annette stein interview (mors) · and annette stein. mr. arthur ÔÔartÕÕ stein,...

Calhoun: The NPS Institutional Archive

Faculty and Researcher Publications Military Operations Research Society (MORS) Oral History Interviews

2014-19-01

Annette Stein Interview (MORS)

Stein, Annette

http://hdl.handle.net/10945/49262

CLARIFYING NOTE

The Preface, below, was written in 1997when the intent was to publish thethen seven histories developed under

the Army Oral History Project. Subsequently,a decision was made to delay publication.By the time I retired from the Office of theDeputy Under Secretary of the Army (Oper-ations Research), ODUSA (OR), I found thatWalt Hollis, FS, had originally commissionedDr. Wilbur Payne, a well-known and highlyregarded Army operations analyst, on Wil-bur’s retirement from his last Army positionas Director, TRADOC Operations ResearchActivity (TORA). Unbeknown to any of us inthe ODUSA (OR), Wilbur had completed twointerviews before passing away in 1990. Thesubjects were Dr. Hugh Cole and Floyd Hill.The tapes of those interviews were found byDr. Dan Willard when he was asked to gothrough Wilbur’s papers and files by MaryFarley, Wilbur’s widow. I had the tapes tran-scribed and the transcriptions reviewed by thetwo subjects who approved their publica-tion. The added interviews brought the totalnumber of histories under the Army projectto 10. The interviewees were Hugh Cole,Margaret Emerson, Abe Golub, Frank Grubbs,Floyd Hill, Hugh Miser, George Schecter,Nick Smith, Annette Stein, and Art Stein. Asof now, the following who have died are HughCole, Abe Golub, Frank Grubbs, Hugh Miser,George Schecter, Nick Smith, and Art Stein.

After my retirement from federal ser-vice, I recommended that MORS pick upthe ‘‘baton’’ and perform additional oralhistory interviews concentrating on pioneerMORSians and early MORS Fellows of theSociety. MORS established the Oral HistoryProgram in 1998 and has since capturedmore than 50 oral histories of distinguishedMORSians and prominent members of themilitary OR community.

This volume of MOR presents a Fore-word written by Walt Hollis and two of theabove-mentioned oral histories: Art Steinand Annette Stein. Mr. Arthur ‘‘Art’’ Stein,FS, was the third President of MORS, servingfrom 1967 to 1968. Dr. Annette Stein, Art’swife and one of the human ‘‘computers’’during World War II, worked for Dr. Johnvon Neumann. Dr. Bob Sheldon, FS, whomI mentored in the MORS Oral History Pro-gram starting in 1998, helped edit these oralhistories.

—Eugene P. Visco

PREFACEThe oral history project was assigned to

me by Mr. Walter W. Hollis, Deputy UnderSecretary of the Army (Operations Re-search) in the winter-spring of 1992. Hisinitial guidance to me was to seek out sup-port from other Army elements, most spe-cifically the Center of Military History inWashington, DC, and the Military HistoryInstitute at the Army War College, CarlisleBarracks, Pennsylvania. Together, Walt andI developed a provisional list of candidatesfor interview. Our criteria were simple: thepeople we were looking for would have be-gun their military analytic careers no laterthan World War II and they should nowbe ‘‘vertical, ambulatory, and cogent.’’ Upto two of the last three, somewhat light-hearted, criteria could be waived. As Waltpointed out in his Foreword, we had alreadylost many of the stalwart founders of thepractice of Army operations research. Thereare still, fortunately, many who still meet thecriteria. So many, in fact, that the oral historyproject has a rich trove of potential contribu-tors. Because of time and money limitationsonly seven people were interviewed for thisbook. We hope that this little book will be butthe first and will open the door to manymore like it; time is of the essence if furtheroral histories are to be gathered from theArmy founders.

Following Walt’s guidance, I met withBG Harold Nelson, then Chief of MilitaryHistory, COL Thomas Sweeney, Chief, Mili-tary History Institute, and Dr. James Williams,then Director of the Oral History Program,Military History Institute, and now Cura-tor, Armor Museum at Fort Knox, Kentucky,to develop a plan and methods for the oralhistory project. Subsequently, Jim Williamscollaborated with me on development ofthe Army OR Oral History project. Togetherwe developed a short information paperused to introduce the project to intervieweesand help get the dialogue moving. We con-ducted some of the interviews jointly, withJim taking the lead at the outset, while Ilearned from his considerable experience.

Six of the seven histories are includedin this volume; the review of the seventhby the interviewee was not completed be-cause of complexities in the interview andsubsequent family problems. Maybe someday that interview will be available forpublication. The six histories come from

MilitaryOperationsResearchSociety(MORS)Oral HistoryProjectInterview ofMr. ArthurStein andDr. AnnetteStein

Eugene P. Visco

Retired from the Office of theDeputy Under Secretary ofthe Army (OperationsResearch)

MILITARY OPSRESEARCH HERITAGEARTICLE

Military Operations Research, V19 N1 2014, doi 10.5711/1082598319169 Page 69Military Operations Research, V19 N1 2014, doi 10.5711/1082598319169 Page 69

Drs. Annette Stein, Hugh Miser, and NicholasSmith; Ms. Margaret Emerson; and Messrs.George Schecter and Arthur Stein. The last twogentlemen are now, sad to say, deceased, result-ing in great gaps in the military analysis commu-nity. Hugh Miser and Nick Smith both begantheir careers during World War II as membersof Army Air Forces operations analysis sections.Later, both served with distinction, Hugh withthe new US Air Force and Nick with the JohnsHopkins University Operations Research Office,supporting the US Army. Annette and Art Steinbegan their professional careers during WorldWar II at Aberdeen Proving Ground. MargaretEmerson worked on the Army staff in the Penta-gon during the War. George Schecter spent thefirst half of his distinguished research and analy-sis practice in the Army’s arsenal system. Jimand I jointly interviewed the Steins and GeorgeSchecter; I interviewed Margaret Emersonand Nick Smith; and Hugh Miser was inter-viewed by LTC Lawrence S. Epstein, an ArmyReserve officer working under our directionfor this project.

The lives presented in this book span a widerange of Army analysis activity and are a goodand interesting introduction to the origins ofArmy operations research.

FOREWORDFrom the time I was appointed Deputy Un-

der Secretary of the Army (Operations Research)in 1980, I have been concerned with the heritageof the practice of operations research in theUnited States Army and in the larger defensecommunity. When Dr. Wilbur P. Payne, a premierArmy operations analyst and the founder andfirst occupant of this office, retired from full-timeservice as a Senior Executive Service official, Isaw an opportunity to contribute to the record-ing of our heritage. I prevailed upon Wilbur toundertake interviews with distinguished analystswhose careers began during and just followingWorld War II, recognized as the beginning ofthe modern practice we call operations research.Sad to say, Wilbur became quite ill shortly afterour discussions and soon passed away, leavinga major gap in our community. When Mr. EugeneP. Visco, once a colleague of Wilbur’s at the Johns

Hopkins University Operations Research Office,the Army central operations research organiza-tion from 1948 to 1961, came under my direc-tion in 1987, I saw fit to assign the oral historytask to him.

We realized that it was already too late tocapture the thoughts and personal recollec-tions of many of the early leaders and contrib-utors to Army analysis. Dr. Ellis A. Johnson,founder and only Director of the ORO, wasno longer with us. Dr. Dorothy K. Clark, a finemilitary historian and analyst who mademajor inroads in the relationship between ca-sualties and success on the battlefield andwho worked in the field in Korea, Vietnamand Thailand, had passed away. No longer withus were James W. Johnson, one of the architectsof pentagonal division, Dr. George Pettee andLynn Rumbaugh (the latter two were mainstays in helping Ellis Johnson design and de-velop the ORO), and many other distinguishedpractitioners. Nevertheless, there are manyanalysts still with us whose words need tobe preserved.

This little book represents the beginning tothat task of preservation. Although the Armymay not be able to continue the oral history pro-ject, it has set in motion a process that I ferventlyhope will be continued by others. The MilitaryOperations Research Society, for which I amthe Army Sponsor, is an admirable choice topick up the torch. With Gene Visco now retiredbut still active with the Society, that may come topass. If so, I am sure the oral history will benefitby expansion to include analysts from the otherservices. I note that two of the analysts whosehistory is included in this volume got their startwith the Army Air Forces (now the US AirForce) during World War II.

I commend these pages to the younger andeven the middle-aged analysts now practicing.There are important lessons to be learned aboutsearching for operational data, understandingthe operations being analyzed, and presentingresults to decision makers. I recall the advicegiven by P. M. S. Blackett, one of the principalsin the founding of operational research in theUnited Kingdom during World War II: ‘‘I thinkthe essential prerequisite of sound military ad-vice is that the giver must convince himself thatif he were responsible for action he would

MILITARY OPERATIONS RESEARCH SOCIETY (MORS) ORAL HISTORY PROJECT INTERVIEWOF MR. ARTHUR STEIN AND DR. ANNETTE STEIN

Page 70 Military Operations Research, V19 N1 2014

himself act so.’’ Good words for analysts to keepin mind.

Read and enjoy!Walter W. Hollis, Deputy Under Secretary of

the Army (OR), The PentagonApril 1997

OPERATIONS RESEARCH ORALHISTORY PROJECT INTERVIEW OFMR. ARTHUR STEIN, FS

Gene Visco, [email protected]

Dr. Jim WilliamsU.S. Army Military History Institute

INTRODUCTIONMr. Arthur ‘‘Art’’ Stein was the first Secre-

tary Treasurer of MORS in 1966. He then servedas Vice President of MORS and later as the thirdMORS President, from 1967 to 1968. He waselected a MORS Fellow of the Society (FS) in1991. Mr. Stein earned a bachelor’s degree inmathematics from City College of New York(CCNY) in 1938. Art began his early career inoperations research at the Army Ballistics Re-search Laboratory. He was named head of Op-erations Research at the Cornell AeronauticalLaboratory and remained in this position untilhis retirement in 1974. After his retirement, Artworked at Falcon Research and DevelopmentCenter until December 1983. After a second re-tirement, he became a consultant at the Institutefor Defense Analyses (IDA). Mr. Stein passedaway on June 24, 1995. The interview was con-ducted at IDA, Alexandria, Virginia, on April 2,1992 and at Mr. Stein’s home in Williamsville(near Buffalo), New York, on June 5, 1992.

Gene Visco: Would you begin by giving yourname, position, and address?

Art Stein: My full name is Arthur Stein—nomiddle initial. I am presently a consultant hereat IDA and have been since 1984. I retired twice.After about 15 years’ service, I retired at the be-ginning of 1974 from the Cornell AeronauticalLaboratory, which had been sold and named

CALSPAN Corporation. I then opened a branchof the Falcon Research and Development Com-pany, whose headquarters was in Denver. Falconwas wholly owned by the Whittaker Corpora-tion, a conglomerate. I retired from Falcon in De-cember 1983. I turned it over to one of the otherfellows in the group. That company is now anindependent company; they bought themselvesout. They are now called ANSIM (Analysisand Simulation) Corporation. Until two yearsago I maintained a separate office in the Buffalo,New York, area. Now I have my office in myhome in Williamsville, New York.

Gene Visco: I’d like to ask you some basicquestions about your background as it mightpertain to operations research. Did you haveany military, family, or educational backgroundthat was relevant?

Art Stein: With respect to military experi-ence, the highest rank I held was staff sergeant.My Army specialty was mathematician. How-ever, I had two main activities during the timeI was in the Army. One, I served at the BallisticsResearch Laboratory (BRL), at Aberdeen Prov-ing Ground, Maryland. I came back to that lab-oratory as a private, after having been a civilianthere for three years. Also, half of my time inservice was on a ballistics team. This team,which was a 12-man squad, had the responsibil-ity for calibrating artillery weapons in the field.So, I did that. About two or three months beforeI was discharged and left service, a new projecthad started. I’ll tell you something about thatlater. As a result of that new project, I was askedif I would assist. They needed a statistician—somebody to do some test planning. It soundedquite interesting. That’s how I got into the air-craft vulnerability business, because the projectwas associated with obtaining information fromwar-weary aircraft. That was very shortly afterWorld War II was over. My Army service wasfrom July 14—Bastille Day—1944 through June1946. My childhood and education: I was bornAugust 5, 1918, in New York City, where I wasraised and educated. I received my bachelor’sdegree from City College of New York (CCNY)in 1938, with a major in mathematics. This maybe of interest to people in operations research:At that time the possible fields for people inmathematics were very restricted comparedto today. All we knew—and perhaps we were


Military Operations Research, V19 N1 2014 Page 71Military Operations Research, V19 N1 2014 Page 71

a bit provincial—was that you could teach it oryou could become an actuary. There were ap-plied areas, such as astronomy and engineer-ing. However, they would require that I go outof town to some university and that was out ofthe question in my particular circumstances. Itwas my intention to become an actuary eventu-ally. It was very difficult and time consumingto go through that actuary sequence, becausefirst you had to be working in an insurance com-pany and then take seven sequential examina-tions, normally spaced over almost 10 years.There was a depression on then, and openingsin insurance companies weren’t there. I tried.As a backup, I took some undergraduate edu-cation courses, as well, and then did some prac-tice teaching in mathematics at a now-defunctand very unusual high school in New York—Townsend Harris High School. This was a schoolfor bright young boys. Upon successful comple-tion of Townsend Harris, which they could do inthree years, they could enter CCNY without hav-ing to take any examinations or have any partic-ular grade level. Others going into CCNY had tohave a very high average—93 or 94—to get intothe college. I entered CCNY based on my gradeaverage. Incidentally I graduated from CCNYat age 20, which wasn’t unusual in those days.In fact, I would venture to say that, in that partic-ular college, almost half of the students graduat-ing were probably on the order of 20 years old.After that, I went to Columbia University. WhenI graduated from CCNY, there were no positionsavailable because of the Depression. I had takena number of civil service examinations and waswaiting to hear from them. One was for meteo-rologist. I studied and learned something aboutmeteorology. I took some local civil service exam-inations for clerical positions. Lo and behold,one came along. In September 1939, I went toColumbia University to major in mathematicalstatistics, which was in the School of Econom-ics. What turned me toward that particular areawas a conversation with a second cousin. Fami-lies were very close-knit in those days. This sec-ond cousin was really the fair-haired boy in theextended family. He was going into medical re-search and had studied two years of mathemati-cal statistics to equip himself to do that kind ofwork. When he heard about me, he asked tosee me and talked about the opportunities that

would exist in mathematical statistics. So thatopened up some new possibilities, and I thoughtthat made good sense. I knew I wasn’t interestedin theoretical mathematics, per se. I was inter-ested already in problem solving, and mathemat-ical statistics seemed to be a way to do that.I went to Columbia, which he had suggestedand which was fortunate because it was inNew York. I could continue working in the cler-ical position with the New York City Departmentof Welfare, which was the largest department inthe city in those days. Professor Harold Hotel-ling, who was one of the best-known academicsin mathematical statistics, was at Columbia.(We may come back to him later in connectionwith some papers for the National Defense Re-search Council, which he was involved with.)I was fortunate, as well, in that all of the coursesoffered in mathematical statistics could be takenstarting at 5:10 p.m. That is, half the courseswould be during the day and half would be dur-ing the evening. The next year they inverted thetwo. So it was possible in two years to get yourmaster’s degree. That was what I did. Columbiadid that to accommodate people who were work-ing. Many of the people in my classes were work-ing in various insurance companies. It wasinteresting. In fact, some were actuaries. Duringthat period, the thing that influenced me most to-ward what we now know as operations researchwas a course I took just to fill out the requirednumber of credits. I had taken all the coursesthey had in mathematical statistics. This othercourse was given in the School of Business. Itwas called internal statistics; it related to statis-tics of the firm or company. The first week con-sisted of the very simple statistics that youmight see in business statistics. After that, itturned out that what this man was presentingwas solving various problems—scheduling,queuing, and optimization—and much of thiswas analytical or trial-and-error, because wehad no computers and no theory to fall back on.It was trying to think through the process—how you would arrive at the best way to pro-ceed. That I found to be one of the most inter-esting courses I had taken. The course wasgiven by Dr. R. Parker Eastwood.

