don't want to change to C they show that C has terrible run times If you want to use C they show that c doesn't have to be slower than straight C Here is the critical line of code used in each of the three versions Straight C ddiadic quapar vector vector len len len C fast quapar covar_calc rowvector C slow quapo r rowvector transpose product rowvector Unfortunately the easiest to read code is also the slowest C attaches the data to the code that processes it For example it is very easy to give each matrix a name so that the code can report not only a error but the matrices involved This means that if you tried to add two matrices of different sizes together the code would not just report Error matrices are different sizes but could also tell you which matrices are involved This makes for very friendly code The straight C code used for the test follows the traditional math programming style and doesn't check for any errors WHY TIlE C CODE IS SLOWER I believe that the creation and destruction of scratch variables is the major reason that the C code runs slower Look again at the critical calculation The C expression causes the following steps to be done 1 Create a data vector and put data_vectorT into it 2 Create a matrix and put data_vectorT data_vector into it 3 Destroy data vector containing data_vectorT 4 Add matrix from step #3 to covariance_matrix 5 Destroy matrix containing data_vectorT data_vector Each time the line is executed two intermediate values are created used and destroyed By replacing the line with a call to a subroutine the need for the intermediate values is eliminated It would be interesting to see how much time could be saved by using some sort of cache for intermediate data elements That is instead of destroying a intermediate data element put it aside in case you will need it the next time through the loop CONCLUSION USE C t The two major steps of real time programming are first make the code run right and only then make it run fast C has significant advantages for writing code that runs right and as this experiment shows C can run fast 2nd Issue 1991 CRYPTOLOG page 26 POR OPPIOtAiJ USB ONLJPY aCID 4010108 WHY IS ZORTECH C - - FASTER THAN mE TURBO Ctt The test code had a verbose option that would print out each time a data element was created or destroyed For example the critical calculation caused the following lines to be printed when the Zortech Ct version was run column_vector rowvector transpose 90 1 created matrix rowvector transpose matmul rowvector 90 90 created matrix rowvector transpose matmul rowvector deleting data destruyed column_vector rowvector transpose deleting data destroyed The format of the line is type of matrix the matrix's name in parenthesis the size if created arid a comment saying created or deleted An intermediate data element takes the name of the original element with the name ofthe operation added For example the line column_vector rowvector transpose 90 1 created says that a column vector with name rowvector transpose with size 90 1 was created Here is the corresponding output when the Turbo C version is run column_vector rowvector transpose 90 1 created column_vector Copy of rowvector transpose 90 1 created column_vector rowvector transpose deleting data destroyed matrix Copy of rowvector transpose matmul rowvector 90 90 created matrix Copy of Copy of rowvector transpose matmul rowvector 90 90 created matrix Copy of rowvector transpose matmul rowvector deleting data destroyed matrix Copy of Copy of rowvector transpose matmul rowvector deleting data destroyed column_vector Copy of rowvector transpose deleting data destroyed row_vector Copy of Row 20 of means deleting data destroyed Notice all the Copy of lines generated Turbo C - is creating copies of matrices that the Zortech code didn't need Why are all these extra copies made The difference between the two is caused by how the compiler handles return values For example look at this subroutine Matrix dummy void Matrix X X o return X II X is an automatic variable II Put a value in X I I Return X The above subroutine returns a variable X that is automatic and goes out of scope when the subroutine returns Turbo C creates a copy of X and destroys the original when the subroutine returns Zortech C postpones the dastruction of X until the subroutine's return value is no longer needed At least one of the reasons that Turbo C runs slower is because of all the extra copies it makes 2nd Issue 1991 CRYPTOLOG page 27 FOR OFFIOJIkL US13 ONLY DOCID 4010108 20 reached through correspondence - German Polish French British and American They included cryptanalysts U-boat and ship captains and crew members intelligence officers radiomen and others Seizing the Enigma The Race to Break the German U-Boat Codes 1939-1943 David Kahn Houghton Miffiin Company Boston 1991 Reviewed by Vera Filby D9 Many NSA people have read or know of David Kahn's first book The Codebreakers a massive and erudite world history of cryptology With this background of knowledge an Oxford doctorate in history and the acquaintance of many who remember the events recounted Dr Kahn was well equipped to tell the story of the German naval Enigma cipher machine and the struggle to tum it into a weapon against the U-boats Readers who have followed the cryptologic history of