50 Years of Army Computing
From ENIAC to MSRC
Thomas J. Bergin, editor
A Record of a Symposium and Celebration
November 13 and 14, 1996
Aberdeen Proving Ground
Sponsored by the Army Research Laboratory and
the U.S. Army Ordnance Center & School
Approved for public release; distribution unlimited.
This volume is dedicated to Michael John Muuss (1958 -
2000 ), whose video record of these precedings made this
publication possible. His loss is felt personally by many of
the participantshe recorded, but more poignantly his loss
is felt universally, for his talent, his spirit, and his wisdom
cannot be replaced. He was truly a national treasure.
Technical editing and design by Barbara Collier, ARL Technical Publishing
U.S. Army researchers played a fundamental role in the inauguration of the
modern computer age. The urgent need for high-performance computing during
World War II led the U.S. Army Ordnance Corps to fund the design and
implementation of the world's first high-speed electronic automatic computer,
the ENIAC. The price tag for this groundbreaking invention was $486,804.22,
but its lasting impact on Defense science and technology, and indeed on the
world today, is incalculable.
What began with our early technology pioneers, both military and civilian,
continues today with numerous contributions in hardware, software, networking,
and computational methods. The Army Research Laboratory, the Army's premier
research organization, celebrates its heritage with the publication of this
volume. I hope it is a welcome addition to your library.
Robert W. Whalin, PhD, PE
U.S. Army Research Laboratory
In November 1996, a symposium was held to commemorate the 50th anniversary of
the dedication of the ENIAC. The commemoration also marked another important
milestone in the history of Army computing, the ribbon-cutting ceremony for the
Army Research Laboratory Major Shared Resource Center (MSRC). This large
high-performance computing facility features heterogeneous supercomputing
systems, massive near-line storage, robust high-speed networking, and
scientific visualization with video production capabilities. The symposium
provided a wonderful opportunity to recognize the contributions and dedication
of the Army computing pioneers and to trace the growth of an industry and its
impact on the Army over 50 years. The accomplishments of the early computing
pioneers, many of whom participated in the proceedings, are an indelible part
of the legacy of the Army research.
As the successor to the organizations described in this book, the ARL
Computational and Information Sciences Directorate proudly carries forward the
tradition of its distinguished predecessors. On behalf of our Directorate and
the ARL MSRC, I am pleased to present this volume as a historical perspective
on the Army's role in the birth of the computer age.
Dr. N. Radhakrishnan
Chief, Computational and Information Sciences
U.S. Army Research Laboratory
A symposium and celebration was held at Aberdeen Proving Ground (APG),
Maryland, in November 1996, to recognize and commemorate seminal Army
contributions to the birth and development of modern computing. Primarily
inspired by the 50th anniversary of the invention of the world's fi rst general
purpose electronic computer (the ENIAC), this two-day event also celebrated the
dedication at APG of significant new computational resources provided by the
Office of Secretary of Defense. On this occasion, scores of "computing
pioneers" gathered at APG to reminisce about the accomplishments that stemmed
from the Army's computation needs during World War II-in particular, the need
for the firing and bombing tables that were essential for accurate targeting of
groundand air-delivered ordnance.
How did this grand celebration happen? Pretty much by accident! In August of
1995, a letter arrived at my home from Dr. Judith Rodin, President of the
University of Pennsylvania. This was a generic letter to "parents of
University of Pennsylvania students" which, much to my interest, announced an
upcoming celebration of computing to be held in Philadelphia. The focus was to
be the 50th anniversary of the invention of the ENIAC, and the festivities were
to be sponsored jointly by the University and by the Association for Computing
Machinery (ACM). As part of the event, Vice President Al Gore would
ceremonially activate a small piece of the original ENIAC.