Shortly after I had started at Columbia, onlyabout three or four weeks into Hotelling’scourse, he introduced a little man who came



down from the back of the room. Hotelling saidhe was going to India to teach for a year. Thecourse would be taken over by this little man,Dr. Abraham Wald, who was a Hungarian refu-gee. Wald eventually became the leading mathe-matical statistician in this country. Unfortunatelyhe had an untimely accidental death in an air-craft accident in India. I guess he was in his 40swhen that happened. He was responsible forthe development of sequential analysis. It’s inter-esting that when he first started to lecture, wewere so disappointed because we couldn’t un-derstand his English. But he did much writingon the blackboard. After the third lecture wedidn’t care particularly if Hotelling didn’t return.Hotelling was a very nice guy but rambled. Waldpresented everything in so logical a fashion andmade the difficult seem simple that it was reallyinspiring. He was very, very modest—an unas-suming person and the kind of professor peoplereally grew to love. He was my professor for per-haps half of the course work I had at Columbia.

Gene Visco: When and how did you first startdoing work that became associated with opera-tions research?

Art Stein: As I was nearing the end of thetwo-year period at Columbia, I could not getmy master’s degree because there was a requiredseminar that was not given in the evenings. Itwas the one exception to the ease of schedulingall those courses. It was early 1941. Things werebeginning to heat up. The war in Europe had al-ready started and things were already starting inthis country in terms of supplying the alliedforces. Hotelling suggested that the only manhe knew who was really doing any work in ap-plied mathematical statistics was Captain LeslieE. Simon. Hotelling thought maybe he was a ma-jor by then. Also, applied work was being doneby people at the Bell Telephone Company. Theywere working principally in probability, andWalter Shewhart had just started some of hiswork in quality control. Anyway, Hotellingthought I should write to Simon at the AberdeenProving Ground. So I wrote to Major Simon. Itturned out Major Simon was Colonel Simon bythen, and he had just become the director of theBRL. He was a very remarkable man. He hadwritten a book, An Engineer’s Manual of StatisticalMethods, which was a very forward-looking, in-teresting book on sampling and quality control,

while he was a lieutenant at Picatinny Arsenal,New Jersey. Incidentally, I have a copy of thatat home, along with some other things, such asthe comments of a team that went to Europe afterthe end of the war.

Gene Visco: Was that connected with Alsos?That team was not associated with the AlsosProject, which was strictly nuclear and was ledby a Dutch physicist who could not be clearedfor the Manhattan Project. He had a team of mil-itary police, whose mission was to capture theFrench and German physicists they could find.That way, if there was a German nuclear wea-pons program, it would be stopped, so that therecouldn’t be a surge at the end. Those guys wereall incarcerated in a nice facility in England.There’s some interesting stuff connected withtheir reaction to the bomb—a separate part ofhistory.

Art Stein: There was another group, knownas the Lehigh Project, that was concerned withassessing bomb damage. It eventually becameknown as the Strategic Bombing Survey. Well,I went to work at BRL on October 2, 1941. Iwas assigned to a group called the MathematicsUnit.

Gene Visco: What were some of your earlyprojects or tasks? What do you consider yourmost interesting or important early work?

Art Stein: The principal work of the Math-ematics Unit was the generation of bombingtrajectories to make bombing tables. At thattime the bomb trajectories, where you hadlarge curvatures, were done by finite-differencemethods—quadratures or numerical integration.The study of finite differences had been re-stricted to actuaries and astronomers. Therewas no such course when I went to school.The only places they taught that, I believe, wereat the University of Michigan and the Univer-sity of California. The need for me, then, wasto study. The only really good book was by aFrenchman, Jordan, which had been translatedinto English.

Gene Visco: I took that course in 1949 atBoston University, and there was no textbookthen.

Art Stein: At any rate, about half of whatthey did in that unit was the start of these trajec-tories. That’s where they were the most difficultto apply. Once you got through the first 15 or so



lines—we’d have a little template, marked qua-drille pads, and Marchant or Monroe calculators—and you’d start this trajectory for a particularbomb and particular altitude and speed. Thesewere for standard conditions. At that time, ifyou had nonstandard conditions, you appliedcorrections using Galois equations to get thedifferential effects for winds, errors in altitude,tipping, air density variations, and yawing onrelease. That also turned out to be half of whatI did. Most of the people there worked full timeon starting these trajectories.

Gene Visco: What do you consider your mostimportant or interesting work in these earlyyears?

Art Stein: I got into somewhat more interest-ing work because I was new and did not havethe knowledge of all the various bombs and theirnumbers that others did from having worked onall these trajectories. About two weeks after I ar-rived, the man who was in charge of this unitgave an examination to the people who’d beencoming on board. In the examination he askedvery specific questions: For a specific bomb,whose nomenclature was Mark or M so-and-so,dropped from an aircraft travelling at a specifiedvelocity at a specified height, the trail of thebomb would be such-and-such. Or he wouldask what the range would be. I didn’t knowone bomb from any other bomb, so all I didwas calculate what the vacuum trajectory wouldbe. Because the answers were multiple choice,I always took a little bit less. I apparently got agood grade on his examination. As a result Iwas given some very interesting assignments.

Some of these assignments, other than whatturned out to be quite boring in starting thesevarious trajectories, were very interesting andmight have been called operations research, ifthat discipline existed. For example, an ord-nance officer at the Ordnance School, a lieuten-ant, had devised a method of firing a rapid-fire40 millimeter gun wherein the tube would movein a circle. He had a cam set for this so that hecould change the diameter of that circle thatthe muzzle would be moving in. The purposewas to induce dispersion. What caused his be-coming interested in this particular device wasthat, at that time, there was a great fear of an in-vasion by Germany. The war had started. It wasthought that there would be some sort of very

fast landing craft, doing a lot of maneuveringon the way in. If you tried to track them withthe linear tracking systems we had, you wouldhave very large dispersions. If you had veryhigh precision, you would precisely miss, as itwere. So he thought of this as a way to inducedispersion. The question was what is the proba-bility that you’ll hit one of these things? Andhow should you design the diameter as a func-tion of range and some subjective estimate ofhow much maneuvering was going on? I washanded the problem of determining the proba-bility of hitting with this device. From then on,I got various coverage-type problems relatedto bombing accuracy and effects, how manyyou wanted in a stick of bombs, how to straddletargets, and varieties of problems of that type.They were assigned to this Mathematics Unitat BRL. I even got to get my hands on things.We designed, and had the machine shop atBRL make, a copper plate with all kinds of mov-ing arms for doing vector additions to put in theeffects of winds aloft that a bomb would have.That would get corrections to determine thetrue ballistic wind. We made that gadget. Theyhad a problem calibrating the timing device todetermine the time of flight of the bomb. Theydid bomb testing to get the ballistic coefficientsof various types of bombs. They would dropa bomb and obtain a signal of when the bombleft. Then they would take pictures of wherethe splash area would be. These pictures weretaken with a Western Electric camera clock thatwas originally developed for horse races. It hadone lens on a rapidly-moving dial and the othersaw the scene. The clock started when the bombwas released, and the picture told you when thebomb landed. Meantime you had telemetry thathad altitude readings and so on. That clock,which was good to a thousandth of a second,was much more precise than the chronographthat had been used, which was good to a hun-dredth of a second. But every once in a whileit would be off by a second or two. So I was told,‘‘Fix it.’’ I couldn’t see, with the gears meshingthe way they were in that clock, why such an er-ror would occur. However, just to see whether itwould help, I remember putting in an electricbulb to dry out the entire mechanism. We said,a couple hours before we were ready to go, we’llturn on the bulb. We did that and it took care of


4 Military Operations Research, V19 N1 2014

the problems. But I still don’t really understandwhy. I’m merely saying we were given a varietyof tasks at that time; they went from analytical tohands-on work, such as with the ballistic windscalculator and doing something about cameraclocks.

Now something else happened in the firstweek that I was at BRL that would be of great in-terest, from the point of view of operations re-search. All that existed in the way of electronicmachines were accounting machines. IBM hadaccounting machines that were over in head-quarters. They were used for payroll and such.However, at BRL the man who was in chargeof that unit thought that it should be possibleto adapt these accounting machines to deal withthese finite differences and to do that job. Theywere merely a series of additions—actually, tak-ing differences, then taking them in another col-umn, and more. So he went and talked with theman who ran the accounting machines. Thisman was Mr. Gillespie. I’m not sure whetherhe ever finished high school. He was almosta caricature of the Tennessee mountain man—tall, lanky, chewing tobacco—but he had an in-sight into the operation of that machine thatwas really fantastic. They had written to IBM,when BRL wanted to know whether it couldbe used for this purpose of doing quadratures.IBM said, no—that was impossible with thosemachines. Mr. Gillespie wired up a machine.He took the board apart, despite instructionsnot to do this and that. Sure enough, he was do-ing it. The BRL unit had a lady over there whowas doing the roughing of the cards and actu-ally going through the operation. It was nowan ongoing operation, using the accountingmachines, to do some of the calculations. Thislady got sick; and, during my first week there,I was sent to continue the work. Gillespieshowed me what to do. About the third day Iwas there, a delegation came from IBM to seehow this thing was happening, despite the factthat their people told them it couldn’t. The menfrom IBM were all obviously VIPs (very impor-tant persons), because they all had stripedpants and homburgs [Editor’s note: a homburgis a formal felt hat]. There was a large number—must have been about a dozen of them. They allstood around while this man from Tennesseeshowed them how to wire the machine to get

these differences. At any rate, to my knowledge,outside of the normal calculating machines likethe Marchants and Monroes that were in use,that was the first time that the machines wereused. That preceded the high-speed computersat Aberdeen by several years. Before that, allof the calculations had been done manually ina room filled with girls working with large qua-drille pads—ruled paper. They still had to, be-cause the machine wasn’t that fast. They hadthese quadrille pads and had three shifts going.The night shift, because there was no air con-ditioning, had the windows open. Tiny gnatswould come in through the screens, fall on thequadrille pads, and look like decimal points.So, if any of the bombs went astray, blame iton the gnats! After I was there about a year,it turned out there was another problem. Onegroup had been doing surveillance testing ofammunition to see how the ammunition wouldstand up in storage and whether it was stillgood to use. At the initiative of Colonel Simon,that group teamed with a group from Bell Tele-phone Laboratory. The Bell people includedHarold Dodge and other people who had beenworking in the area of quality control. WalterShewhart was the chief of the group. The mis-sion was to generate sampling inspection tablesfor the very large increase in acquisition, pro-curement of munitions, and other things. Therewere no formal procedures for our inspectors,who were now going out to all the plants. To-ward the end of 1942, I transferred into thatgroup. For the next two or three years I was in-volved in developing these tables. As a result,when they had that seminar at Columbia Uni-versity and I could finally attend, my master’sthesis was on most-powerful double samplinginspection plans, because you could take differ-ent ratios in the two samples. We wanted to getthe best sampling plans to put into those tables.During that same time, Wald had started sequen-tial analysis, which gave the most-powerful mul-tiple sampling plans. Initially, you just took onesample, tested it, and then saw whether yourmoving mean went above or below these linesto accept or reject the lot of ammunition or what-ever you were sampling. I used that before it wasreally available and published. Because I wastalking to them, I got some of their initial notesand used it, since it was the optimum. It was



impractical, though, to take one at a time and goback for another. We were working in batches. Sowe were using group sequential plans. LaterWald actually wrote about group sequentialsampling as part of the book on sequential anal-ysis. Incidentally, Walter Shewhart was the fatherof statistical quality control. He was still at BellTelephone. They had developed double sam-pling inspection plans. Because they were withinthe same company, one section would do a de-tailed inspection of a rejected lot. Their outgoingqualities were very different from the govern-ment inspection, because they were involved inthis sort of screening operation. With the govern-ment inspection, if it was unacceptable, it wouldgo back to the manufacturer, who had to fix it,change the process or something, and resubmit.So both the average outgoing quality of thoseprocedures and the average amount of inspec-tion were different. The latter type was moretypical of the type of procurement the govern-ment would be doing. That became the basisfor the Ordnance Sampling Inspection Tables.Later, with slight modifications, they becameMIL-STD-105, which is still used all over theworld for attribute-type sampling. In additionto the work on this type of sampling, becausethis surveillance group had the statisticianswithin BRL, it also received other tasks duringthe war. For example, I became the fellow whohad to answer queries from the artillery. Thequestion would come in: Why don’t we getthe same results you see in the firing tables?There would be some kind of mismatch. Theywould describe what they were doing andthought they should get a result they weren’tgetting. I would have to prepare the responsesto that and give it to Dr. Theodore B. Sterne,head of the Terminal Ballistics Laboratory, orto Colonel Simon to send out. Usually those wereall quick responses. They needed it within a dayor two. From the South Pacific the type of ques-tion was often something like, how many shotsshould we expect to have to fire to breach a caveor bunker on the islands, realizing that we haveto get succeeding shots into the same hole? Noone impact would be sufficient to cause thebreach. That would have been a simple problemif each one were independent. Unfortunately,they weren’t. It appeared to us to be a much morecomplicated problem. All during that time at