the Battle of the Atlantic will find much in this book that is familiar but they will enjoy it all the more for that reason and will appreciate it for its new information its entertaining style and its novel approach The theme is the recovery of Enigma materials from U-boats through circumstance of war and from ships by chance or by planned seizure Information encrypted in naval Enigma was desperately needed in Britain's deadly battle against the U-boats and the struggle to break the system was long and hard Veterans of that effort were among the more than 70 people interviewed and Among them were Gustave Bertrand wartime head of French espionage Norman Denning head of German intelligence in the Operations Intelligence Centre and Harry Hinsley who as a Cambridge history student joined the G C C s now GCHQ at Bletchley Park taught himself traffic analysis and soon demonstrated its operational intelligence value He became a naval intelligence analyst and many years later was chief author of the UK Government's official British Intelligence in the Second World War Dr Kahn is reputed to be a skillful and persistent interviewer with a knack of asking questions to which he already knows the answer or part of it perhaps in the hope of confirming it or eliciting more information Other sources besides many of the books and articles published on the subjects covered included ships' logs reports memoranda and personal papers in German British and American libraries and archives Dr Kahn's accounts ofthe ship captures are full of color excitement and suspense He is a great teller of sea stories The parallel story of the Enigma machine and its ramifications he treats with thoroughness and clarity beginning with the day 15 April 1918 when the inventor Arthur Scherbius offered his cipher machine to the Imperial German Navy up to the end when the Enigma decrypts were no longer needed He recounts vividly the Polish French British and American experience with special attention to the Polish story He describes in detail the construction and workings of the machine the means and methods of atttack and associated codes and ciphers with the focus always on the naval Enigma Luftwaffe Enigma was first broken on 22 May 1940 and thereafter air traffic was read regularly But the naval remained resistant Naval Enigma was more complex and naval communications more disciplined and only limited and temporary successes were achieved until December 1942 2nd Issue 1991 CRYPTOLOG page 28 F6ft 6FFleIXL tJ'SI ONLY DOCID 4010108 With the Enigma buttoned up the analysts at Bletchley worked on and broke lower level systems One of these was the Dockyard Cipher which was used for messages to and from shore stations and patrol boats minesweepers auxiliaries and other vessels In time the analysts discovered that messages sent in Dockyard were sometimes sent also in Enigma to ships equipped with Enigma thus producing cribs which they called kisses GIFTS FROM THE SEA A bit of help for the analysts in Bletchley's Hut 8 the naval cryptanalysis section came from an unexpected source early in 1940 when the minesweeper H M S Gleaner encountered and attacked the German U-boat U-33 at the entrance to the Firth of Clyde 12 February The skipper ofthe damaged submarine decided to surface and make a run for it Enigma parts were distributed among crew members to be dropped into the sea ifthey had to abandon the submarine The Gleaner spotted the U-33 and continued the attack In the danger and confusion one seaman forgot to sink his piece-a wired rotor On 26 April a British destroyer returning from Norway stopped and boarded a trawler identified as Dutch and named Polares A crew member threw two canvas bags overboard but the boarding party was able to capture one ofthem When the contents arrived at Bletchley they were found to contain Enigma keys for four days in April Nearly a year passed before the struggling cryptanalysts received another gift from the sea in the form of key tables for February In March 1941 a destroyer accompanying troop ships carrying commandos for a raid on German-occupied Norway encountered an armed whaling trawler the Krebs in the Lofoten Islands Documents retrieved from the ship after it was fired on and destroyed included gridded charts the February Enigma keys rotor settings and the plugboard setting With their help the Hut 8 analysts decrypted messages which included weather reports and information on the movements of weather ships to and from fixed areas where they were stationed for periods of several months The analysts were soon getting kisses from weather messages WEATHER Germany could not do without weather information from northern waters It was essential for current data and forecasts for all purposes but especially to support air operations against England Only ships at sea could provide a steady flow of observations so fishing vessels were used to serve as weather stations Even U-boats were pressed into service To deny the data to the British reports had to be encrypted and the system used to do it was the Enigma Weatherbirds will appreciate the significance of this Not many NSAers know about the Weatherbirds-an informal club of SIGINT weather analysts Weather is usually a low priority target for SIGINT-until there is a crisis then everybody wants