At the time of the Rodin letter, many of us at the Ballistic Research
Laboratory (BRL) had just been sensitized to our own Army ENIAC history. In
September 1992, the old BRL, along with a number of sister laboratories, was
reorganized into the newly established Army Research Laboratory (ARL). During
a contemporaneous retrospective1 the history of Army computing at APG was
reviewed, including the concept, design, and construction of the ENIAC,
activities sponsored and managed by the Army. To ensure that the Army would
not be overlooked at the Philadelphia celebrations, I contacted Dr. Rodin's
offi ce. Fortunately, both the University and the ACM were happy to refi ne
the planned program to give the Army the credit it deserved. Over the next few
months, ARL worked with contacts at the University of Pennsylvania (Mr. Steve
Brown and Dr. Greg Farrington) and with the ACM (Dr. Bert Herzog, Dr. Dianne
Martin, and Dr. Tim Bergin); fi nally, we met with Dr. Herman Goldstine
himself, who had been the Army technical representative overseeing the ENIAC
project 50 years earlier.
One result of these activities, due particularly to Dr. Bergin, was that an
Army history panel was added to the ACM History Track. And thanks to Dr.
Farrington, Dr. Goldstine assisted Vice President Gore in the ceremonial
restarting of the ENIAC. In addition to the assistance of Dr. Bergin,
enthusiastic support was received from Dr. William Moye (ARL Historian), Mr.
Harry Reed (former BRL Division Chief), and Mr. Mike Muuss (Senior Computer
Scientist at ARL). After the Philadelphia celebration, those of us who had
witnessed the event were so impressed with the importance of the Army story
(and the signifi cant contributions of our computing pioneers) that we
dedicated ourselves to sponsoring an APG-based symposium. For this event, we
wanted to focus particularly on the people and activities that had typically
been ignored by the extant articles and monographs on the history of computing.
We additionally wanted to celebrate a new beginning with the dedication of the
DoD-sponsored ARL Major Shared Resource Center. With the encouragement of Dr.
John Lyons, then Director of ARL, an ad hoc committee was formed to bring about
the celebration documented in these pages. The activities of the committee
were supported by Major General Robert Shadley, Commanding General, U.S. Army
Ordnance Center and Schools, and his staff, who sponsored the military review
and award presentation on the second day of the celebration.
Although more than three years have passed since this event, time has not
dimmed the importance of the role played by the Army. We are pleased that
these pages can finally be shared with the Army family and all those who may
be interested in the roots of 20th century technology. In no small measure,
credit for these pages goes to Dr. Bergin, Dr. Moye, and Dr. Barbara Collier
(ARL Technical Publishing Branch). Finally, thanks go to Dr. Robert Whalin,
Director of ARL, and to Dr. N. Radhakrishnan, Director of the ARL Corporate
Information and Computing Directorate, for ensuring that this story was
Paul H. Deitz, Ph.D.
US Army Materiel Systems Analysis Activity
392 Hopkins Road
Aberdeen Proving Ground, MD 21005-5071
1 For insights into early ballistic
studies at BRL, see Klopcic
and Reed (1999).
My part in this story began in the summer of 1995, when a colleague asked if I
would serve as the "liaison to some people working on the ACM's Computer
Science Conference." When she put it that way, how could I refuse? Over time,
I found out that the ACM1 was about to celebrate its 50th anniversary and
wanted to have a commemorative program on the history of computing. After much
e-mail, I decided to have a "history track" for the Conference, as well as a
half-day commemorative program on the day before the Conference.
My fi rst task was to invite a number of prominent computer historians to a
twoday planning meeting in Arlington, Virginia. At this meeting, we outlined
the commemorative program, assigned historians to topics, and outlined four
panels for the history track: (1) "Hardware History," (2) "Software History,"
(3) "The ENIAC," and (4) "Antecedents of Personal Computing."
In November, I received e-mail from David Rutland, who had worked at the
National Bureau of Standards and wanted to propose a panel on NBS's early
computer efforts. So the fi rst history track in the history of the ACM now
had fi ve panels, including "Early Electronic Computing at the NBS," which had
Harry Huskey2 as a speaker.