BRL, in the early days of the war, there existeda very unusual, multidisciplinary growth in theattack on problems—very different from whatwas usual before. Simon had induced the emi-nent mathematician, Dr. Oswald Veblen, as his‘‘recruiter’’—one of two well-known brothers.Veblen, as the recruiter, obtained almost the en-tire University of California math faculty to cometo work at BRL. Edward McShane, who was thehead of the math department at Virginia, wasa famous man in integration theory. He had[Edwin Powell] Hubble, the astronomer, on theScientific Advisory Committee. Part time, hehad [John] von Neumann, who was there twodays a week. [Isador Isaac] Rabi, the physicist;[George] Kistiakowski; Dr. [Bernard] Lewis fromthe Bureau of Mines. It was quite an assemblage.Many of these people did what I would todaycall operations research, but it wasn’t called thatthen, at all. They all had problems within theirspheres that were types of optimization prob-lems, although they were dealing with veryhot, current problems. For example, in abouttwo hours, von Neumann was given and solvedthe problem of how to correct for spin of a shapedcharge: What was the optimum configuration fora liner to compensate for spin? The cross sectionwas the spiral of Archimedes. It would compen-sate, because a spinning projectile normallydoesn’t give you the same kind of penetrationthat a nonspinning projectile would with ashaped charge. The jet is moving around ina spinning projectile, instead of staying in oneplace. It was fantastic to see these minds atwork. During that same time he was very muchinvolved with the concepts for the originalcomputer—the ENIAC. I didn’t know it thenthat he was spending the rest of his time on de-veloping the atomic bomb. Many of these menwho came and went were contributing in dif-ferent ways. At BRL there were problems in op-timizing fragmentation of projectiles: Whatkind of fragments should you have? What sizes?What kinds of errors could you stand on timefuzes? I worked on the variations you wouldget from a powder-train time fuze: What werethey due to, and how could you get rid of them?That was very interesting. There was a lot of thattype of work. BRL then was very small. In factmy badge number, when I got there, was 78.Many of the people with lower numbers had left.



So, initially, it was quite small and started in oneof the wings of the administration building.It was called the Ballistics Section. All thatgrowth took place in the early years of the war.They got an air-conditioned addition to thebuilding by claiming that the Bush differentialanalyzer required conditioned air. They wereable to get the whole addition that they builtair-conditioned. I should correct something Isaid earlier. I said the only machines beingused were the calculators—the Marchants andMonroes. For artillery projectiles, where youhad these simple and relatively flat types of tra-jectories, they could make use of Sciacchi equa-tions. You didn’t have to use the numericalmethods that were used for the bomb trajecto-ries. They made use of the Bush differential an-alyzer, which was a nonelectronic machine. Itwould occupy an area about 40 feet by 70 or80 feet. It was about waist high, with long rodsand gears at various stations. People wouldstand at these stations. As the calculations weregoing on, somebody would read off a time offlight; somebody else was getting the range.They would take it off at various times, thenthey would match up these different readingsand do interpolations to get them all for thesame times. That was used for artillery, flatter-trajectory firing. The name came from VannevarBush, who had one of these at Harvard. This wasthe second one.

One of the reasons I mention the activities ofthe academics at BRL was that their generalbehavior was very focused on the solving ofapplied problems. Many of these involved op-timization of procedures. To my knowledge,there had been no work on the kinds of opera-tions research that dealt with large forces. With[Dr. Oskar] Morgenstern, von Neumann hadpublished a book on the theory of games andeconomic behavior. We very much wanted totalk to him about things that he had in the book,because the book was not applied to military-type problems. We wanted to see how we couldadapt his game theory to military problems.But it was difficult at work to get him to do that.Von Neumann was the type of fellow whonever wanted to talk about what he had done.He felt somebody else could come along and sortof mop up. He was always interested in talkingabout something new. However, we attended

the meetings of the Scientific Advisory Commit-tee, which met in the evening. Preceding dinnerthere would be a cocktail party, and he lovedManhattans. So those of us who were interestedin the theory of games would gather aroundhim. One fellow would come back with twoManhattans. He’d take one and give one tovon Neumann: ‘‘Dr. von Neumann, a toast!’’Von Neumann would enjoy a Manhattan. Thenanother. I must admit we did this with maliceaforethought. Soon he would talk about his book.Those were the only times we could really get hisattention to those types of problems. Incidentally,this was the BRL Scientific Advisory Committee.It had a number of Nobel Prize winners on it andwas very powerful.

Gene Visco: Do you know how and whereColonel Simon got his training?

Art Stein: I don’t know where he got histraining after West Point. It would be interestingto see. One man who might know is Dr. FrankGrubbs, whom I was going to suggest you con-tact. He’s one of the very few people I knowwho’s still technically active and who wasaround during that period. He was an officer—acaptain, then a major for most of the time I wasthere. Dr. Grubbs is a mathematical statisticianwho authored many BRL reports and later theArmy Engineering manual on systems analysis.I think he still lives in the vicinity of BRL. An-other man now works on the West Coast; hisname is Dr. Alexander Charters. He was in-volved with the spark photography range forgetting drag coefficients and behavior of projec-tiles, and with many of the scaling studies thatwere done. Dr. von Neumann would consultalso for some of that work. My wife actuallywas one of the computers. In those days peoplewere called computers. My wife had been a jour-nalism major and we were married in Decemberafter I started to work at Aberdeen. She camedown to Aberdeen and after three months orso she was stir crazy and wanted to get a job.She thought she could work on the post paper.She went into the civilian personnel office there.They said, ‘‘You’re a college graduate? Did youhave college math?’’ She said, ‘‘All I had wasthis one required course.’’ He said, ‘‘But it wascalled College Math?’’ She said, ‘‘Yes.’’ ‘‘Well,then, you should go to BRL.’’ So he sent her toBRL. Dr. Louis S. Dederick, the associate director,



said, ‘‘You know, you really haven’t had muchmathematics, have you?’’ She said, ‘‘Well, that’swhat I tried to tell them. But they said I shouldcome here because I had this course calledCollege Math.’’ Dr. Dederick looked at herand her paperwork. Then he said, ‘‘Stein? Areyou related to Arthur Stein?’’ She said, ‘‘Yes,he’s my husband.’’ ‘‘Oh, that’s all right, then.He can help you at night!’’ So that’s how mywife became a computer. She first worked onthe second parts of those bombing tables. Therewere a very large number of people, all calledcomputers, doing the rest of the bombing ta-bles. They were actually putting them together.Later she was loaned out to assist various otherpeople. When von Neumann came, he used toask for her to help him. She would tell me whata real gentleman he was.

Gene Visco: Would Chester Clark be the per-son whose name you couldn’t recall before? Youhad given me some names earlier.

Art Stein: No, but Chester Clark would beanother person who would be good to speakwith. He was the executive officer. I think he’sstill alive. Mr. Robert (Bob) Kent was an associ-ate director of BRL and head of the Exterior Bal-listics Laboratory. He served as the technicaldirector. Kent worked in many of the exteriorballistics and artillery fragmentation problems,among others. He was a brilliant man. FloydHill came there around the end of the war, inearly 1946. He could tell you some of the stufffrom a different viewpoint. Floyd precededDave Hardison, looking at tank problems. Floydwas in the Proving Ground Detachment withme. When he got into it, I’m not sure. He’s a veryclose friend, lives in this area, and is currentlya consultant at IDA. Walter Hollis, FS, knowshim because Floyd used to be at the Army’sOTEA [Operational Test and Evaluation Agency].At any rate, one of the reasons I’ve spouted per-haps too long about these men, rather than whatthey may have done specifically in operationsresearch, is that I think it would be very desir-able and useful for these archives to obtain atleast two different kinds of things. One, the bib-liography of BRL reports. That would triggerthe memories of the people with whom you’dspeak, including myself, on some of the activi-ties and those activities that might be consid-ered pre-OR. Two, the correspondence files on

the different projects. There were many lettersand reports that came in from the field. Therehad to be responses to these. In those days, un-like the Navy, I don’t know of any people whowere with the troops doing operations researchor who were stationed at various commands do-ing that. At least I don’t know of that being donein the Army. The Navy did, and the Army AirForce had something going on in England. Wewould later receive large three-view drawingsof a B-17 (Boeing B-17 Flying Fortress) withholes marked, where they came back with frag-ment holes. From that we would try indirectlyto deduce that those areas where there wereno holes were more vulnerable, so those planesdidn’t come back. So, in addition to seeing whatthe distribution might be over an aircraft, wetried this type of indirect estimation of vulnera-bilities of aircraft, by looking at the damage tothe returned planes. I didn’t have any opportu-nity after the war to verify the basic premise.I never thought of myself as doing operationsresearch then or of how broad the term wouldbecome. I remember when the journal (Journalof the Operations Research Society of America[JORSA]) first came out. The early practitionersseemed mostly to be mathematical statisticiansand a few physical scientists. Some of the earlypapers had to do with discrete distributionsand Poissons and binomials, and various typesof applications of this sort. The early journalhad a much narrower focus than was true later.I would guess it was about 1946 before thephrase, operations research, first came to myattention. It was used earlier. There was a ques-tion, are you doing operations analysis or oper-ations research? That was kicked around. WhenI get into the last part, that deals with the wholequestion of aircraft vulnerability, optimizationof when you should open or close fire, air-to-air duels and the things that I was talking toShephard [Professor Ronnie Shephard] abouthaving gone on in England. I don’t want to getinto that because it’ll take a little longer. In theseother areas, until I got into that aircraft vulnera-bility bit, I personally was involved primarilywith ballistics, and sampling inspection andquality control. By virtue of some of the earlierwork, I was also asked to look into malfunc-tions because they were involved with testsand sampling. Something would malfunction



and I would have to go see why it was happen-ing. What kind of a test could we run to findout? I became familiar, particularly with fuzes;they were the most amenable to malfunction orat least to give results that were somewhat un-expected at times. Later we were visited, oncea year, by classes from West Point. There wouldbe a series of lectures on different subjects,and I was asked to talk about ammunitionmalfunctions—what kinds of things could ormight happen. That was a kind of test-designwork. We set up experiments and tests to deter-mine what the causes would be. I think youwould still call that statistics, rather than oper-ations research, but I don’t know. It is very im-portant that we obtain the reports of HerbertK. Weiss, who came to BRL toward the end ofthe war—about 1945 or 1946. I would say hewas the principal man doing operations researchat BRL. I believe he left BRL about 1953 or 1954 togo to Northrop. The last I knew he lived in PalosVerdes, south of Los Angeles. Prior to coming toBRL he worked at the Army’s Air Defense Schoolat Fort Bliss, Texas. He has an antiaircraft gunsight, the Weiss Sight, named after him. I knewhim to be brilliant, imaginative, and a prolificwriter on many diverse military OR topics.Dr. Paul Deitz, head of the Vulnerability Labo-ratory at BRL, has promised to send me a listingof Weiss’s BRL reports. If he is still alive, hewould be an excellent man to interview. FloydHill may have his address and phone numbers.

Jim Williams: Art, what is your perspectivein this oral history?

Art Stein: What I have been describing is ei-ther work that I did or was involved with, or myperceptions of work being done and in which Ihad some interest and awareness. However,back then there were Black Programs, just asthere are today, and I was quite junior in the or-ganization. As a result, there’s much I didn’tknow about. In addition, there were some sec-tions that I just never interacted with, such as In-terior Ballistics. We didn’t do very much withthem during the time I was there. I had muchmore to do with that area later on. On the otherhand, I think that there is very little of what wemight term operations research or operations re-search-related activities going on in a section orbranch such as Interior Ballistics. They startedout as sections and later as the staff grew at

BRL, they were made branches and later in factwere called laboratories all on their own. Well,what I have related so far essentially concernedthe work in ballistics, per se. I was originallywith a mathematics unit and some of the prob-lems I described were during that first year atBRL when I was with that unit. Then I trans-ferred into the surveillance unit, which wasmuch more statistical in nature and that’s wherea good deal of the sampling inspection, qualityassurance, and ammunition malfunction workwas done.

I went into the Army in 1944 from that sec-tion and returned as a private. I believe I havedescribed some of that. Initially when I returnedit was back to the same desk. I think I mentionedthat and I had a number of WACs (Women’sArmy Corps) and even a couple of lieutenantswho worked for me during the day; but beforewe started work and after we finished, thenwe reversed roles.

The next thing I did was go to a ballisticsteam known as the T-6 Chronograph teams.They were developed in 12-man squads to cali-brate artillery weapons in the field. We usedDoppler radar, but the entire operation wasn’tthat well mechanized. It was necessary to havea mathematician on the team, because the radarwas not located right behind the gun itself orclose to it, but was offset. Therefore, you hadto make corrections for the amount of offsetwhen you were determining muzzle velocity.

You couldn’t take chronograph screens asyou would use at a Proving Ground where youfire through an initial screen and then a laterscreen in order to get velocity and then workback to the muzzle. These were Doppler radars,very mobile, and you could use them to calibrateguns in actual combat.

We had three port calls to go to various the-aters and each time they were canceled becauseneeds changed. The war was starting to winddown. At the end there wasn’t very much forthis team to do. We had finished putting every-thing back in Cosmoline [Editor’s note: Cosmolineis a rust preventative] again and actually we weretrying to find useful things to do.

It was just at that time—in late 1945—thata letter from the Office of the Chief of Ordnancecame to the Development and Proof Servicesat the Proving Ground. It said there were 20 to



30 obsolete aircraft there at the Proving Ground,and it was desired to fire a very large numberof types of small arms and small caliber ma-chine gun types of ammunition at them at dif-ferent velocities and angles. The types wouldinclude ball or shot, high explosive, high explo-sive incendiary, armored piercing, armoredpiercing incendiary. The purpose was to deter-mine which caliber would be optimum for fir-ing against the aircraft and to obtain somenotion of the effects.