the weather support SIGINT weather can provide identifications and locations contribute to traffic analysis give tipoff's to related activity reveal information about the entities it serves reflect such events as airfield openings and closings and indicate future events In Seizing the Enigma Weatherbirds will learn how the Enigma encipherment of weather led to its own undoing Losses of ships carrying war materiel food and other necessities of survival were increasing and help from Bletchley was urgently needed Harry Hinsley studying the weather messages along with his other traffic had a stunning idea Why not go after a weather ship His suggestion went through channels and the Muenchen which was about to sail to a grid square northeast of Iceland was selected A task offorce of destroyers located the ship on 7 May and shelled it Before abandoning ship the radioman who had been transmitting a weather report gathered up the Enigma machine the current keys and other materials put them into a lead-weighted bag and threw them overboard But the boarding party did succeed in collecting some documents Among them Bletchley identified the Short Weather Cipher Enigma settings and the June keys The Short Weather system reduced synoptic weather ele- 2nd Issue 1991 CRYPTOWG page 29 6ft 6 PtCfA L USB efftN DOCID 4010108 ' - ments to single letters which were then enciphered in Enigma and could be transmitted in 15-20 seconds naval traffic thioug h6ut 1942 Naval Cypher No 5 replaced Nos 3 and -lin June 1943 arid from that pointon Brltishnaval was denied to them Meanwhile ihe British lost the Enigma when the Recoveries from the U-llO captured only two days M4 model wils introduced on 1 Februry 1942 after the seizure of the Muenchen and from the Salvation came from the destruction of the U-559 targeted fishing vessel Lauenburg on 28 June brought gold to Bletchley-keys an indicator book by the destroyer Petard in the eastern Mediterranean in October Men from the Petard boarded and enciphering instructions the U-Boat Short Signals Book machine instructions settings and the sinking boat with bodies floating all around them They succeeded in getting many documents the July home waters keys out but an officer and a seaman went down with In contemplating the damage the encipherment of the U-boat The documents they had given their weather brought on the Germans one cannot help lives for were current editions of the Short Signal wondering why they risked their most powerful Book and the Short Weather Cipher system for it Why not some other cipher The objective was to secure the weather information The sequence of events that followed their arrival but more important was concealing the fact that at Bletchley led to a break into the Atlantic U-boat the information was weather specifically weather key SHARK in December By January and Februfrom ships at sea But why put Enigma in double ary 1943 sinkings were reduced to half of the jeopardy by using it for that purpose and sending losses in the previous two months In June 1944 an American task force captured the U-505 and it out on small vessels vulnerable to capture retrieved from it the grid position Adressbuch There is another mystery Short-form Enigma This filled the last gap and thereafter the Allies weather reports from U-boats were decrypted and read naval Enigma currently sent to the weather central in Germany Selected To complete the picture Dr Kahn devotes a chapreports were reenciphered in a weather system ter to the American role in the naval Enigma and broadcast to ships at sea and shore stations How could the COMSEC authorities have allowed Once coordination was achieved the naval analysts at Nebraska Avenue worked closely with this to happen The only answer is that the Germans had absolute faith in the impenetrability of those in Hut 8 exchanging cribs and recoveries Enigma They did conduct investigations when In a final summation Dr Kahn writes that compromises were suspected but the findings ULTRA's greatest gift was that it saved lives always concluded that the source could not have Not only British and American lives but German been Enigma lives as well That is the debt the world owes to Bletchley codebreakers that is the crowning THE BATTLE OF THE ATLANTIC human value of their triumphs As Dr Kahn notes the Battle of the Atlantic was the only battle that lasted from the first day of the war to the last Thoughout much of that time the numbers of ships lost to U-boats reflected the status of solution of enciphered commuications on both sides The picture could be modified by other influences-improved DF techniques for examplebut the effect was always there sometimes draWill the person who borrowed Editing Your matically Sinkings in the last half of 1941 Newsletter in looseleaf format kindly return amounted to 600 000 tons In the same period of it to the CRYPTOLOG office Ops-1 2N018 1942 losses reached 2 600 000 The Germans No questions asked broke Naval Cypher No 3 and read much of the I BULLETIN BOARD 2nd Issue 1991 CRYPI'OLOG page 30 F6R M'FICtA L t J 6NLY I DOCID 4010108 On the Enigma P L 86-36 ENIGMA an off-line cipher machine was the first wired wheel machine to be used to any significant extent It appeared in Gennany about 1925 as a commercial offering at which time it had three