By now, dear reader, you may be asking, "Why is he telling me all this?" Well,
in December, I received a phone call, at home, on a Sunday evening, from Paul
Deitz, who identifi ed himself as the Chief of the Army's "Vulnerability and
Lethality Division." That title sounded ominous, so I decided to try to help
this fellow. He said that he had a story he wanted to tell about Army
computing, and he had learned that the Association for Computing Machinery was
having a history track at the 1996 Computer Science Conference to be held in
Philadelphia. Paul asked if there would be room for a panel that explored the
early days of Army computing at the Aberdeen Proving Ground. I told Paul that
it sounded like a wonderful idea, but that I would have to get approval from
the Chair of the Program Committee for an additional room. In two days, we had
an agreement for a sixth panel, entitled, "The Army, the National Need, and the
After the ACM Conference ended, all those involved agreed that it had been a
wonderful conference, and that the History Track and the History Commemorative
Program had added much to its success. Indeed, after the fi nal panel, "The
Army, the National Need, and the ENIAC," Paul Deitz, Harry Reed,4 Bill Moye,5
and I talked about the need to do more. The Army did indeed have a story to
tell, having been a pioneering organization in the earliest days of computing.
Most importantly, we agreed that the people who worked at Aberdeen needed to be
recognized and given an opportunity to tell their story.
Well, through the efforts of Paul Deitz, Harry Reed, and Bill Moye, a Program
Committee was assembled to plan Fifty Years of Army Computing, from ENIAC to
MSRC, the Commemoration documented in this volume. The results of their hard
work were presented on November 13 and 14, 1996, at the Top of the Bay (Offi
cers' Club) at Aberdeen Proving Ground. The Commemoration honored the memory
and accomplishments of the earliest computer pioneers at Aberdeen, as well as
celebrating the future of Army computing with the dedication of the Major
Shared Resource Center.6
The Commemoration also included two exhibits, a photographic display and a
collection of artifacts. The photographic display was created for the ACM
Conference by Sharon McCullough and her colleagues at Special Events, Inc.
Sharon invited me to create an exhibit of artifacts to accompany the
photographs (possibly inspired by a visit to my rather cluttered offi ce).7 The
two exhibits appeared at both the ACM Conference and the Commemoration.
On behalf of everyone who attended the Commemoration, I would like to thank the
members of the Program Committee: Ray Astor, Bill Barkuloo, Hal Breaux, Paul
Deitz, Tad Edwards, Bob Eichelberger, Carol Ellis, John Gregory, Sharon
McCullough, Bill Moye, Mike Muuss, Charlie Nietubicz, Harry Reed, and Jill
I would also like to acknowledge the tremendous effort it took to plan and
manage the Commemoration itself. Without the efforts of the Administrative
Committee, we would not have had such a wonderful two days: Virginia Bailey,
Elizabeth Barber, Judith Celmer, Patricia Cizmadia, LouAnn Conway, Connie
Gillette, Rodger Godin, Dave Jennings, Judy Johnston, Thomas Kile, Angie
Levrone, Charles McDevitt, Ronald Mihalcin, Bob Reschly, Jr., Brenda Rice,
Kathryn Sorensen, and Edward Starnes. We owe all of them a big "thank you" for
a job well done.
And last, but not least, I want to thank Bill Moye, the ARL Historian. Bill
has worked with me through all the rough spots in this project and I appreciate
his assistance, support, and friendship.
I wish to apologize for the delay in the production of this volume. A week
after the conference, my wife, Diane, was diagnosed with ovarian cancer. I
completed the fi rst draft in late fall 1998, and received it back for fi nal
editing in February. Diane died on March 5, 1999, and I have had a diffi cult
time fi nding the time and focus needed to fi nish the job. I would be remiss
if I did not thank the people at Business Plus, Inc., especially Jon Morell,
for their support, patience, and understanding during the preparation of this
volume. I have been fortunate to have worked through this fi ne company.