It used to be that people would be mostlyconcerned with the probability of hit and oncethe projectile or a round hit whatever it wasdesigned to or supposed to hit, it was assumedit would do the job. This was true earlier with ourtank munitions and also with the anti-aircraftmunitions. But obviously this wasn’t true now.This program was requested by the Chief ofOrdnance. Well, there was some concern. Thishad not been done before. There was some con-cern on the part of Colonel Carr at the Develop-ment and Proof Services as to whether he couldactually accomplish all the different tasks andangles and types with so few aircraft there. Theywanted to do tests with gasoline in tanks and ina sense this was a forerunner of the current livefire testing that was asked for by Congress fornew types of aircraft.

Actually it was Joe Sperrazza and anotherchief Proof officer working for Colonel Carr atthat time in Development and Proof Serviceswho were going to conduct the testing. Theyasked me if I would be able to help them by ex-amining the statistical meaningfulness of doingdifferent shots under the various conditions.They had to get a reply quickly. I really didn’thave more than a day to look at that problem.

At that time, as I may have mentioned, I wasa staff sergeant. The problem was one thatrequired some assessment of the likelihood ofdifferent types of results and an assumption re-garding what might be a permissible error in theanswers. I did some calculations and madesome estimates of what kinds of variations theremight be among a group of projectile types. Onthat basis it appeared to me that if we wanted toget the answer to the questions raised by theChief of Ordnance within 10 percent, we wouldneed more than three or four thousand aircraft.Assumptions would have to be made regarding

how many shots we might be able to take againstan airplane, even with fixing it up, before wewould no longer have it to use.

But if you wanted to get answers within 20percent, you would need something like 1,200airplanes. That’s what I told them. I neverexpected that anything like that would happen,but they asked me to look at that problem andthat’s what I did and those are the answers I got.

Well, I guess before Colonel Carr sent hisletter back to the Chief of Ordnance with thosenumbers in it, he called around a couple of pla-ces and saw nothing that suggested there wasanything wrong with what I had written. So hesent it in.

At this point we switch to the Office of theChief of Ordnance and the letter is on theChief’s desk. He’s first seeing it. I knew someof the people there. Later I got to know themquite well, one woman real well. She was a very,very fine analyst in her own right. Melvin Millerwas there and they all told me that when theGeneral saw this letter he just about hit the ceil-ing [Laughter] and he put a call into the ProvingGround to Colonel Carr. He wanted a conferencecall and he wanted Colonel Simon from BRL,who was a scientist in the Army Officer Corpsat that time, and the Director of the BRL to bein on the call.

Colonel Simon came to Colonel Carr’s officeand there was a group of people in ColonelCarr’s office when the call came in. I was sittingthere as well. The General asked Colonel Carr,‘‘And who made the analysis? Who came upwith this number of 1,200 airplanes?’’ ColonelCarr said, ‘‘Why Sergeant Stein, sir.’’ [Laughter]

Then he wanted to speak to Colonel Simon.The General said, ‘‘Les, what do you think aboutthat estimate? Do you know who this SergeantStein is?’’ And Colonel Simon—I loved him for-ever afterward—said, ‘‘Sir, if Sergeant Stein saysyou need 1,200 aircraft, that’s how many aircraftyou need.’’ [Laughter]

Jim Williams: Do you remember the name ofthe Chief of Ordnance at that time?

Art Stein: No, I probably could dig it upsomewhere. Believe it or not, they started ship-ping in 1,200 airplanes. They all came flying intoAberdeen Proving Ground and that was thestart of the largest systematic investigation ofaircraft vulnerability. The cost of the testing



was provided by the air people, not by ord-nance, because they wanted to know how tomake aircraft less vulnerable. The ordnancepeople wanted to know how to make better mu-nitions for shooting against aircraft. So therewere those two interests.

I believe the amount of money provided bythe Air Force was on the order of two milliondollars a year for that program. We had abouteight ranges going at the same time and we alsohad quite a number of supplementary tests.

Let me say that very shortly after this pro-gram started we scrapped that original plan oftesting all these many different types becausewe knew that if we didn’t get a particular typeof damage with a larger and faster projectile,we weren’t apt to get it with a slower one. Westarted to do our test designs incorporating se-quential types of subplans, you might say.

In addition, we felt that we could get a greatdeal of information from the aircraft withoutinteracting with the loaded fuel tank. If we firstdownloaded the fuel and did some structurestesting, we looked at component damage forpower plants, fuel, fuel tanks, probability of get-ting fires, personnel, damage, casualties, andstructures.

When I left the Army I came back to BRL,because I had been a civilian there to begin with.And the program I was responsible for was theaircraft vulnerability program.

Gene Visco: Did you mention the date whenyou made that estimate?

Art Stein: The actual date?Gene Visco: The approximate year.Art Stein: 1945.Now organizationally this program then

was transferred away from Development andProof Services to BRL. The responsibility wasput there and I was responsible for that pro-gram. It was embedded within the TerminalBallistics Laboratory headed by Dr. Ted Stern.

Gene Visco: I had forgotten about Ted.Art Stein: He passed away.I was put together with a new employee at

BRL for whom I worked directly, namely HerbWeiss. That’s when I first met him. Our originalstaff consisted of the two of us, and a computer.

Gene Visco: This was a human computer?Art Stein: A human computer person who

was Francis Hill—Frankie Hill—who was Floyd

Hill’s wife. Floyd Hill was also an enlistedman in the BRL Detachment. And a LieutenantBritton.

It wasn’t long before the requirements of thestaff grew so large so that we had to make signif-icant additions. Now the test planning was doneby BRL. The test execution was by Developmentand Proof Services. So there would be a Proof of-ficer. They would get the gun crews, the ammu-nition together and the like. The test analysisand test report were done at BRL.

Each time there was a test conducted by De-velopment and Proof Services there would bea firing record written by them that was in con-sonance with their procedures for all other typesof tests.

These firing records constituted the rawdata that we had to work with for vulnerability.We were starting in this area with a blank pieceof paper. We appreciated that there would bedifferent types of damage, immediate kills somewould be interested in, attrition kills to enemyaircraft, types of kills that would interfere withmissions but not necessarily be killed, types ofdamage that might cause extensive requirementsfor repair before the aircraft could be used againalthough they weren’t actually killed, and thelike.

We had 5-minute kills to represent relativelyquick damage that might prevent a bomber fromreleasing on target at the time it was intercepted.These were notions we had for organizing the de-scription of the damage. They were not entirelyindependently generated by us. We, I think,had a big part but there also had been other workgoing on for other reasons at the New MexicoSchool of Mining and Technology located at thattime in Albuquerque, New Mexico.

They were operating under a contract forthe Navy to examine the effectiveness of anti-aircraft shells. I believe they were 5 inch 38s [Ed-itor’s note: this was the Mark 12 5"/38 caliber gun]with Variable Time (VT) fuzes. To do that theywould raise aircraft between two towers sus-pended by cable, fire the rounds, have their air-bursts, and quite precisely map the locations ofbursts around the airplane. That was their primepurpose, to investigate the VT fuze functioning.

However, and quite incidentally, when theshell detonated, as a secondary objective theysaw where the fragments went. They wanted



to know whether you would get hits or not withthe various locations of bursts and at the differ-ent remaining velocities that these projectileshad at the time they burst.

They actually then recorded where on theairplanes the fragments would hit and theywould divide the hits as to whether they werevery serious or critical types of hits or not so crit-ical. In other words, they had already started tothink about some procedure for this.

We were quite interested in finding outwhat they were doing and to borrow or usewhatever concepts they had that would applyand be useful for us.

Before I continue with what the effect of ourvisit there was, I should describe how we didour assessments because we wanted our dam-age assessors to go to the New Mexico Instituteand see how they were doing damage assess-ment there and what could be borrowed.

We had Naval and Air Force officersassigned to the Proving Ground who did thedamage assessments for us. They would seethe damage, which cables were cut, what hap-pened to an engine and the like, and we wouldhave some assumptions regarding the scenariofor the combat aircraft. They would then maketheir assessment as to what the damage signi-fied and they would write it up. They alsotraced the damage, and described what it wasthat was cut. They traced the path and wroteperhaps a half page on the average descriptionof what was observed.

They would then categorize the damages towhether it was an immediate kill, a 5-minutekill, attrition kill, requiring air type. However,I appreciated the fact that damage to that air-plane, in that scenario, either was a kill or itwas not. It wasn’t a probabilistic matter eitherway.

Nevertheless, if I insisted that they tell mewhether it was a kill or not, yes or no, and theyhad entertained some doubts as to whether itwas a kill, I might be forcing a very large errorto get into the assessment procedure.

I had a system set up whereby they woulduse fractional numbers as well, reflecting the de-gree of confidence they had—how they leaned.They didn’t have to say they were certain thiswas a kill, which would be 100 percent, but theycould say 80 percent, which meant that they

were fairly sure it was a kill, but it might notbe. They weren’t certain, or 10 percent, for someparticular damage they observed.

The underlying assumption for these frac-tional kills was that the expectation over manysuch assessments would be the expected num-ber of kills you would have had if indeed someall-knowing person had put 1s and 0s. That’show that assessment procedure was establishedthroughout all of the firing tests against the air-craft vulnerability program.

Jim Williams: You were talking about how itwas that you came to be able to get essentiallya pretty high confidence level in the reliabilityof different types of test results without forcingpeople into making the categorical distinctionsbetween kills and non-kills.

Art Stein: Yes. Now our principle interestwas in the damage itself, true enough. However,we were working with an assessment plan whichwould require that we put all of our informationtogether and obtain a probability of kill on thatairplane — estimate of probability of kill basedon the empirical results and assessments.

To that end we had to divide the types of fir-ings we did into two types; one was the randomhits where we could do that collection—randomhits over the presented area of the componentsthat were being examined. But then we had se-lected tests where we were attempting to deter-mine the causes of damage and the means thatwould enable us to generalize from the damagewe observed to other aircraft. That becamea very important issue later.

We had some problems obtaining randomhits on aircraft, for example, shortly after wefirst started shooting against engines, whichmight have a presented area of about six to eightsquare feet. I observed that just before lunch orquitting time the engine would be killed and Ifelt this was happening more often than onewould expect just by random hits over the pre-sented area of the engine.

I went down to observe the procedures thatthe gunner was using and, as we had gonethrough previously, he would correct the wind-age and then he would do a number of shots to-wards the 1,500 yard range or whatever and quitefar in order for us to get the proper slow down.

But before lunch is when he recalibrated,taking extra care, and used his Mann barrel



(which is putting the barrel in a V-slot cradleand is used for very accurate shooting as com-pared with conventional combat). He was ableto place the round, for example, right into aP-47 (Republic P-47 Thunderbolt) oil cooler.The inlet of that would be something like a6-inch diameter circle and he could do thatfrom considerable range. That would cause oilstarvation and freezing of the engine and weknew that by then, but it always happened beforelunch or before quitting time.

We had to devise something to correct andget around that problem. So we took essentiallydiagrams of the presented area of the engine atthe angles that were being employed, the off an-gles in the particular series of shots, and we di-vided that presented area into boxes of areasequivalent to that of this oil cooler, and we la-beled each box A, B, C, D, and so on. Then froma table of random numbers we got the orderwith which we were going to challenge the gun-ner to shoot at this engine.

Now he might shoot H first, then B, et cetera,and he actually aimed and we weren’t depend-ing on accidental dispersion now at these farranges. But he aimed to get the round into thepresented area as on the diagram he had, thatpart of the engine and the order in which wehad picked from a table of random numbersand that was the challenge. We complimentedhim on his ability to do that throughout the seriesand we got over this problem of the biasing re-sults before quitting and lunch.

Jim Williams: Did you ever find out what itwas that motivated the extra care at that partic-ular point in time, right before lunch and quit-ting time?

Art Stein: Oh, they didn’t want to drag over.You couldn’t stop the task and let things hang.Suppose he started a fire? People would haveto go down and they’d have to assess it andput it out. I mean there’s a lot of work to bedone. So that nothing interfered with the order-liness of life, if you went and finished it off andkilled it, that was it. You could make things a lotneater.

There are many small things of that kind thatarose and we were learning as we went alonghow to properly conduct the tests.

The assessors, as I mentioned, were Navaland Air Force officers, and I went with them

to Albuquerque to speak with the people there.The people we met were extremely capable andeven though they were concentrating on a some-what different problem than the one we werefacing, they were very, very helpful and afterthat we were able to continue to cooperate agreat deal. We had established a good relation-ship with them. It was an experience for the as-sessors and for me to see an experimentalprogram somewhat like ours that had been go-ing on for a period of perhaps two to three yearsand where some of the procedures had maturedand they had also observed some of the thingswe had started to see.

For example, we also did tests not just firingthese rounds that the Office of the Chief of Ord-nance had talked about, but we wanted to getinformation that was important for air bursts,anti-aircraft shells, and missile warheads.

To do this for our purposes we developeda controlled fragmentation shell. We had nineshells, three different initial velocities by con-trolling the charge-to-metal ratios with theseexperimental warheads, and three different frag-ment sizes. These were obtained by notching thewarhead.

The men responsible for the design of thesecontrolled fragmentation shells to give us theinitial velocities and fragmentations that wewanted were Mr. Shaw and Noah Tolch. Theywere both very capable and fine people andwe worked quite well together with them.

They were able to obtain very good control.The shell was made in the machine shop at BRLand when we fired them, we fired them stati-cally. They would sit in the cradle and we’d det-onate them. They might either be horizontalor vertical depending upon what we had in anarena. We used an arena with several aircraftwith engines running. We had components outon fuel tanks on stands. We had collected boxesto show us where the fragments were that werecoming off from these warheads. Each warheaddetonation had done a tremendous amount ofdamage.

One of the things that we observed was thatwe had masses or fragment sizes up to the orderof something like three-eighths inch square—Ithink 700 grains was the largest mass we had.We had about 8,000 feet per second. That wouldbe a very large test of warheads.



In thousands of such fragment impacts, wenever obtained a kill that was due to structuraldamage alone. That was a rather robust findingwe felt. So that there’s no misunderstanding,that did not include damage in control cablesor any portion of the control system, per se.

