Historical Document · 1943
Navy account of Friedman's Hebern solutions, Serial 682,096, Rowlett's stepping maze, 1935 disclosures, index maze (5,855 vs 120), 3 Feb 1940 acknowledgement, 'Rowlett is entitled to full credit' (5 pp)
.t 2J. Mr. ;alliam F. Friedmc'lli, Prit~cipal Cryptanalyst of the Signal Intelligence .::service, and interested o.fficers at Signal Corps Headqua.l."ters were f..wdliar . with the various ff.odels of the HCl;., but not with the prospective changes which t.te r;avy had concaalei:l f!'cn~ Eebern. In f::..ct, on ~r. Friedman's recon.mendation, the Uignal Corfs purchF..aed b;o of Hebern' a early 5-wheel nonprintlng moqels late in 192.3. At the req~eat of the· ~Javy ~crartment, Friedman under~pok a crn;tunalytic test of tlJe HC?..i in the spring of 1924, being £umished a set of 10 test cr:,rptograms prepared by the Code ar;d Signal Section. Friedman was successi'ul, and develope"i crypt::i.na.lytic. · technic,,..ues whereby,, under certain conditions of meter action, solution ceiuld be achieved even without possession of the code wheels. Again at the req_uest of the Navy Departrnent, in April 1932 Friedman undertook a second test on the r,1uch improved 19.30 model of the HOM. This t:.n~e he W'a.S furnished the J1iachine J a description or the general system employed in setting up the message indicators, and a series of test messages .. AJain he was successful, with the aid of three or four of his assistants. As the test r1essages tr;ere enciphered with l!ebern' s stepping action and not with th'9 irz·ecular code-wheel stepping produced by the HCl.f Ad.apter (CSP 535), the solution did not worry us particularly. These solutions were very in1portant, in three ways, namely:I. They shotted the weaknesses of the meter action of the 1923 HOM and of 6 of the 30 optionaJ.. stepping actions of the 1930 llC!:L. II.. The 1924 solution was tbe basis of further analysis by tl:e r;avy which disclosed ste:µpi115 actions that would block arlalyticc.J. solutions or short-cut solutions based on possession of the code wheels.. Friedrran arrived at similar concll.4.Sions, independ-· cntly.. Othervfi5e, ue would have had to abandon the Elt!ctric Cipher 1'.achine as bei.ng deficient. in inherent security. III. In recent ysars, U.e principles and techni'1_ues of these solut.icns were instr\.Um:!nta.l in the solution of certain systel!l.S \'•hich are still us in?. ~ • U" a meter action.· 1 The first soluticn ( t!.at of 192.4) was written up by Friedman in a secret, technical pa.per completed early in 1924, which was not printed, hc·;rever 1 until .1934, under tt.e title 11.Analysis of a !.~echar,ico-:5lectrical Cryptograph--Pa.rt I. 11 l':t.e second solution (that cf 1932) was also written up by hi.1: in a second secret p3.por canpleted in 19.'.33 but not. printed until 1935, under the title "Ar~aJ.:,""Sis of a. I.:ec.lvlnic0-:Uectrictl Crypto.-r:r3f.-h--I'urt II. 11 Beth papers ·nere very carefully 3afe.;ua.rded at a.11 tin.es and were err.ployed only in the SIS for t.h.e advanced tra.in:i.ng.lo.f u. very liri1ited nuri.1er of students. The docuraent.::i wer-a given no disseminstio11 except that, the Navy De:t.iartment was furr.i&hed copies. Jut, because i~ was not consulted \Vith regard to the advis. ability of printin.:; these papers, combined with a serious ;r.istrust of the Gove1•nment Printing Office, the liav~' Department entcrt.:i.ined some apprehensions o:;.s to security an·:i this 'led to an orde1• fror:i t~.e ::).l?.C. thttt the .::>ignal Corps ·was not to be stown the l•:ark I ECli: or to learn any of its. deta.ils.. This order, which Ha.s not revoked ur. til J .