In addition, I would ask that all attendees look in their attics, basements,
and garages, to see if they have materials or artifacts to donate. The
Computing History Museum at American University is a direct outgrowth of this
Conference and the ACM Conference which preceded it. Indeed, Armand Adams, one
of the participants, has already donated artifacts from his personal
collection. The Museum is an opportunity for you to continue to contribute, by
sharing your memorabilia with present elementary, secondary, and college
students. Please contact me if you have something to donate.
Finally, I want to reiterate what I learned many years ago: Wonderful things
happen when a group of dedicated people works together to achieve a goal. This
was true for the pioneers who were honored during these two days, as well as
for the people who worked so hard to make this Commemoration possible. It was
an honor to work with all of you.
Thomas J. (Tim) Bergin, Ph.D.
Director, Computing History Museum
Professor, Computer Science and Information Systems Department
4400 Massachusetts Avenue, NW
Washington, D.C. 20016-8116
1 Association for Computing Machinery.
2 ENIAC team member. See Presenter Biographies, this volume, p 138.
3 A transcript of this panel is included in this volume (pp 142-158).
4 ENIAC team member. See Presenter Biographies, p 139.
5 ARL Historian. See Presenter Biographies, p 139.
6 A Major Shared Resource Center is a Department of Defense high-performance
computing asset available to the DoD R&D community. See also Bergin and Moye
7 With the addition of some first-rate artifacts donated by members of
the Program Committee, as well as by attendees at the Army Commemoration,
this exhibit became the backbone of the Computing History Museum
at American University. See "ENIAC 50th anniversary continued,"
in "Happenings," IEEE Annals of the History of Computing 18, No. 3 (Fall
1996), p 75.
1. Opening Session: History of Early Computing.................... 8
Announcements................................... ................ 10
Keynote Address: A Short History of Computing or
How Did We Get to Aberdeen?...................................... 12
2. ENIAC: Development and Early Days..............................24
3. Women Pioneers.................................................38
4. Digital Computing at BRL: 1938-1969............................50
1939-1954: ENIAC and the First Computer Survey....................51
BRLESC I and II...................................................58
ENIAC Put to Work.................................................66
5. The Early Computer Industry....................................74
6. Recent History: Supercomputers and Networking..................86
7. Military Ceremony.............................................102
8. Civilian Recognition and MSRC Ribbon-Cutting Ceremonies.......104
Appendix. ACM History Track Panel: The Army, the National
Need, and the ENIAC..............................................142
Epilog. High-Performance Computing at ARL........................160
Side Bar Articles
Mauchly and Eckert............................................... 26
Adele Katz Goldstine and stored-program machines................. 27
"Intermediate programming"-converter code........................ 30
His engines...................................................... 32
Bletchley Park................................................... 33
The "first draft report"......................................... 34
How a core memory element works.................................. 56
Machine language................................................. 62
FORTRAN and FORAST............................................... 62
JK switches...................................................... 63
ENIAC at BRL: Homé McAllister, Winifred
(Wink) Smith, George Reitwiesner,
and Ruth Lichterman............................................... 1
Mauchly, Barnes, and Eckert peruse ENIAC
manual in February 1946........................ .................. 2
Console of BRLESC I computer.......................................3
ARL Major Shared Resource Center.................................. 4
Photo exhibit at 50 Years of Computing celebration................ 9
Computer technology stamp........................................ 11
Martin Weik's "computer tree".................................... 53
Truman visits ENIAC, 1951........................................ 66
Muuss and Weaver with PDP-11 displaying wire
frame diagram of XM1 tank........................................ 92
Uncrating the Cray Research XMP.................................. 95
The Ordnance Corps honors two of its own................... 102, 103
Barnes and Brainerd in front of ENIAC
Function Table "A".............................................. 109
Ribbon-cutting for ARL MSRC..................................... 113