Jim Williams: You’re talking about the basicairframe.

Art Stein: The basic airframe and skin.There were many other interesting factors.

We found that you could put an average of 2050-caliber projectiles into the R-2800 piston en-gine, which was on the P-47, on the B-26 (MartinB-26 Marauder) light bomber, and on theNavy’s F-4, before you’d stop the input of theengine. These were extremely robust and in factthe chief way you obtained the kill at all was bya shot into the oil cooler. To know what thatwould signify, we ran off-line tests where weshut off the oil to an engine and found that itwould run about 15 minutes before it froze.We found in our firing tests that this kept hap-pening with not that much dispersion aroundthe 15-minute freeze point.

We did the same side tests with hydraulicsand we got to understand that an in-line liquid-cooled airplane like the P-51 (North AmericanAviation P-51 Mustang) was extremely vulnera-ble compared to the P-47. The P-51’s enginewas up there in the front where it had two circu-latory systems. It had the oil and the hydraulicsand a hole in either one would cause a kill, an at-trition kill on that airplane. And because theengine is jacketed, you couldn’t miss them ifyou hit that anywhere in the presented areaof the engine.

The P-51 may have been more maneuver-able and had an advantage that way, but theP-47 certainly was a lot more robust. It was themost robust airplane I’ve seen. When you lookat the presented area and what you could hitwith just those two small circles for the oil coolerunderneath and the top of the pilot’s head if hewas coming ahead.

The early versions didn’t even have any fuelin the wings. The fuel was behind the engine inan L-shaped tank coming down and then belowthe pilot. All in all it was a tough airplane, butthose were all findings that we would neverhave expected. As far as hits with high explo-sive shells were concerned, we did that. We

cut off sequentially different amounts of pro-pellers to see when we would get sufficient im-balance for there to be kills.

We detonated small blocks of explosives totake out cylinders and in different sequenceson the engine to see what they would signify.When you had a running engine and you puta bullet into it, it seemed to be as though it werealive. You’d hear the roar of the engine and thenit seemed to grind out any pebbling or damagethat the bullet may have caused. It would havegrounded that cylinder that might be ineffec-tive, but it sounded as though the engine spitthe projectile back out again when you listenedto it.

We measured the ability and the power ofthe engine by pulling a dynamometer that wehad in the airplanes. We did a great many inves-tigations offline to obtain incendiary function-ing distributions for probabilities of fire to seewhat the delay times were for fuel spray to comeout from tanks after they were hit. We even didsome tests just before I left Aberdeen in thestratosphere chamber just to see whether wecould get any ignition at extremely high alti-tudes and we did, although the kind of fire weobtained was very different from anythingI had seen. It was an orange transparent fire,very thin, which made me think that we shouldhave been doing an altitude chamber examina-tion to get a better feel for what could occur ataltitude.

Unfortunately, they still haven’t done thattoday after all these years. All in all we were,as I said, learning as we went along. The bigproblem, of course, was we were going to knowa great deal about these obsolete aircraft fromWorld War II. How does that help us? That wasthe subject of an inquiry. A small board wasset up by the Air Force to see whether the$2 million per year they were spending atAberdeen was worthwhile. That question wasjust one of the questions raised then. How doesthis shooting against obsolete airplanes helpyou for the future?

The members of the committee that came toBRL were Colonel Charles Lindbergh, JimmyDoolittle, and a Navy officer whose name I can’tremember.

Jim Williams: Was that committee givena particular name?



Art Stein: I don’t know. Lindbergh was themost outspoken of the three. He asked verysharp, good questions. We got to this principalquestion of how can you estimate new airplanesthat we were shooting at.

This was essentially what we did and how Idescribed it to them. We didn’t look at aircraftthe way they did. To us, the aircraft was a conve-nient collection of components. Some of thecomponents of the new aircraft will still exist, al-though perhaps in different sizes, but physicallycomparable to the components in these olderairplanes. Some of the skin or particular typeof aluminum may not be in the same place onthe new airplane, but how it affects the function-ing of the ignition of the fuze and of the incendi-ary, we learned from the older ones.

Aside from the assessment of the kill ofthose older airplanes themselves, the descrip-tions of the damage of what happened is whatwas so important—what was going on physi-cally. I looked at the airplane as having perhaps50 red bricks and 200 brown bricks where thebricks represented different componentry ormaterial and the numbers represent their rela-tive size.

The new airplane doesn’t have 50 redbricks, but it has 20 red bricks. It has 600 brownbricks, but also has some purple bricks that theold one didn’t have. Now we have to see whatfraction of the total presented area is repre-sented by purple bricks and what differencethey would make. In other words, the physicalthings that we don’t understand, that wecouldn’t predict, what difference would theymake in the overall assessment? If the differencewas not large, we would estimate it as best wecould from what we knew physically. But if itcould be significant and important and if wewere looking to say something about one ofour own airplane’s vulnerability, I would gener-ally go visit the chief designer. Say on a new jetengine, we didn’t have any modern jet enginesto shoot at. But we had these World War II air-planes. We had some P-59s (Bell P-59 Airaco-met, the first American jet fighter aircraft) withcentrifugal flow engines. But we had no actualflow jet engines left to shoot at. I went to thechief designer for these engines. They were verycooperative and we’d have detailed drawingsof the engine and I would start in the front

and I’d ask, ‘‘Well, what if we had a pin holehere? What would happen?’’

He would say ‘‘Oh, nothing.’’ I said, ‘‘Well,suppose it were a half inch hole?’’ Well, hedidn’t know. I said, ‘‘Suppose it were a threeinch hole.’’ ‘‘Oh, it would be terrible.’’ I wouldclose in from the two ends to try to find wherethe zone of mixed response might be on his part.Where was he certain and sure and where washe not?

Then I would proceed backward throughthe engine, station by station, looking for holes,chips off of particular turbines, holes in the tail-pipe, and obtain his best overview of what thesignificance would be.

Now we knew from tests on the materialshow big the holes were that we had made, butwe didn’t know what the significance of mak-ing them would be for the new components.In some cases, we didn’t know what the holesor damage would be if the material was quitedifferent. For example, we had metalite typeof control surfaces. Metalite was really a typeof plastic. We did these tests involving air-craft that were sitting on the ground, but in ac-tual fact they’d be in an air stream if they wereflying. One question was, ‘‘How much differ-ence would that make?’’ We would cut notchesin ring materials and put them in a wind tunnelto see whether we would get any increasein damage. We found that the raised petal ma-terial would flutter and then break where itwas bent and wouldn’t cause any increase indamage.

But we weren’t sure that would happenin metalite, and indeed, metalite did rip andit was much worse when you did that kind oftesting.

We had to rely on many supplementarytests to obtain this information. It’s interestingthat today in terms of live-fire testing programsthat various models have been set up for predic-tions. The important thing with respect to themodeling of the damage that we learned wasthat we could not have a universal model whichwould exist for all the new aircraft to come andmerely require that we modify geometry anddeal with the damage estimation that way. Theintroduction of new materials, new control sys-tems, much larger power plants all meant thatthere had to be a continuing surveillance and



vigilance that the models would be current forthe new types of aircraft emerging.

It was for that reason that we wanted to em-phasize obtaining an understanding of the phe-nomenology of damage. For example, given thedistribution of length and duration of incendi-ary flash on hitting skin at different obliquitiesand different thicknesses; then coupling thatwith the delay times for fuel sprays to emergefrom a fuel tank that was hit either below the liq-uid level in the tank or in some cases from aboveand going through. We determined that wherewe had an intersection of the flash and theemerging spray we would have ignition andfire, but where the distances might be largeand the flash located very close to the skin, itmight not be available any longer. It would havegone out by the time the fuel spray reached it orthe fuel air mixture that is caused by the fuelspray would reach it.

Those were principles that enabled us to goto a different type of aircraft and take what dif-ferences existed for it, but apply the principleswe had learned regarding ignition and be ableto estimate what we would expect to get on thisother airplane.

At all times we wanted to obtain the feelingfor the physical principles involved that wecould use in order to make those predictions.We saw that it would be very dangerous justto rely on the empirical data themselves.

Now a number of things would occur.Sometimes we didn’t have all the aircraft onhand when we wanted to do the firing andwe had some close calls as to being able to con-tinue our efforts. We were engaged in theseR-2800 engine tests, and as I mentioned, they’reon a number of different aircraft. We had essen-tially expended all that we had there at theProving Ground, but we couldn’t completethe tests without the Navy F-4s arriving andthey were due to come in, but they hadn’t ar-rived. I remember we were going to have tolet go of the gun crews and the teams we hadassembled to work together and it was a matterof great concern. The planes would be arrivingat Phillips Field. I asked the man at the tower,‘‘Let me know if any of those F-4s come in to-day because then we can hold the group to-gether.’’ I told him that the Proof officer wouldset it up at whatever the attitude was to be such

as 13 degrees elevation, 20 degrees off an azi-muth, for example.

Sure enough, late in the morning we get thecall that an F-4 is coming in and I tell him to callthe Proof officer and get it down there at thatparticular range. I forgot about it and went backto my other work.

Well, later in the afternoon there was a call forme from the tower. ‘‘Are you Mr. Stein?’’ ‘‘Yes.’’Well, the fellow on the other side sounded just likeDonald Duck. I mean he was mad. This was aNavy liaison pilot who had flown that F-4 inand he went over to the administration buildingand was doing his business and when he cameback to get his airplane it was down at the endof this range sitting 20 degrees up, 13 degrees el-evation. [Laughter] Luckily we hadn’t started toshoot at it yet, but he was one mad fellow.

Gene Visco: Art, just a quick question. Itseems to me at this time there were two differentF-4s. Was this an F-4U (Vought F-4U Corsair) oran F-4F (Grumman F-4F Wildcat)?

Art Stein: F-4F.Gene Visco: Do you remember if the wings

were straight and not gull?Art Stein: Yes. And it had the same Pratt and

Whitney R-2800 engine. Yes, I’m pretty sure. Infact, I better check now—F-6F. [Editor’s note:The Grumman F-6F Hellcat replaced the earlier F-4F Wildcat.] Well, then we had many occasionswhere we just barely didn’t commit terrible, ter-rible errors in the conduct of the tests.

Despite all the variations in our testing andour goals, the fact that a good number of aircraftwere devoted to obtaining information for ex-ternal blast tests, we were making good use ofthe 1,200 airplanes that had come in.

I must say this and you can see what oftenmight happen in describing modern scientificstatistical design of experiments. When I hadthat estimate for 1,200 airplanes, there wasa meeting of the Scientific Advisory Committeewith many of the people from the Laboratory.Colonel Simon at that point was talking aboutthe aircraft vulnerability program. He was talk-ing about how this was not a hodge-podge pro-gram, but was scientifically designed from thevery beginning and we predicted that we wouldneed 1,200 airplanes.

Then he turns to me and I’m sitting in theaudience and, ‘‘Mr. Stein, how many airplanes



did we actually use?’’ ‘‘Oh, we’ve used 1,190.’’And he turns to the audience and says to them,‘‘And that’s what you get with statistical de-sign.’’ [Laughter]

As I said, earlier, I really love him. He wasvery good at what I would call scientific sales.The tests we did were completely different fromthe ones that we based an estimate on and weused all we had and if we had 1,500, then wewould have used 1,500.

Jim Williams: You had mentioned earlierthat you did some kind of comparison of thesetests against actual data from combat.

Art Stein: We had received from the EighthAir Force pictures or drawings of B-17s withholes, the idea being to see where the holes wereand the inference being that if you found certainsections where there were no holes, that meantthat those particular places were very, very crit-ical. It was an indirect way of estimating whatparts were critical because no airplanes cameback with hits on those spots.

That was for the B-17s. We had very poor in-formation except gross information from com-bat. For example, that over 80 percent of theairplanes that went down were aflame and thosewere in consonance with the sort of things wewere doing.

However, there was a much more mean-ingful prediction not made during that time,but at a later time during the Korean War.At that time I was at the Cornell AeronauticalLaboratory and I had become interested inthe total aspect of survivability or evasive ma-neuvers, countermeasures, total weapons ef-fectiveness, and the like. We had estimatedwhat the relative losses would be in fights be-tween the F-86 (North American F-86 Sabre)and the MiG-15.

This was before any data was available. Weassessed the fact that there were essentially nolosses with dog fights, per se. If one of the air-craft wanted to not be hit, he could do so—hecould get away. They didn’t have the ability toreally go after somebody who was trying toget away and be able to bring him down.

That seemed to have been borne out and, infact, most of the kills would occur owing to sur-prise from a formation of one type by the othersdiving in on him, going through them, and thengoing away.

When I was at Cornell Aeronautical Labora-tory we had conducted this fuel analysis that Ihave just been describing. To make the estimate,we assumed that half the time the F-86 wouldsee the MiG-15 first and half the time the MiGwould see the F-86 first. The tactic that was mostimportant insofar as losses were concernedwould be to dive through the formation firingand then after passing the formation, to keep go-ing. We didn’t have the complexities of the dogfight to gain more estimates.

I was told by a number of officers that thatwas an appropriate assumption. Then the thingsthat made a difference would be the rate of fire ofthe 50 caliber weapons of the F-86, the vulnerablearea of the MiG-15 to those projectiles, where theopening and the closing range was, and howmuch they carried or could carry and mightuse in their pass.

Similarly, the 23-mm weapon on the MiG-15was a higher caliber than the 50 caliber for sure.But this early version of the 23-mm had a verylow rate of fire and a very poor ballistic coeffi-cient and the net effect was that during thetwo types of attacks by the MiG-15 and theF-86 and vice versa, we estimated an advantageof the F-86 of about 8 to 1. It surprised a lot ofpeople that it was that much. The results werepresented during a keynote speech I gave atthe first Aircraft Passive Defense Conference atWright-Patterson. I think it was held around1950 when I presented the results of that study.

Well, lo and behold, when the war ended,the losses were totaled up of air-to-air combatbetween the F-86 and MiG-15, the ratio was somuch closer to what I had predicted than anyof the errors would be. I mean it was somethinglike 8.1 instead of the round 8 that we had. Theactual number was extremely close, closer thanthey said in the errors. It could have been 8.5 oreven 9, and I would have been very happy. But8.1 was so close!