~u1uary 194(}, was responsible for la.ter mis understr:;ndings. Cer.t.!:l.in Jigna.l Corps represei£ta.tives,, includin6 Friedlnm C:l.!ld ::r. Frc.nk B. Rowlett, had been shovm the pilot 1r.odel oft.he ?.iD.rk I Ec::Lir sor.uetic:le iu tl:e wint.er of 1934-35, befvre the order we..s issued, so ·t.hcy were not entirely ienorant of what. the l!avy was doing along tl':.ese lines. Declassified and approved for rele~!c-by NSA on 10-24-2013 pursuant to E. 0. 2i.. t~rpewritten, r -REF ID:A60492 - -...._........-- 11 25. From 1924 to 1932 the Signal Corps appeared more interested in the Teletype Scrambler than in the HCM as a practical cipher machine which would meet Army requirements. However, under date or 25 July 1933, the Chief Signal Officer filed on behalf of Friedman a patent application (Serial No. 682,096) covering a cryptographic system and machine in which the stepping or the code wheels was very irregular and under the control o! a keying tape. Electric control thus made its first appearance! Friedman made a complete assignment of his invention to the War Department and one or two preliminary models were built in 1935-36. These were successful and an order was placed with a relatively small and inadequately equipped manufacturer for a few machines, which were designated as Converter M-134A. It took a comparat~vely long time to build these few machines but by 1938 some of them were delivered and placed in service for communication between the War Department and the Commanding Generals of Overseas Department~. Later, additional ones were delivered and placed in service for intercomrrrunication among the War Department and Corps Areas and between the War Department and the U.S. Military Attache in London. The first model of this machine was shown to me by the Signal Corps sometime in 1937. This machine indicated the reliability of electric control but the undesirability of the particular method (perforated tape) used in the Signal Corps machine • . 26. Shortly before 15 June 1935, during the interval when preliminary models of the foregoing machine were being built, Mr. Frank B. Rowlett, principal assistant to Friedman, conceived the idea which constitutes the basis of the "stepping maze" in the present ECM. His concept was based upon the principle of sending an electrical impulse through the circuits of a code-wheel maze to generate a long, irregular sequence of events which could then be used for various purposes, such as keying. Rowlett and Friedman then jointly developed Rowlett's novel idea of a key generator as applicable to the Signal Corps machine and reduced it to more practical form in drawings. No model incorporating their ideas was built by the Signal Corps, however, because the Chief Signal Officer was conunitted to the type embodied in the Converter M-l.34A, pre-producation models of which were then under manufacture, and he wa~ reluctant to make any change in design, despite Friedman's urgent recommendations that this be done. The inventors proceeded to incorporate the results of their theoretical studies and their drawings, reducing the new principles to practice in a patent application filed in the Patent Office on 23 March 1936 by the Chief Signal Officer on their behalf as joint inventors (Serial No. 70,412). The inventors made a complete assignment of their invention to the Secretary of War on 2 April 1936 and the application was processed through the Patent Office, though, of course, it is held in the secret status. Nearly all of the claims (39) have been allowed in the case. -11- - -."""·"',__=""l="""""'=t . 1---r~-~!!!!-!l!i!~~-M!;_;:'!_-~-~----~-~--~--~~!!!"!!!!!~~=~~;::!!"!__!'!!!!!!!!!!'!!.!!""!!!___ ""!.!!!"_."!!!--~-!!"!!!!''!'!'!.. !!!!!!!!!!!!!!!!!II ___ ...:;:..::.-If'..--·--:------·---·--·--·~:.