wheels and a reflector Umkehrwalze In 1926 the German Army introduced a version which had different wirings a different reflector and a new feature the plug board Stecker-verbindung or stecker between the wheels and the input-out functions The Poles attacked the German Army usage and made limited progress by discovering that some from of ENIGMA was being used and that the first six letters of each message were probably indicators In October 1931 the French intelligence services developed a source within the Cipher Bureau of the Gennan Ministry of Defense who supplied them with copies of code clerks' instructions for ENIGMA and subsequently at regular intervals with copies of daily key lists but no wirings The French Cipher Service looked at these and immediately declared the machine impossible to solve French intelligence then obtained pennission to give the infonnation to France's allies and to suggest a common attack on the problem GC CS Government Code and Cypher School the predecessor of GCHQ was given first chance but they Filed their copies of the documents and did not respond to the offer of cooperation Gordon Welchman however believes the British had more effort against ENIGMA before the war than they indicated to the French The French then approached the Poles who accepted with enthusiasm and promised to share results of their work However all cryptanalytic efforts failed At that point 1 September 1932 those in charge of Polish intelligence brought in three mathematicians Marian Rejewski Henryk Zygalski and Jerzy Rozycki In the middle of October 1932 Rejewski was put to work on ENIGMA and using information in the documents from the French which Above the Enigma machine a mechanical device for coding invented by the German Arthur Scherbius Its keyboard is similar to a typewriter but the letters strike rotary disks that choose substitute letters at random to create encoded messages showed how the Grundstellung indicator system worked he was able to in about a month to develop a theoretical method for recovering wheel wirings based on the fact that message settings of the three wheels were enciphered twice to produce the six-letter indicator sent at the beginning a a message this was the Grundstellung indicator system Unfortunately the amount of work needed to carry out the calculations was prohibitive and might be prohibitive even with modern computers Then Rejewski was given two more documents that had been procured by France these 2nd Issue 1991 CRYPTOLOG page 31 FOIl OFFl6 USE OPHK DOCID 4010108 The ENIGMA as it was at that time looked like this U S 1 I A B @ 1__ 1 Z M A B STECKER F A B A B W I Z I_I Z - 1 - 1 - 1 - 1 Z containing two monthly schedules of daily settings Wheel orders Grundstellung wheel settings used for enciphering message setting and plug board connections At this time wheel orders were being changed only every three months but by luck the two months provided by the German source fell in two different quarters All this information permitted the theory to be simplified sufficiently for Rejewski to recover wirings of the three wheels and reflector by the end of December 1932 There were three 26-point wired wheels which could be inserted into the machine in any order but not backwards There was a reflector CUmkehrwalze which had input-output contacts on only one side that were connected to each other in pairs Each wheel had one notch on it which controlled the stepping of the wheels The fast wheel F advanced one position for each encipherment of a letter the medium wheel M did not advance for an encipherment unless either F advanced off of its notch or unless M itself was at its own notch in either of which events M advanced on position The slow wheel S did not advance for encipherment unless M advanced off of its notch in which case S advanced on position The reflector U did not step during the encipherment of a message and in fact was not settable The stecker between the maze and the input-output mechanisms could be changed by the operator but remained constant during the encipherment of a message it consisted at this time of 14 points connected straight through and 12 points exchanged in pairs INPUT- -I -I KEYBOARD -I -I _OUTPUT_ LIGHT PANEL In addition to all these variable the alphabet ring on each wheel by which the wheel could be set in a specified position at the beginning of a message was rotatable relative to the core of the wheel and could be put at anyone of 26 offsets Thus setting the wheels so that the same letters appeared at the bench marks would produced different settings of the wirings themselves if the alphabet ring offsets Ringstellung were different Once all these variable elements were appropriately set up the code clerk typed each plain text letter in tum and for each one wrote down as cipher text the letter which lit up Depressing the key also caused the wheels to advance appropriately In July 1939 the British and French were informed of the Poles' success on ENIGMA and GC CS accelerated its own partly s ccessful work against it ENIGMA was used by the German Army Air Force and Navy among other government elements Although rather heavy it was small enough for one person to carry and it operated from batteries so it was suitable for field use The Germans continually improved their