So, insofar as that kind of prediction wasconcerned it worked out well. You’ve got to ap-preciate that these were two airplanes that weknew something about and that we had madethese estimates for.

Jim Williams: I was reading some materialyesterday flying up here. I ran across a referencethat the Germans had installed combat camerason their fighters and that they had about



a thousand sorties worth of combat film footagethat, after the war, apparently found its way hereto Aberdeen, and Herb Weiss had done someanalysis of that. Did you folks use that as well?

Art Stein: Well, it was used to illustrate a dif-ferent point that I think you would appreciate.The combat films were matched against thedebriefings of officers after they had returnedfrom combat. The officer in his debriefing, thepilot, would typically say, ‘‘I opened fire at say400 meters and pulled away at 150 meters.’’Then they would compare it with the camerafilm to see what it actually was.

Well, yes indeed, in the first couple of yearsor so if he said, ‘‘I opened fire at 400 and pulledaway at 150,’’ maybe it actually was 430 or 320that he opened and when he pulled away at150 it might have been 200 or something likethat, but it was in the ballpark.

These first ones occurred when there was notail turret on the B-17. Then a tail turret was puton the B-17. With the addition of the tail turret,the B-17 apparently seemed much closer to thefighter coming in. Now when he was debriefedand he said, ‘‘I opened fire at 400 and pulledaway at 150,’’ what the camera film showedwas that in fact he had opened fire at over2,000 and pulled away perhaps at 1,200 or1,500 because somebody was shooting at him.That, coupled with the fact that some of the pi-lots toward the latter stages of the war were lessexperienced in Europe, caused a very wide var-iance between the debriefings and the camerafilm in contrast to those that were observed ear-lier during the war.

We’ve often talked, for various types ofweapon delivery, about the comparisons be-tween testing at Proving Grounds and combatdata, about what the Pucker Factor might be.In that case you could express the Pucker Factorin terms of the opening and closing ranges. Weused to use a rule of thumb of something inthe order of 3. That is, the standard deviationfor fire would be degraded to three times whatit would be in Proving Ground testing.

That was your question that I sort of jumpedahead to. I was waiting to talk about Herb Weissand his part of the work. I might as well do sonow.

Let me interject something just to completethe aspect of the vulnerability studies. The blast

work, external blasts particularly, were done byother people at the laboratory. These includedBill (Wilfred) Baker, Jim (James) Sarmousakis,and Joe Sperrazza; and one of the things wedid in order to get more modern aircraft wasto act like ghouls and try to obtain any crashedairplane which would be more modern thanthe ones we had, anything where a function-ing portion of the airplane could be useful fortesting.

That’s how we were able to get informationon the vulnerability of integral fuel tanks to in-ternal bursts by explosive shells. A B-36 (Con-vair B-36 Peacemaker) went into the lake thereat Dallas/Fort Worth and it was essentially be-ing junked. When it was pulled out, there werevery large sections that were intact and good. Sowe immediately had called to try to see what wecould get of that. There was no way we couldmove what was there elsewhere. So we wentthere with our explosives and we detonated in-creasing sizes, from quarter-pound, half-pound,and so on going up inside within the very largeintegral fuel tanks on the B-36. We got a greatdeal of information. It was the only informationwe had for a long time and that only came aboutbecause of the fact that unfortunately there hadbeen that accident. That also was a source of in-formation and data for more modern aircraftwhen they had an accident of that type.

Jim Williams: Did you get pretty goodcooperation?

Art Stein: Oh, yes. In fact, everybody therefelt so bad that this happened and there’s sucha loss. The fact that you could retrieve somegood from it and get some information madethem feel a little bit better. Yes, we got good co-operation.

Now the Air Force was concerned aboutthis money they were giving to our team. I thinkwhat really stuck in their craw was the fact thatafter we were finished with the airplane, we hadit taken away by a scrap contractor, located onthe field, who was paying the Army about an-other million dollars a year just to dispose ofand to take the scraps.

The people at Eglin then attempted to geta transfer of our group to Eglin to do this work.None of the group wanted to leave, but eventu-ally an additional group was set up at Eglin forvulnerability work. That was quite a while after



I had left Aberdeen, but the first attempt wasaborted.

Before leaving the discussion of the aircraftvulnerability efforts at Aberdeen, I should men-tion that since the initial start of the group that Ihad described with Herb Weiss and two otherfolks, the group had grown to where it included,as one of the key people, Mr. Morgan Smith,who had come from the New Mexico School ofMines to work at the BRL. He was essentiallyin charge of all of the fieldwork that was con-ducted. Mr. Harry Kostiak’s general supportserved somewhat as Executive Officer withinthe group itself.

Then I had asked a number of fellows, whowere all mathematicians, what their choicemight be. Who wanted to become our experton power plants and who wanted fuel systemsand the like because I would then be expectingthat any time I had questions related to those,they’d be able to come up with the answers.So we had Roland Bernier in the area of fuel sys-tems and so on throughout all the subsystems ofthe airplane.

Interest was not only on the matter of thevulnerability of the aircraft being hit, but onthe entire question of weapons selection. In theAircraft Weapons Branch the aspects of duelsbetween aircraft, the questions related to hitson aircraft, the questions related to the optimumtime to open fire and close fire, were all dealtwith directly by Herb Weiss. He had one ortwo other people who would come on boardto assist him. Even though he was the managerof the entire operation, he nevertheless was veryprolific in writing papers of his own in thoseareas.

In this sense, much of what we know todayas being OR has greater similarity to the workthat Herb Weiss did than to what we were doingin the aircraft vulnerability part of the group. Heleft BRL in 1952 or 1953 to go to Norton and sub-sequently served on the Army Science Board, onthe Scientific Committee for BRL.

It’s interesting the way in which Weiss oper-ated. He ‘‘felt’’ the correct solutions, initiallyintuitively, and then he would proceed to dem-onstrate them. Often his demonstrations mightnot be mathematically rigorous and there werea number of people around the country whomade it their business to make rigorous the

things that Herb Weiss had observed and hadheuristically demonstrated. This was true in theOperations Evaluation Group (OEG) in the Navyand other places around the country.

During this period, the needs of various or-ganizations multiplied insofar as aircraft wea-pons effectiveness was concerned and thewhole question of games and theory.

The Air Force initiated a contract at the In-stitute for Air Weapons Research at the Univer-sity of Chicago and these folks were located inthe Museum of Science and Industry in Chi-cago. That’s where I first had the opportunityfor meeting Clayton Thomas, FS, Tom Caywood,Walt Strauss, Ed Schiller, and the like. They cameto Aberdeen to get orientation with respect tosome of the terminology we were using, e.g.,what we had found out insofar as the vulnerabil-ity of aircraft was concerned and weapons effec-tiveness. They then were to go back and go onfrom there pursuing various Air Force-specificproblem areas.

I’ll assume that at some point in the archivesClayton Thomas will be asked to describe theactivities of that group themselves.

With the expansion of the effort after theend of World War II, all efforts then were di-rected toward what the future might be like.When we got into the Korean War, a numberof other types of problems surfaced. One ofthese had to do with the ballistic quality of theartillery ammunition being used in Korea.

Apparently 40 percent of the 105-mm How-itzer rounds fired were fired for observationrather than for effect. This was extremely highand was due to the fact that there were so manysmall lots at the gun positions. Each time youchanged from one lot number to another youhad to reregister and refire because of the dif-ferences in ballistics behavior from one lot toanother.

I was asked to go to Joliet and head a groupof ordnance people to serve with people fromBell Telephone Laboratories who were calledin to attack this problem. It was one of the mostrewarding technical experiences I had.

Rarely does one have the opportunity toparticipate in the examination of requirements;into R&D activities to determine differentialeffects and which variables really drive disper-sion; into matters related to procurement,



inspection, testing, and evaluation for ballistictests, production, and production control; andone of the most important aspects—distribution.One of the problems that existed is every timethere would be a distribution point whereammunition lots were divided among freightcars and among trucks. At the end there werevery few large lots that would arrive at any oneposition.

The average ammunition lot size in thosedays was on the order of 9,000 rounds. Yet,one powder lot alone could serve 150,000 com-plete rounds of 105-mm Howitzer ammunition.Within the space of less than two years it waspossible to produce very large lots. The firstlarge lot was JAA-1. I still remember the lotnumber and giving the go ahead for loadingthat lot out. In that lot, the dispersion was lessthan that which existed in the smaller lots previ-ously.

Another activity that existed at Joliet, and inwhich I found myself being involved more andmore frequently, had to do with the investi-gation of various problems in the munitionsprocurement, aside from this question of the105-mm Howitzer.

The Korean War period was one of rapid en-largement of the production facilities, and thetransfer of people among plants into positionswhere they had to learn not only characteristicsfor the new munitions themselves, but also newmethods of manufacturing process control. Itwas during the time when there was increasingemphasis on statistical process control and itsuse in the munitions industry.

I recall being asked to go to Kingsbury Ord-nance Plant in Indiana due to a problem withthe 20-mm ammunition scheduled for use onthe B-36 bomber. There was not one lot acceptedand the causes were several. On the one hand,lots would be rejected or suspended becausethey did not meet bullet pull tests. The roundwould be tested in a tensile machine whereinthe cartridge would be pulled one way andthe projectile the other to see how much bulletpull is required to break the crimp. The bulletpull had to be controlled within fairly goodlimits because if it was too hard you would rup-ture the case and get erratic muzzle velocitiesand also foreign particles in the chamber ofthe weapon. If it was not crimped hard enough,

you might get too low striking velocities or vari-able striking velocities. So there had to be goodcontrol on bullet pull.

The typical manufacturing process in-volved one machine wherein the shell and pro-jectiles would come in along two different belts.They were weighed. Propellant would dropdown into the cartridge case. The projectilewould sink into the neck of the cartridge case,pass a crimp bar, and eventually be the com-plete round.

When I visited the plant I saw a very consci-entious man hovering around this machine likea mother hen. Every few minutes he would pullfrom several projectiles, go to his tensile ma-chine and see what it took to pull it. If the pullwas above the nominal value, even though itwas within the specification on this projectilethat he had collected, he would try to fine-tuneback down to the nominal value in terms of thepressure on the crimp bar.

If it were below or above that nominalvalue, based on those samples of one, he wouldmaybe make these corrections. In effect, if hehad had a practice that was controlled wellwithin the two specification limits with a meanright on the nominal between those two specifi-cation limits, and the tail of the distributionreaching out to the lower and upper specifica-tion limits, he should have left the process alone.However, due to a random variation, he pickedone sample and said that the sample might beone or one and a half sigma below or abovethe nominal, now he fine-tunes to get that onerandom reading back to a nominal. In effect,he’s shifted the whole distribution now. He’sthrown a large part of the tail of the distributionoutside its specification limit, where if he’d leftit alone everything would have been fine.

The next time he gets a reading, perhaps inthe other direction, he shifts the distribution theother way. The effect is that he’s taken the orig-inal process distribution and broadened it, andas a result far too large fractions are outside thespecification limits. The lots become suspended,even though the process was inherently okay.

This overzealous operator should have beentaking larger samples and working with themeans of those samples if he wanted to try touse them for machine process control. Problemssuch as these existed in many of the plants in



the industry and it became necessary to write in-structional manuals that could be disseminatedthrough chief inspectors to their staffs and thenproduced within the various companies.

Our greatest moment in that whole activitycame from reports back from the field in Koreaas to the behavior of the 105-mm large lots ofammunition that were being sent there as a re-sult of the ballistic quality control project.

We had never expected that the batterycommanders would have the time or inclinationto answer some of our questions regarding theammunition. To our surprise, however, wefound they would write reports, some of them50 to 100 pages long, about their experienceswith this ammunition, how much they liked it,how much they felt it saved them in terms oftime and effectiveness, and getting on targets.

At no other time did I have the opportunityto see the results of an R&D activity materializein actual combat, and then be recorded upon bythe user the way we did with the 105-mm am-munition.

Gene Visco: Do you want to talk a little bitabout the international meeting?

Art Stein: To come back to the aircraft vul-nerability area for the moment, one of the thingsthat was possible to get initiated quite early inthe process was cooperation among England,Canada, and the United States in sharing of in-formation and in working together and coordi-nating efforts. These were known as workingconferences on aircraft vulnerability.

In the first working conference, for example,small teams of two to four people were set up tolook at specific problems that the group asa whole felt should, and could, be examined inthe short time that the group was together, bypooling the information that various attendeesbrought to the meeting.

For example, one group examined the prob-lem of decompression in cockpits due to holes inthe cockpit, in the canopies, problems related topressurized suits, and what effects would be atdifferent altitudes or different hole sizes. Otherslooked at what the needs might be in fuel andbeing able to predict the results of fuel fires.

We incorporated the best of what we couldfind in the various efforts going on in the threecountries into the proceedings of the workingconferences, which appeared as BRL reports.

Skipping back to the work at the OrdnanceAmmunition Command, experience indicatedthe need to build up a competent staff that workcould be delegated to with high trust.

In the 105-mm Howitzer Ballistic QualityControl Project we found as a result of numbersof experiments and trials that the sources of dis-persion in range and muzzle velocity are dueto the shell itself: the rotating band diameter,the rotating band tighteners, the surface finish,dynamic unbalance, and to some extent, otherfactors relating to the boat tail [Editors’ note: Aboat tail bullet has a slightly tapered base at its rear.].Insofar as propellant was concerned, there wasa problem related to moisture content, and itscontrol. Different amounts of ‘‘dry’’ powderwere found in bags even though the total weightwas the same when weighed out.

We found that the moisture content of pow-ders varied sufficiently so that one could obtainup to 11 feet per second variation in muzzle ve-locity from one day to the next. Powder is notweighed in climate control conditions.

It was necessary to obtain the voluntary co-operation of the companies producing 105-mmshells because there was no time for renegotia-tions of contracts. There were about 20 manu-facturers of the steel shell using variousprocesses and they were all paid, and theirproducts accepted, if they were in conformancewith the original agreement.

However, they could do 100 percent, or de-tailed inspection on the variables that we hadrecognized as being important. They could re-ject or rework shells that fell outside of the spec-ification limits, and the net result of all thatbeing a fairly uniform distribution within thespecification limits rather than the more idealnormal distribution peaked near the centerand tailing out at the specification limits.