··-·- ..:._ ............ .:..f-~--·"'" .... !'!"'. REF ID:A60492 27. In Cctobcr 19'.";.;, Friedr.:..Lan anti Lii::utcnar.t ·i.t·n,:::ar (of tiit:: Gode anti. Sisnal .Section) hblli a e.;enE-ral tiisc-~ osion on cipner rr.act.ine:s. ~.cn,::;er e-.xprescea consh.erc::.bltl l!.issat.:':_ol'actLn id.th t.i:.e ...'..i:.rk I .uC2.~ anci asked Friecl.tan whcti1er ti:.tii .3i;n!..l ~orps had. any 11 good" it;.eas alone; t 1~eso lim:is. FritiC:.:i:a.n indicated tt•at tl.iere \;ere &ei.•;;;ral iceas -w."ldcr. the Sic;r~al Corps was not exploitine but wldch he was not .at liot:.rt;;: to disclos1..·, since t.i:e:.- .-.ad teen pl&cec.. ir! t.he secret category. Fried!1i£1n turt.ilE>r indica.te<.i tiiat i f :ier!.[;.::r so C.esired, :;>1:>rr.-.ission l.o C.if:clos<:. tbc::_r, l.o tn!::l Havy ,.,ould be requecLeC.. ·.-~elli:ur asked ti.e:.L ti.is ;,;e done. .h.ccordin2·l:t, Friedman reqt<.edE;d antl was granteci pt:.r!l'i.ission b:,· ilis St;periore to G.isclose 1..r.e ci.et~ils of tr1e Friedr.an-!fowleLt pu.tent ap;llico.ticm to representativE:.s of tlie !:a-..;y Departrnt::nt. ,··Tr;.erefore, on ;a Oc;tober 1935, at a conferf::nce in Frieci:;an•s office, tne Gdails \'Hlre clisclo.::cd to Co!T'.!£Lander 1,:cClaran and Lit."Ut1;;nant ~:ell£cr, who we1·0 s .. ov.n the drawings ti.atJ for::icd tte basis of the pat~r,t application Serial !fo. 70,412. Cn ,'.jl Octoi:cr 1935, a seco1.d and sbiilar disclos:-re v;as !'.!ia.Ce to Co:r.r11ander ;i.cClaran Lieutenant ::en.:;er, and Lie:t;.tunant Harper. A t!".irc>_ C.isclost;.re was :nad1:; on 1 I:ovemter 1935 to Lieu"l.t:;rumts ;.ocd and Dugan, also of tr,e Code and Signal Section. Friec.ima.n and Roule.tt -~;ere:: told very little as to the t;a.v;,r Department 1 s reactLm to th~ disclos-ures; in fact, ti.ey \;ere told ti:.at tne principles disclostd \':ere of no interest to the r:av:: at that ti.:1;.e - which was t~e truth of ti;.c matti::r. 28. ;.:_y first-hand knowled<:"'.e of tiie Fri edT:a.n-Rowlet t invention be.Jan in the win"'..er of l<t~6-.37 wt.en we were preparing initial specificc.tions for tile if.art II .:!.Cl:. :;enger stated that Fri .. dman had an idea i'or an electric control v;:!iich had very intere.stin5 po~sibilities anti procuc~G fro.in his safe a single si.e~t of cross-section pa.per containin[; t!;ree el1:;,1;,entc..ry .-.-iiring-deagra·.. s by mcar:..s of wi1ich clt::ctric control of an ..1~:·.~ coi..:.lu. be acllievE:.d ti·lrough an .i!:CY. r..a.zc. This paper wus dated and signl-d (as I ri::_i.ec:.1ber) by J!a.rper, -;;engcr, anci ·;:ooa., and by Friedman and il.owlett. C-,!e hav1;; been unable to loc&tE:: ti.is paper since 1940.) I k':iediat13ly raalize;d that electric control e;ave us the answ(..r to tr.any of our ti.nsolved proble::Hs anC: ti1eref ore i.ad to be incorporate(: in the new !!1acl:..ine. I was under orders not to discuss or sl.ow eitl1e:r ·tht:: ~!.ark I :ii.Ji~ or t.Le Yark Il :"!.c:. : to t.ile Signal Corps and, tiiere.f'ore,vi,d.opted electric central and further developed ti:ic basic idea witi·~ot.;.t tuE:; knowledge of ti:.e original invt::ntors. ln Janm;ry 1940 the .::.:ark II ~Cl~ was offered to ti.e ·::ar Departi11ent for Joint J...rmyI:av-.r 1.1.ee ana also for pti.rely Army uce. It was exp],ained that t!·.e mr.cha.nical fcat-i..res were well developed a.nd "frozen11 in design, and that \·,e belioved the Army woula be well satisfied with tl:.e cryptogr¥hic principJ.t~s involved, but triat we were willing ·t.o disct<ss any sect;.rity f eatl:.res in 01 der to get a 1}1B.Chine t~-.a.t woul..;. be satisfactory to both services. ""t":e wan tee. the ArrriY to join us on tr.e first order :for L.he macLine in order to further tr1e idea of using identical cryptograp:1ic systems in t.