usage of ENIGMA by changing wheel and reflector wirings a number of times by increasing the number of available wheels from which daily wheel orders could be selected they went from three to five to eight in the course of time bu changing the indicator system several times by increasing the number of steckered letters from 12 to 20 by devising the number of steckered letters from 12 to 20 and by devising nonreciprocal steckers o 2nd Issue 1991 CRYPTOLOG page 32 FOR OFFIGIM USE ffi'aN DocrD 4010108 4 nllhui lna hinebuilt Il S ' l kn 1lI17l ii tBreaKing into Our Past nigmas Cryptology embraces two ancient practicesconcealing the meaning of one's communications from unintended eyes and deducing the hidden meaning of the communications of others One we term cryptography the other cryptanalysis code making and code breaking Cryptanalysis we say yields communications intelligence COMINT But COMINT is both a product and a process Historically cryptanalysis was a matter of circumstance of the chance capture or interception of a cryptogram No thought was given to organizing systematic efforts to capture cryptograms-the communication technology would not prompt thought along those lines or support an effort How many daily hours of cover would one devote to capturing enemy couriers But the technology changed in the 1860s in America-not with Morse's telegraph which required a physical tapbut with a system of visual signals invented by an army surgeon Albert James Myer The Myer system a binary code by the way made battlefield telecommunication practical It also introduced a communication technology susceptible to interception The rudiments of cryptology are evident in the American Civil War A little more than a decade after that war the telephone came on the scene and it was enthusi- of 9Lnotlier 1 j na astically adopted by our army signal corps But from the standpoint of cryptology this was little different from Morse's electromagnetic telegraph-interception required physical contact with the wire involved The invention of wireless communication near the turn of the century changed everything Technology would now support an organized systematic COMINT effort even as it was demanding a new look at security Modem cryptology is built on the foundation ofthe First World War when military use of wireless first became widespread The advanced nations maintained some interest in the subject during the 1920s and '30s albeit in varying degrees For the United States this was the era of Herbert Yardley and his American Black Chamber and William F Friedman who created a scientific basis for cryptology But it is to World War II that we look when we think about the origins of modem cryptology and for good reason cryptology continued from hot war into cold war without dropping a beat In fact it thrived Now a half century separates us from the exploitation of the German Enigma and the Japanese Type 97 machine Americans called PURPLE That generation has largely passed from the scene Sons and even grandsons yes 2nd Issue 1991 CRYPrOLOG page 33 FeR eFFICIM GS G DOCID 4010108 and daughters and granddaughters have displaced them The passing of the generation that established modern cryptology has removed from our ranks the pioneers in what we now consider our profession We have lost some ofthe lore some of the technical knowledge and some ofthe enthusiasm that they possessed when ciphers were broken by individuals and when each accomplishment broke new ground The process has become less individualistic and more mechanical Mathematicians and scientists have replaced students of the humanities Sometimes it is difficult for newcomers to understand their role in the greater scheme that engulfs them In the fall of 1989 we moved to remedy this by putting greater emphasis on recalling to memory the past accomplishments and the pioneers who laid down the trail we follow The Center for Cryptologic History is an effort to rejuvenate our history program and to make it relevant to the needs of a new generation of young professionals and modern decision-makers The scope of our inquiry extends from a better understanding of the sweep of American cryptologic history and its European antecedents all the way to the lessons of DESERT STORM While the recent subject may be of more pressing interest our present purpose is to share a puzzle from the more distant past It is somewhat like the missing link to anthropologists in our case it is an unidentified cipher device that may extend our knowledge of our craft The Cylinder-Cipher Enigma In 1890 a disgruntled French cryptographer revealed a device he had been unsuccessful in persuading his own government to adopt The Bazeries cylinder-cipher comprised 20 disks or wheels-rotors one might say-arranged in a specified order on a shaft or axle The letters of the alphabet were placed around the rim of the disks scrambled The cryptography was that of polyalphabetic substitution an improvement on the familiar Vigenere in that the alphabets were mixed To encipher one aligned the disks to spell 2nd Issue 1991 the plain text then selected and transmitted as his cipher text any other line of mixed letters On the receiving end one set up the cipher text then simply turned the cylinder until plain text was revealed This concept and virtually