To obtain this voluntary compliance and toapprise the industry of how it was doing vis-a-vis other members of the industry, I took it uponmyself to publish a newsletter. I sent it around tothe various companies and I indicated how manyshells were rated A, B, or C in terms of qualityfrom the standpoint of ballistic uniformity for eachof the shell characteristics that we had found to beimportant and that were impacting on dispersion.

Some of the companies responded veryquickly and were producing the AA rated shell



within less than a month after the program wasstarted. However, this was not true of all and, inparticular, one company called me. The presi-dent called me to complain that his company’sC rating had been disseminated to the rest ofthe industry through this newsletter.

I told him that we had received a number ofrequests from people who wanted to know howthey were doing vis-a-vis the others and we feltwe were providing a service. However, if hedidn’t want to be included, we wouldn’t in-clude him and just make a note that the Suchand Such Company didn’t want to be included.He said, ‘‘That’s blackmail,’’ and I said, ‘‘But it’strue, isn’t it?’’ I said, ‘‘But why should we havethis argument regarding whether or not one isincluded in the letter when it would be so easyfor you to always have gotten the highest rat-ing?’’ I said, ‘‘I understand how this has comeabout and if I could speak to you I think youwould agree that this was possible and also de-sirable.’’ I visited this company CEO and de-scribed how much we thought of the capabilityof his plant manager, who had written a numberof articles in Machine Design. The articles dis-cussed how to control diameters using lathesversus centerless grinders, and how much hecould obtain at a lathe. And it was true, he couldget more at a lathe than others seemed to be ableto, but they did not and could not produce aswell as some of those grinders did.

It seemed that it was about time to move to-ward the use of the centerless grinder, whichthey did, and that enabled them to producethe highest score of the ammunition in the field.

We had a number of those instances wherefor one reason or another it was necessary topersuade the people involved. I found, for ex-ample, that foundry workers, who were a breedapart from all other men, had been very difficultto convince or to have any concerns at all aboutwhat the shell ratings per se were, or any of theother advice. But when explained to them interms of the impact poor dispersion had on in-creasing numbers of casualties, it meant a greatdeal to them and their complete cooperation wasobtained. They were eager to participate andmake suggestions as to how to improve theweight and cavity of the projectiles, which haddetermined the piercing operation of the bulletsthat they started with.

Similarly, we had problems with the largestshell plant, which was producing almost asmuch as the rest of the industry together. Itwas possible to demonstrate that one man ona centerless grinding operation could handleboth the rough and finished grind and wouldhave little or no inspection if he was able to in-corporate the suggested process-control proce-dures. There was a great deal of oppositioninitially from the plant manager who said, ‘‘Ei-ther you fellows show me how you could saveme money and I produce the same quality, orhow I can improve my quality and not cost meany more.’’ He said, ‘‘But I don’t believe anyof this quality control stuff.’’

Well, we had selected the centerless grind-ing operation and were able to demonstratethese savings. Their process had a line goingpast each of the 12 banks of rough and finishedgrinders and on this belt would go many, manyshells to the ‘‘hospital’’ for rework and there wasa great deal of scrap produced in the way inwhich they had been operating. He was skepticalof savings until after using our process-controlsystem design to remove the scrap. It turnedout that after a full day’s operation there werejust three pieces that were slightly oversized.The plant manager was completely won over.

The 12 banks were operated three shiftsa day. Each grinder had an operator and therewas an inspector for each. By going throughan inspection scheme where they would leavethe process alone and not measure again for fiveminutes, it was possible to have the same in-spector operate two of the banks with no diffi-culty whatsoever. The reset limits were basedon new, medium, or old wheels in terms of therate of trending that you get with the wear onthe wheel and the process was very easy to adapt.

The addition of the two readings, the sum,which was in effect the average of two, the dif-ference between them was the range, whichwas an estimate of the dispersion, and all theoperator had to do was to put in the numbers,the sum or the difference of the two and if theywere higher or lower than the reset limits, thenhe reset, otherwise let it go for five minutes.

The plant manager liked the savings inpersonnel so much that it turned out he wantedto use the same system on all the other opera-tions in the plant. This is where he ran into



trouble because while the people from the cen-terless grinding operation could be distributedamong some of the others, there came the pointwhere there just were too many people andsome people would have to be let go. The unionobjected to the whole procedure that was pro-ducing these results. The plant manager felthe had less trouble staying with the old systemthan to buck the union on this issue.

At the same time, there was a problem atone of the well-known wheel companies thatwas making the 105-mm Howitzer shell. Therethe union had a grievance and complained thatthe system should be used throughout the plant,not just on artillery shells. Their men were beingpaid on an incentive basis based on how manygood pieces they made in a day. The processcontrols that were put in for the 105-mm How-itzer resulted in those men making more moneythan the others in the plant. This resulted in dis-satisfaction to the degree that they insisted thatthe Army help them put in the same kind of con-trols on everything that they did there in theplant, not just the 105s.

The period was very interesting. Many in-teresting side problems were examined, suchas how to detect whether operators were eageror shied away from finding a defect that wouldreject a large lot of ammunition or which wouldshut down a line. There had been indicationsthat some inspectors would shy away fromreaching that decision point, and others wouldleap forward and were only too eager to findthe serious impacts on their operation. Wefound it very interesting regarding the statisticalmethods that were used to separate these out.

All and all it was a very active period, bothin the growth of quality-assurance methodsgenerally and the use of analytical techniquesin the control of the internal operations in thefirm. This had to do with the examination of dif-ferent vendors, whether they were responsiblefor the outgoing quality of the plant, how muchwas contributed internally within the plant,how to manage a quality-control operation mosteffectively, and ultimately how to write govern-ment specifications so that the government couldreceive a product that had been produced underprocess control.

In October 1955, after quite a bit of discus-sion on the part of various companies, RAND,

Cornell Lab, University of Chicago, and others,I decided to join Cornell.

The areas of aircraft vulnerability and sys-tems analysis really broadened after I joinedCornell Lab. I was involved with all aspects ofaircraft survivability and the examination of re-quirements for tactical air as well as tacticalground operations.

However, by this time the entire area of op-erations research had expanded significantlyand was much more well-known insofar as thepublic record is concerned. So I’ll stop here in1955.

In thinking back about those early days andthe various activities in which I have been in-volved, it would appear that those activities re-lated to the sampling inspections and processcontrol have continued development along verymuch the lines that we had been implementingwithin ordnance.

However, when it comes to the matter of op-erations research with respect to certain optimi-zation problems, questions related to aircraftvulnerability and weapons effectiveness, I be-lieve that we have both gained and lost some-thing with the advent of the computer.

What we have gained, of course, is the abil-ity to tackle complex problems and to do sowith far more different types of cases thanwould have been possible otherwise. Many ofthese could only be solved by numerical meansand the computer has enabled the whole areaof finite element analysis to be used for the de-piction and explanation of complex interactiveevents.

It is, however, also true that the tool hassometimes obscured the object that was beingfashioned and that we find many modelersand analysts who do not have a great deal ofsubject matter knowledge. They cannot recog-nize the reasonableness of some of the outputsthat occur, the difficulties or means whereby ap-propriate inputs would be obtained, and theloss of ability to analytically conduct at leastbounding investigations to see how worthwhileit is to go into a large-scale examination.

We have seen people react to a new problemsomewhat like Pavlov’s dogs and immediatelystart to model the problem without thinkingthrough, at least for a few hours, as to whetherit was possible to determine a much more



general appropriate solution analytically. Or as Isometimes have said to people, ‘‘How wouldyou do the job assuming you didn’t have a com-puter available to you? What would you do togive me the answer tomorrow? What wouldyou do to give me the answer next week if youdidn’t have a computer available to you?’’ I havefound that just in thinking through thosemethods, people have found more efficient waysof employing the computer rather than bruteforce, head-on attempted simulation.

It was very interesting to us to find the de-gree to which we could impact, by analysis, theoperation of extremely serious problems relatingto the survival of our personnel and equipmentand the effectiveness of our operations.

Little did we dream then that the entire areaof operations research and management sciencewould mushroom to the degree that it did froma group that may have numbered no more than50 or 100 working nationally.

DUSA(OR) PROJECT INTERVIEW OFDR. ANNETTE STEIN

Dr. Jim WilliamsUS Army Military History Institute

INTRODUCTIONDr. Annette Stein was the wife of Mr. Arthur

Stein, FS, the third President of MORS. AnnetteStein was one of the human ‘‘computers’’ dur-ing World War II and worked for Dr. John vonNeumann. Von Neumann, a true polymath,was famous for being a principal member ofthe Manhattan Project and for the developmentof game theory and the digital computer, amongnumerous other accomplishments. Dr. Stein re-ceived her PhD from the University of Buffaloand then taught on the faculty at Buffalo StateCollege for 15 years, specializing in diagnosisand remediation for children and psychomet-rics, the measurement of their abilities and dis-abilities. Dr. Stein is now a Professor Emeritusat Buffalo State College. This interview wassponsored by the Office of the Deputy UnderSecretary of the Army (Operations Research),

DUSA(OR). The interview was conducted onJune 5, 1992 at Dr. Stein’s home in Williamsville(near Buffalo), New York.

Jim Williams: Dr. Stein, I’d like to ask youjust to give a little bit about your background,where you were born, raised, and that sort ofthing.

Annette Stein: I was born in New York Cityand grew up there. Most of the time I lived inthe Bronx, and I went to school there. I went toHunter College, which at that time was a collegefor women and was also in a growth period; somy first semester I went to school at Number 2Park Avenue, an office building in downtownNew York. Then we went to the new BronxCampus, so I was able to walk to school fromhome and I was there for a year and a half. Thenthe newest of the buildings, the one that waserected on 68th Street and Park Avenue, wasready and I attended there for the second twoyears. I was a member of the first graduatingclass from that particular building.

I was an English and journalism major and Istudiously avoided taking courses that I didn’tparticularly enjoy. I took the required coursesin math. I did not take any education courses.I really didn’t take any of the difficult sciencecourses either. I had physiology and anthropol-ogy and all kinds of good things like that, a lot ofeconomics and political science, and I thought Ihad a pretty well-rounded bachelor of arts.

Jim Williams: Did you enjoy anthropology?Annette Stein: Yes. I had a wonderful pro-

fessor, Dr. Elizabeth Keuer. She was marriedto an archeologist who was on the faculty ofColumbia University and the two of themwould spend their summers in the Southweston various archeological digs. I was very fortu-nate I thought, and still do, to have been invitedto her home one time for afternoon tea and I wasreally impressed. Her home was beautiful and itwas filled with treasures of Native Americanart. I had known that she had wonderful silverand turquoise jewelry because she always worethem to school and there were always differentpieces—beautiful belts and squash blossomnecklaces—just marvelous. I took her coursesin 1939 and I graduated from Hunter in 1941.

The art of the American Southwest reallywas not very well known in a big city likeNew York at that time. It later became very



popular. But when I went to her home for tea Iwas overwhelmed by the beautiful artifacts thatshe had, rugs on the walls and on the floors, andpottery. It was just like being in an informal mu-seum. That was the beginning of my interestin things like that. So imagine what a thrill itwas when Arthur returned from a trip to NewMexico some years later after we’d been mar-ried for about six years, and he had a pair of sil-ver and turquoise earrings for me and severalsmall pots from the Pueblo Indians, which westill have today.

That was the beginning of a love affair withthe Indians of the Southwest and we’ve returnedmany times and we keep adding to our smallcollection. They are a continuing source of realpleasure and joy, I think for both of us, but Iknow for me.

Jim Williams: When did you get married?Annette Stein: We were married December

28, 1941. Actually we had originally plannedto be married over the Thanksgiving holiday.Arthur was already working at Aberdeen Prov-ing Ground, Maryland, and so our plan was tobe married at Thanksgiving time. He would behome for a long weekend, home meaning NewYork, but he was not able to get any housingfor us until the end of December. We delayedour wedding and we were married, instead, De-cember 28. We didn’t expect that between thosetwo dates that we would have the beginning ofour involvement in the war with the bombingof Pearl Harbor.

Jim Williams: What is your recollectionabout the Pearl Harbor bombing?

Annette Stein: Well, Arthur was home for theweekend and he got tickets for a New YorkGiants football game at the Polo Grounds. Iguess that doesn’t exist anymore—no, it’s gone.[Laughter] But at any rate, we were at the foot-ball game and there was an aura of excitementall through the game, as there usually is, butthen something different became apparent aftera while. For one thing, on the public addresssystem there were calls made for General Some-body and Colonel Somebody, and the name that Iremember hearing was a Colonel Doolittle andthese people were asked to report to the officeand that seemed a little bit ominous. But we re-ally didn’t know what had happened, thoughwe knew something had occurred. We didn’t

know what it was until the game was over. Weleft the stadium and by the time we got to thestreet we heard that Pearl Harbor had beenbombed and we could hardly believe that.

I remember so well the next day when I wasat work. I was working for a printing companyat the time and everything stopped when Pres-ident Roosevelt made a speech on the radio.We all listened to it and I don’t think I’ll ever for-get that particular day and how people lookedand how frightened we were. It was such a scarything for us. It was an experience that you’re notlikely to forget. So that was where I was. I was ata football game when the bombing occurredand heard officers being called. There werea lot of officers at that football game. Then thenext day at work everything stopped so thatwe could all listen to the President.

Jim Williams: You said that it came as quitea shock. What do you recall about the build-up though? The war, of course, had been goingon in Europe for some time and I would expectin New York City folks were pretty well awareat least of some things that were going on.

Annette Stein: Oh, surely, but somehow theidea that the Japanese would fly over Pearl Har-bor and bomb our Navy—I don’t think that hadoccurred to anyone as a possibility, certainly notto me. Of course, I guess I was pretty young atthe time. I was just 20 and even though I wasvery much aware of the war in Europe and verysad and worried about that, somehow I neverexpected that we’d have such a terrible thinghappen to us at Pearl Harbor. It’s hard to believeeven now, even though I’ve been there severaltimes and stood right there at the monumentover the sunken ship.

Jim Williams: You went to Aberdeen at theend of December then, when you were marriedand when Art was able to get housing?