-.e two services, as had already been done with the Strip CipLer Device. Another reason was to share the overhead i'or tooling-up ana ti:iereby give us a better price. It had been previously suggestt-ici tliat tne Arm;:_.- and Navy .zet together on the Signal Corps machine or the-i-tark I EC.M. We advised that neither macldne 'l't"a.S acceptable because of mGcli.anical deficiencies but that we were develo1>ine a new 1r..achine and as soon as we had a 1:orking rr.oiel we would endeavor to get permission to :n.ake it, available as a corn.'D.on Army-Ifavy :na.ctine:.. -12- a ..... ~,......___________lliTlll!Ei"-il1iiiri~.c~--F'.·-~--=·~·-!·'!·!"·'!'?~.!~'~?~·~12!-4.i~J~·~~IP~--;-#~f.r.--~-.~. ."'f'o--·~·! ...... ,. ' ··-··· II '·'--RE:F- 1 IID:Ai60492 . ' 29. en 3 Febrnar~r 191+0, i;.dr:1iral I:oy'3s (J.i; .~.) invited. 3c1:~ral ~.:auborzne (Jf\ief Sie;r:al (;f!'icer), Captuin ':}ook·, :.fr. Friedr1len, ond otlier Sicn~l Corps reprP.sent.=.tives t.o inSF·ect a }:-i:!.vt. model o:!" the ! ·urk :u :2GJ..·. (;n that occas:'.(Jn I ackno•sl~d.;ed to l;:r. Friedman, in tho presence of Gener<!.I .i.•c.uborgnc .::~nd zi..dmirc'.:ll ?foycs' our t;.Se of his invantion. Later the.re WC-.5 a sr:acial conference att.ended by }.:r. li.eiher W"!d ~.Ir. Zenr.or of the 'I'n:!.et:rre :Jorporr.ticn, 1~r. Friedman of the ,:ji~;:;.l -;orr.s, Go:;-,n:r~r;d,er Sa.f!'ord arid Lieute;iari t, Zcrn of Kaval Cor.-~'T.un-1.ca.t i.ons, and pnssibl;r ott~rs. The 'blue prints wero carefully examined e:.nd a eencr.s.l discussion or c~rpto.;r.::ph:Lc !'e3.tures follmred. Friedman rointec out thnt tho t:nderlyin~ principles of tl~e control circu:!.ts of the I.lark :a EGL ·11ere those wLich laad been disclosed by .?..owlett and himself to the Navy Depart1,;.F.n\, in 1935, and this was con!'irrr.l:'ld b;r :r.e. The ;.o~r axperimenta.l changes to the Friedr!ia.n-Ro-.'llett. circuit wbi.ch had been ir.a.de by Seiler and myself were discussed and the followint~ eecisions made: m " I. "Index l:aze, 11 which repla.ced the plugho~rd :i.n the :friedrr.anii.owlett invention - Retained. ':'he :•Index 1.:u.ze" accompllshed the sar.-1e cryptogra~hic result as the p:!..u:;board bl.. t. Has n~uch more convenient tr) the oper.:~t.or. II. Grouping of end contacts in the 11,":;tepping !...::i.ze" and in the ~ "Index llaze, 11 which replaced the arranBerr.er:.ts of tLe r'ried!r•.:mRowlet t circuit - Ret.:.i.ine<l. These groupings t""'c;ether ·,dtl1 the ten circuits throui;h the "Index :.iaze 11 cave 49 tim.es as aany steppinJ; combinations u.s was possible with t.l:e Friedn..:.n-P..owlett inventio1. ( 5, 8~5 a.gai.nst 120). III. Subdivision of "Jtep!Jin,!; l:"aze" into tiio parts - Una'1ir.:ous decision to return to the original Friedn1ar1-R.owlE>tt 11.:;tepping I.faze .. ·' Friedmar. prote&ted the sut.division as an unn('ccssary complication. Reiber a.~d Zer.ner did not like it fr~m the viewpoint o:f design and construction. IV. Steppine order for the 11;;>tepping l.:aze'' rroposed by the t~.~Ny was .3-1-5, the other two wheels being dead to si:.1pl:i fy construction. Tte stey;pint; order was chan.ged t.o 3-4-2 upon Fried~an's reco~Liendation •• ·~:ith these exceptions tl,e 11art II Eel:, &S develo1,ed by the i;av:r and Teloty~-.e usine the Fried11an-!towlett "Stepping i;aze,-1 wc..s satisfactory to l!lld accepted by the Arrr.y. Washin5to11 Navy Yard sketch Rr"i68F201, d.oted 2l~ Ayiril 19J,O, used as a basis for specificat.ior.s of the production model, is the earliest-dated dra.•Ning showing the 11.Steppinc'j .lla.ze" and associated _cj_rcuits eXc?ctly in their present rorm. One other con-t:.ribution, !.