the identical device was adopted by the United States Army Signal Corps and introduced in 1922 as its M94 That same year a scholar and former World War I cryptologist who was working among the papers of Thomas Jefferson in our Library of Congress discovered in Jefferson's own hand two papers describing the manufacture ofjust such a cipher device called a wheel cipher by Jefferson Since that time Jefferson has been considered the inventor of this type of cipher device He is so identified in the most recent book on the subject Silvio Bedini's Thomas Jefferson Statesman of Science published in 1990 Bazeries it was assumed independently invented the same device the Signal Corps copied him unaware of the achievement of its esteemed countryman Bedini followed the directions and made a copy But now the plot thickens as they say in detective stories A few years ago there appeared at a sale in the Washington area a device of this very type It was suspected of being of American Civil War origin perhaps Confederate The price was right we bought it Upon examination it was evident that pieces were missing including several of the disks Some disks had been removed and replaced in reverse order Letters were obscure or even obliterated altogether But nothing about the device suggested Confederate use On the contrary accented letters showed that it was intended for the French language Superficial appearance suggested that it was older than the war of the 1860s perhaps early 19th Century or even late 18th The former owner was from West Virginia an area that was part of Jefferson's home state during his lifetime Could this be the source of Jefferson's knowledge of such a cipher machine His undated description is not necessarily evidence of original thinking But how did the device come into American hands and when It retains 35 of a possible 40 disks each bearing 42 mixed CRYPrOLOG page 34 FeR 9FFIGIM YSil 9HIN r DOCID 4010108 letters digits or punctuation Would this larger more involved version be earlier or later than the Jefferson-Bazeries-Signal Corps concept We have no answer as yet The mystery remains But the mystery deepens with another bizarre parallel Notice to Subscribers Distribution for this issue reflects changes received by COB 30 August 1991 Some years back another former cryptologist was working among the historical archives of his own nation He too discovered a description of a cylinder-cipher similar in principle to the one we are studying The researcher was Sven Wasstrom of Sweden The documentation he had found was dated 1786 well before the earliest date postulated for Jefferson s description It described the invention dare' we say of a Baron FridricGripenstierna offered to King Gustav III Working with Boris Hagelin's famed CRYPrO AG Wasstrom assisted in making up a copy of the 18th Century device And there our story ends for the moment It is an amusing curiosity a diversion Or is it more than that In 1865 the United States Army came into possession of a type of cylindercipher used by the army that most threatened the very existence of the United States the Confederate States Army It was simply a mechanical means of using the conventional straight-sequence 26x26 Vigenere square or tableau But suppose each column had been separated by sawing through the cylinder Suppose the alphabets to have been scrambled instead of in direct sequence To change your mailing address please write your name old organization and building new organization and building on a piece of paper put it in a shotgun-not one-time-envelope and mail it to DISTRIBUTION CRYPrOLOG PI NORTH Now come forward in time apply electricity to this manual device The disks become rotors the heart of mid-Twentieth Century cipher machines the embodiment of another Enigma Whether for the old cryppie or the youngest member of the new generation discoveries of this nature illumine the still obscure background of our technical history They hint at continuity behind the scenes perhaps even at international cooperation--or espionage At a minimum they give us a fuller appreciation of our heritage We believe that such knowledge enriches our profession and justifies the modest attention we devote to the subject or send it by electronic mail PLATFORM cryptlg @ Ibarlc05 CLOVER cryptlg @ bloomfield Please do not phone about your subscription because o After every issue there are over 300 changes to the distribution about 300 in the first two weeks after CRYPTOLOG hits the streets o The distribution list is kept on line on a shared computer in another room o 2nd Issue 1991 CRYPTOLOG page 35 6ft 6 FICb L USE O1'HN DOCID 4010108 Editorial Dump Core Recycle Now is the time to come to the aid of the future Just about every analyst has a stash of technical papers working aids and other materials on favorite projects worked long ago that have been saved just in case or that really should be documented as people tell their friends some day when there's time Series a How to TN that's still valid too good to throwaway though it's coffee-stained maybe it should go somewhere but maybe it's filed somewhere Don't bet on it Recycle Send cryptanalytic materials to SAB 2 Door 22 T54111L - J P L SAB 2 Door 22 T5411 NSA Archives There's no better time than the present with reorganizations in the air