Annette Stein: Right.Jim Williams: And you were there some time

before you went to work at the Proving Ground?Annette Stein: Yes, I guess I was there for

about three months. We had ordered our furni-ture in New York City over that ThanksgivingDay weekend and because we were youngand inexperienced, and I guess a little bit stupidabout money, Arthur had paid for our furniturebefore it was delivered. [Laughter] I guess therewas no real compulsion on the part of the dealer



to see to it that it got to us because he alreadyhad his money. But what it meant was that wewaited for almost three months until all of thefurniture was delivered. The day after we ar-rived in Aberdeen our box spring and mattresswere delivered to the house and that’s all we gotfor a month. We had that sitting on the floor inthe bedroom on top of the packing cartons thatthey had come in and that was the extent ofour furnishings.

We had a very kind neighbor who lent usa card table and two chairs so that we hada place to eat. Also, when we had come down,my mother had packed some essentials thatshe was sure we would have to have when wegot there and so we had some linens, a downcomforter, pillows, and a little radio. We carriedall of that with us wrapped in boxes with oursuitcases on the Pennsylvania Railroad downto Aberdeen. I guess we must have looked likeimmigrants getting off the train. [Laughter] ButI was very grateful that we had those thingswith us, otherwise I don’t know what we wouldhave done those first few weeks.

Jim Williams: Did you live on post?Annette Stein: No, we had wartime housing

just off the post, adjacent to the post in an areacalled Swan Meadows and these were small du-plex houses. That is, there were two units in onebuilding. They were single-story houses with-out a basement and in the kitchen we hada hand-stoked coal furnace and a gas waterheater that was not automatic. You had to lightit with a match and remember to turn it off afteran hour, otherwise it would start hissing steam.This was really interesting because both Arthurand I grew up in apartments in New York City.A furnace was something in the basement of theapartment house and hot water came out of thewater faucets and we never had to think aboutwhere it was coming from.

We had another kind of situation in thehouse. We had a coal bin outside the kitchenand we had to remember at night to stoke thefurnace and be sure to dampen it after the newcoals were going; otherwise, we’d get up inthe morning and we’d have a cold house andthat happened because we didn’t always re-member. [Laughter] We also had to rememberto light the hot water heater and to turn it off,but we didn’t always remember to turn it off.

So we frequently had scalding water hissingout of the faucets, but, you know, these were mi-nor inconveniences. It was an adventure. It reallywas. We loved our little house. We plantedflowers and trees and cut the grass. It was justso different from growing up in an apartmenthouse in New York City and, besides, we werenewlyweds. Even a hand-stoked coal furnacewas not too much of a chore.

Jim Williams: Were those housing units builtjust for the people that worked at the ProvingGround?

Annette Stein: That’s right. We were verylucky to get one of those units because shortlyafterward they were all occupied and otherhousing was built at a greater distance fromthe Proving Ground. They weren’t nearly asnice either, if you can imagine something thatwas not nearly as nice as a house with the fur-nace and the water heater in the kitchen. Butwe had the most elegant accommodations thatwere available at the time.

Jim Williams: Were the people who lived inthe other units people that worked at BallisticsResearch Laboratory (BRL) as well?

Annette Stein: No, they didn’t work at theLab. Two of the women worked at the ProvingGround, but not at BRL, and the neighbors onthe other side were Army people—a sergeantand his wife and child. It was a very congenialcommunity. We all got along and enjoyed eachother.

Jim Williams: When did you actually go towork at BRL?

Annette Stein: After about three months. Ev-erything was delivered. We had curtains up onthe windows. I guess I had washed the win-dows and waxed the floors just as many timesas they needed to be done and I felt I could leavemy house safely and go find a job.

I told you earlier that I had been an Englishand journalism major in college. I thought that Imight get a position on the Proving Ground inthe Public Relations Office. And one fine day Iwent to the personnel office and told them whatI was interested in and I was told that there re-ally weren’t any openings in public relations,but did I say I had a college degree? And I said,‘‘Yes.’’ He said, ‘‘Well, do you have any collegemath?’’ I said, ‘‘Yes, the required course.’’ ‘‘Well,we’ll send you over to Ballistics Research



Laboratory to see Dr. Dederick.’’ And off I went.Of course I had heard about Dr. Dederick fromArthur. I felt just a little bit intimidated. I wasout of my depth.

Dr. Dederick also asked me what math I hadhad in college and I told him the one requiredthree-hour course as a freshman, some calculusand trigonometry, and I don’t know what else. Igot an A, but that was all I had. He said, ‘‘Youknow, that really isn’t very much.’’ I said, ‘‘Iagree.’’ He said, ‘‘Stein, are you related toArthur Stein?’’ ‘‘Yes, I’m his wife.’’ He said, ‘‘Well,it really isn’t enough math, but maybe Arthurcan teach you at night.’’ So I got a job and I washired to be a computer. And I knew what a com-puter was then. A computer was somebody whoworked on a calculating machine and did com-putations, but today if you tell somebody thatyou were a computer, that’s really pretty funny.There were many female computers, mostlyyoung women who had started within, aboutsix months of the time that I started and somewithin a few months after I did. Most of themwere graduates of very nice women’s collegesin the South and it was my first introduction tomany different southern accents. I became veryadept at distinguishing between North andSouth Carolina as against Alabama, as againstMississippi, but all of these young women—theywere all about my age. I was 20 at the time andthey were anywhere from 20 to 22 years old.There was one woman from Boston. I rememberher well. We were good friends.

Jim Williams: Had these women that hadcome from the South been specifically recruitedto come up there? Do you know how they cameto be there?

Annette Stein: I’m really not sure, but all ofthese women were math majors and they allhad much more extensive backgrounds in maththan I did—obviously anybody had more than Idid. But they were math majors and I will saythat even though I wasn’t a math major, therewas nothing that I had to do that required meto use anything that was beyond what I coulddo or what I had learned in my one collegecourse. It was fairly routine work and we had,for example, bombing tables that we filled in.We had formulas. We had data and we workedthe calculating machines. It was a question ofaccuracy and speed, and I guess I was just as

accurate and just as speedy as the math majors.[Laughter]

But the interesting thing was that when wewould finish one of these enormous sheets, wewould bring it to a supervisor, Elvin Martin. Iwas always amazed at the way he would takethis sheet and run his finger up and down thecolumns and out of these thousands of numbersif there was one that was wrong, he’d find it andit would take just a few minutes for him to dothis. At that time I had a hard time understand-ing how he was able to do that. Of course in lateryears I was involved in checking the work ofcollege students at the college where I wasteaching. When I reviewed their test results thatthey had done from testing children, I could im-mediately spot numbers that didn’t look right.At that point I frequently thought back to thedays at BRL when Elvin Martin would go upand down those columns in lickety-split timeand find an error, if indeed there was one. Youhave an expectation for what is to be there andif something is not right, it stands out. But Iwas impressed when I watched him do it.

I guess I worked on bombing tables forabout a year and even for a nonmath major likeme, it got to be boring. Not only that, we were sobusy during that year, there was so much workto be done and there wasn’t enough space ormachines. It wasn’t a question of getting morepeople because there was no place to put them.We had shifts: there was night shift and therewas swing shift and there was day shift. I can re-member coming home from the night shift,walking down Liberty Street from the Doodle-bug [Editor’s note: the Doodlebug was a self-propelled railcar], the railroad that used to takeus into the Proving Ground. And as I’m walkingdown the street to my house, Arthur would bewalking up the street to the Doodlebug to goin and we’d say, ‘‘Hi, there.’’

Jim Williams: The Doodlebug was —Annette Stein: The Doodlebug was the little

train that ran from town and made a stop atSwan Meadows and went all the way into theProving Ground. It made stops along the way.We didn’t have a car then and you couldn’tget a car then. So we were really very fortunatethat we lived so close to the Doodlebug. It wasthe equivalent of about a three-block walk tothe railroad to pick it up. But when I was



working night shift it was really amusing to becoming home and Arthur would be going off towork and we’d meet on Liberty Street.

Jim Williams: Did you folks have any daysoff or anything like that, or was this a sevendays-a-week job?

Annette Stein: No. It was a five days-a-weekjob. We did have weekends and I think we weretwo weeks on a shift and then we’d go to an-other shift. I was only on night shift two weeksout of six. The swing shift wasn’t as bad, butthere were two aspects: I guess I was gettingtired of the shift and I was also getting boredwith the work. I asked if I could be transferredto do something a little bit different. That waswhen I went to work for Dr. Alex Charters in Ex-terior Ballistics. I did some calculations there,but I also did some other things. They had pic-tures of projectiles that we used to make mea-surements of the shock waves and of the wavelengths along the projectile.

I have to tell you that I didn’t truly under-stand what I was doing or why, but I did it well.[Laughter] Again, it was work that was preciseand required accuracy and I enjoyed doing it.It was a change from just doing calculationsand it was also an opportunity to do two differ-ent kinds of tasks rather than the same thing allthe time. I found that much more pleasant. Thisis where I had the rare experience of meetingand working for Dr. John von Neumann, whowas on the Scientific Advisory Committee forthe Laboratory. Of course, I had heard Arthurtalking about the Scientific Advisory Commit-tee and I knew who many of the people were,but I didn’t know them personally. Dr. vonNeumann was a legendary name and I was alittle bit apprehensive about doing work forhim, but he was really a very sweet and verygentle man.

After I had done some calculations andsome measurements for him, he used to askfor me to do tasks at other times when he came.That was very flattering, very pleasing. He was,as I say, really such a nice man. Then I’d hearArthur’s stories about meetings that he was atwith Dr. von Neumann and some of the funnystories that he had to tell. That was a whole otherdimension because that was not the Dr. vonNeumann that I knew. But I did know that hewas truly a very special person and it was nice

to know that I was doing something for a verybrilliant man, especially since I really didn’t havevery much in the way of qualifications exceptthat I was a careful and meticulous worker.

Jim Williams: Did you still do computationsfor Dr. von Neumann?

Annette Stein: Yes. He’d bring things for meto do and he’d tell me what to do and I would dothem. So, the human computer. [Laughter]

Jim Williams: Did this require you to learnmore math?

Annette Stein: Not really. You know, I nevercould really understand why we computers hadto be math majors—I was the notable exception—because we didn’t need to know very much todo the job. I will say this, there was no way thatI would ever go beyond the rather menial tasksthat I was doing, whereas there were otherwomen who were able to advance because theydid know what they were doing.

Jim Williams: Was there much opportunityfor advancement for women in your section atthat time?

Annette Stein: Not much. As a matter of fact,I think that at the time I was there, only onewoman had professional rank. All the rest ofus were subprofessionals. I didn’t deserve tobe anything but a subprofessional, but therewere people there who were capable, knowl-edgeable, and well-trained and they still didn’thave professional rank. At the time it seemedto me grossly unfair and looking back, it certainlydoes. After a while there were a couple of womenof professional rank. One of them was Dr. JaneDewey. She was at the Laboratory. I think shecame to work there after I left. I worked for twoyears and then I became pregnant, and in thosedays you really didn’t work up until two daysbefore your baby was born. I left when I wasthree months pregnant and never returned. Iwas a happy housewife and mother.

Jim Williams: Art was drafted, or enlisted,and went into the Army for a period of time.Did you stay there at Aberdeen through all thatperiod?

Annette Stein: Yes. He was drafted. He wasunder 26 and he did try to enlist in the Navy.He had a commission that was promised tohim. He met the people he was going to workwith and was shown his desk. It was all very ex-citing, but was not to be because when he was



inducted into the Army there were papers at ev-ery step of the way that said that he was goingback to Aberdeen; and that’s what happened.I’m not complaining because we were togetherand he got back to Aberdeen about two weeksbefore our daughter was born. So that wasa comfort. He had several port calls after thatand his possessions were shipped out a coupleof times, but somehow or other he never didgo and he was at Aberdeen most of the time.We stayed in our little house on Liberty Streetand both of our children were born there andwe lived there for a total of 10 years—and theywere 10 good years. The house was very inade-quate, but we were happy there.

Jim Williams: Did you leave the house onLiberty Street because your husband left BRLor did you move out for other reasons?

Annette Stein: Well, we had been looking fora house to buy for some time and we hadn’tfound just the right thing. Then Arthur wastransferred to Joliet, Illinois and we moved to Il-linois. We lived in Park Forest for three anda half years while he was at Joliet. Then whenhe was ready to leave Joliet, a big decision waswhether to go back to Aberdeen to BRL or togo to Cornell Aeronautical Laboratory here inBuffalo. It was not an easy decision to makeand I know that Arthur vacillated. He had a hardtime making up his mind, and I remember thatwe drove down to Aberdeen and we lookedaround. Even though we had been very happyfor our 10 years there, after having spent threeand a half years in Illinois in a different kindof community, we were no longer in what wasessentially a company town. We were living ina community with people from many differentwalks of life and I liked that. I also liked theproximity to Chicago and the things that ithad to offer in the way of cultural activities;not that Baltimore didn’t have any, but I think

it was mainly the idea of living in a communitywith a varied population rather than peoplewho were mainly Proving Ground people.

We could have lived in Baltimore andArthur could have commuted, but that justseemed like a lot of traveling. Even his commut-ing from Park Forest to Joliet was less than itwould have been from Baltimore to Aberdeen.We came here to Buffalo and looked things overhere and we made the decision to come here.There are many ways of looking at a decision.I’ve always believed that once you make up yourmind you need to be happy with your decisionand not look back and say, ‘‘Well, what if?’’ Idon’t think that’s productive. We really hada hard time deciding, but once we decided, thedie was cast and we were really very happy here.There were good schools for the children, andthere was an opportunity for me to go back toschool and I went to the University of Buffalo. Ihad a Johnny-come-lately career, which was nicetoo. I did that after the children were in highschool. That’s when I started to go to school my-self so I didn’t have to have guilt feelings aboutneglecting them. That sounds funny today, be-cause today women with small children workand have their careers. But in the 1940s and the1950s that was not the usual thing. I knew thatmy obligation was to be a mother and a wife fulltime and it was when we were in Illinois that Igot my first part-time job. I saw an ad in thenewspaper for a copy editor for a publishingcompany, and I got the job editing college-leveltextbooks in economics and management. Itwas very good for my self-esteem to have a jobthat somebody was willing to pay me for afterall those years and one that I felt qualified todo. I didn’t have to say, ‘‘Well, all I’ve had arethree hours of college journalism,’’ because I’dhad 60 hours of college journalism, literature,and writing. I felt qualified and that was good.



annette stein interview (mors) · and annette stein. mr. arthur ÔÔartÕÕ stein,...

Documents