~ajor Leo n.osen' s 11 Plugboard CoC.e ~iheel, 11 ca.me in 194.3 l'.l.fter t.he Em.: was in sel'"'lice. This was developed by the Sienal Corps for field use, where the d.r.neer of capture w:1.s creater t!·,an in the Navy. The 1'Plugboard Code Viheel" was ~dopted for joint Army-Navy use at the re~uest of the .Army, but is being distributed to all Iia.vy holders of the F.CE. The chief value of the 11 rlugboard Code ~·fueel" to the Eavy is possibly rsychological, but we do have it in caoe of need. 30. ,J . ·-. .: . r· -t-.' ........ ~ •. -··.:.. ~. ,~ .... ~. ' ·I I ·I -13- ···---'-~ . ~ . . . ____ ..;._._~ 4 -)··v"··· ··-' ..f • Jl. Electric control of the ECM by means of the Friedman-Rowlett 11Ste}Jping L:aze 11 is the essential feature that places the Mark II EC11 in a class by itself as regards security. Those vlho have participated in the development of the l:ark II ECU have al-wrs..ys acknowledged the contributions of the Signal Corps. The 11 Index Maze" and grouping of end contacts add to the security afforded by the 11.ltepping l!aze,:1 but would be worthless without it. The importance of electric control can best be estimated by a con~deration of what the :Mark II ECH would have been if Friedman and Rowlett liad not been pennitted to disclose their invention to the ~!avy. Although the "Stepping 1Iazc 11 appears obvious, now that it is in use, no one in the Navy thought of it in a period of 15 years, and no foreign machine employs it. Therefore, the IJavy woilld have continued the develop1r.ent of the older methods and the new ECU: would have used the mechanical stepping control found in CSP 903 or CSF 1700.. We would have had a secure m;achine, superior to anything in use by foreign nations, but definitely inferior to our present ECli. This hypothetical .ca.chine (a.s well as CSP 17C•O) would defy attempts at solution until such time a.s machine and code wheels were captured. Arter -this, ea.ch dey.1 s keys would resist solution for a long time. 11Short-cut11 solutions would be impossible, due to the erratic stepping or the code wheels, but a trial-and-error solution i'tould be within the range or possibility. We could not make the flat statement, as We do for the Mark II ECM, that solution would be utterl~r impossible. In other words, th£' machine would be adequate to take us through Rorld r;ar II but, because we had stopped short of the ultin:ate step, there would. always be the desire to develop a new machine and scrap the old one. Ro\vlett is entitled t() full credit. .for his discovery of the principle of the key generator as ecbodied in the "Stepping :Maze," which adds so rruch to the excellence of the Mark II Ecg_, and Friedman and Rowlett jointly are entitled to full credit for their joint invention of methods of applying and reducing the principle to practical form. I 32. The Signal Carps acceptance of the J.!ark II ECJ.r for Army as well as Joint Ariny-Navy use reflects credit on all who made that decision. The Joint ii.rrny-Navy ~CM Cipher System became effective on 1 August 1941, and the two services had a. co!"llI!'.on high-security cipher ·systerr. in effect and in use prior to the attack on Pearl Harbor. This use o!- an identical machine with interchangeable code wheels has been of great military value, particularly in the early stages of the war when the distribution of machines and code wheels . was incomplete. In the Philippines, Java, .h.ustralia, and even in north Africa, Navy wheels have been used ir. Arny El.""'l;Is, Arnr.r wheels in Navy ECMs; machines have been borrowed back and forth between the two services; Army messs.ges have been sent in Navy EC:i.·: ciphers and Uavy messages sent :1n Army ECl~ ciphers • 1 i I I. ' : l i~ ; 1i ~ I ..---1 .: ~ ' ,' ~ ~ ' •I ( -lh- ...... '··""""·.:;·-· ... L_ __