and rumors of an early out Getting started is the hard part for it's a formidable prospect to document a project years later and it's tempting to put it off to a time when you'll feel up to it 86-36 Send all other materials to I I I I I I I I I IIP L 86-36 To the Editor I just read the article tf0rIllalization by Maybe you should try for Plan B instead of trying Ist Issue 1991 and I wanted I to document the entire project organize your I to compliment him not only on his readable I materials as a resource for future research All writing style but also especially on his mesyou have to do is put the papers or mag tapes or sage whatever together neatly Those red accordian envelopes are just the thing Write a Scope and I have been dealing with several third party Content paper-it could be hand-written on one organizations for years and have even visited sheet of paper This means indicating one of them for a short TDY Most of these what's it all about relationships have existed since long before I why it was important or proved not to be came on board at the Agency and I've always what years thb materials cover dealt with them from this side of the ocean what your role was so I guess I have considered them as bureauany details you can recall It's all right to cratic construction a given that was to be say you don't know dealt with in the context oftoday's world other comments I've never really given a lot of thought to the your name present organization date reasons beyond the obvious monetary ones that motivated our third party partners to There That's not so hard Now set it aside and enter into the relationship in the first place take out your perSUmm' Using it as an outline I dump core Tell all you know about those projects--opening and closing field stations setting up the very first whatever What it was like back in the olden days Add those comments to the surviving papers What about those general working aids and monographs that do not belong to anyone project MILCRYPT III a monograph in the Blue Ribbon I I In the age of the computerization of everything to the point where all some of us is push buttons all day long it's' refreshing to be reminded that our relationships with our partners is really very personal sometimes I I I 3204 L - ------------- CRYPI'OLOG page 36 FOR OFFIOw USE ONLJPY 2nd Issue 1991 t P L J 86-36 DOCID 4010108 CRYPTOLOG Editorial Policy CRYPTOLOG is a forum for the informal exchange of information by the analytic workforce Criteria for publication are that in the opinion of the reviewers readers will find the article useful or interesting that the facts are accurate that the terminology is correct and appropriate to the discipline Articles may be classified up to and including TSC ' Technical articles are preferred over non-technical classified over unclassified shorter articles over longer Comments and letters are solicited We invite readers to contribute conference reports and reviews of books articles software and hardware that pertain to our mission or to any of our disciplines Humor is welcome too If you are a new author please request Guidelines for CRYFTOLOG Authors How to Submit your Article Back in the days when CRYFTOLOG was prepared on the then state-of-the-art a Selectric typewriter an article might be dashed off on the back of a used lunch bag But now we're into automation We appreciate it when authors are too N B If the following instructions are a mystery to you please call upon your local ADP support for enlightenment As each organization has its own policies and as there's a myriad of terminals out there CRYPrOLOG regrets that it cannot advise you Send two legible hard copies accompanied by a floppy disk or cartridge as described below or use electronic mail In your electronic medium floppy disk cartridge or electronic mail please heed these strictures to avoid extra data prep that will delay publication o do not type your article in capital letters o do not right-justify o do not double space between lines o but do double space between paragraphs o do not indent for a new paragraph o but do paragraph classify o do not format an HD as DD or vice-versa-our equipment can't cope The electronic mail address is via PLATFORM cryptlg @ barlc05 or via CLOVER cryptlg @ bloomfield CRYPTOLOG publishes using Macintosh and Xerox Star It can read output from the equipment shown below If you have something else check with the editor as new conversions are being added Be sure to label your floppy or cartridge as to the hardware density format and software you used Don't forget your name building organization and phonel HARDWARE MEDIUM SOFTWARE FORMAT SUN 60 or 150 ME cartridge ascii only TAR or RAW XEROX VP 2 0 5 1 4 floppy only n a n a MACINTOSH 3 1 2 DD disk only MS WORD MacWrite TEXT WriteNow n a IBM Compatibles 3 1 2 1 2 ME disk 5 1 4 DD or HD floppy MS WORD WordPerfect WordStar ascii DCA IBM revisable DOS WANG n a n a n a 2nd Issue 1991 CRYPTOLOG page 37 6ft 6 ICIXL tffll 6NLV DOCID 4010108 jl C J'----- 'fIllS BOCt JMI3N'f CON'fAINS COBI3WORB MA'f'8RIAh NOT RELEASABLE TO CONTRACTORS This document is from the holdings of The National Security Archive Suite 701 Gelman Library The George Washington University 2130 H Street NW Washington D C 20037 Phone 202 994-7000 Fax 202 994-7005 nsarchiv@gwu edu