Speed of networks versus time................................... 118
Speed of supercomputers versus time............................. 119
Computational grid.............................................. 122
Flow field visualization........................................ 122
Computational grid for transonic computations................... 123
Pressure contours on a wrap-around finned projectile............ 124
Image of steel rod penetrating steel plate,
by obtained flash radiography................................... 125
Modeling long-rod penetration with combined yaw
and obliquity................................................... 126
Tank simulation: medium resolution.............................. 128
Applying our expertise to new areas............................. 129
Terrain model showing quantization error ("steps").............. 130
Model of terrain at Fort Hunter-Liggett......................... 131
Corps Main Command Post: detail of truck........................ 131
Energy versus frequency for one pixel in a sensor............... 132
Scene generation elements....................................... 133
Model of tank in smoke.......................................... 133
Software simulation back plane.................................. 134
Energy transport: basic sensor scenario......................... 134
Simulation showing effect of distance on color.................. 135
Barkley Fritz, Herman Goldstine, and Harry Reed
at ACM Meeting.................................................. 144
A simple set of trajectories.................................... 155
Firing table for 105-mm howitzer................................ 156
Quadrant elevation table........................................ 157
This volume is a record of a conference held at Aberdeen Proving Ground,
Maryland, on November 13 and 14, 1996. The conference sessions included talks
and panel discussions on topics ranging from the earliest days of computing at
the Ballistic Research Laboratory (BRL) (now part of the Army Research
Laboratory, ARL) to projections for the future use of ARL's Major Shared
Resource Center, which was dedicated during the event. As part of the
dedication ceremony, awards were presented to honor three of the most important
pioneers, Herman Goldstine, Paul Gillon, and John von Neumann.1
Much of the following material was included in the conference program, along
with some of the accompanying photographs. Because of its historical nature,
we decided to include this program material here for readers who did not attend
The Ballistic Research Laboratory of the Ordnance Department, established in
1938 from the Research Division of Aberdeen Proving Ground (APG), was charged
to produce firing tables for the Army. For artillery, for example, these
tables showed the soldier what angle of elevation was required for a specific
projectile to impact a target at a specified range with a given propellant
charge. The tables also indicated corrections to apply for variations in
atmospheric temperature, air density, wind, angle of sight, weight of
projectile, muzzle velocity, and compensation for drift. Especially in
wartime, firing tables had to be prepared and sent to the field as rapidly as
possible, because without the information, artillery became less effective.
To speed the calculation of firing and bombing tables, BRL's predecessor
organization acquired its copy of the Bush differential analyzer in 1935.
Early in World War II, the lab also contracted with the Moore School of
Electrical Engineering, University of Pennsylvania, to take over operation of
the school's somewhat faster differential analyzer. Hoping to improve on the
Bush analog device as a means to generate firing tables, John W. Mauchly and
J. Presper Eckert, Jr., proposed building an electronic numerical analyzer.
Two Ordnance Department officers, Colonel Paul N. Gillon and Captain Herman H.
Goldstine, recognized the potential of the proposal to build an electronic
computing machine, nurtured it, found money to support the project, defended it
against critics, and helped publicize its achievements.
A contract was signed in June 1943, and construction began in June 1944, with
final assembly in the fall of 1945 and the formal dedication in February 1946.
Later that year, ENIAC was dismantled, and it was delivered to APG in January
1947. It was operational again in August 1947, representing "the largest
collection of interconnected electronic circuitry then in existence."2 With
refinements suggested by John von Neumann and others, ENIAC provided years of
successful service until it was retired in October 1955.
By this time, BRL had acquired two more computers. Beginning in the fall of
1944, the ENIAC team, working with von Neumann, designed EDVAC (Electronic
Discrete Variable Computer). The new device, a collaborative effort by BRL,
the Moore School, the Institute for Advanced Studies (IAS), and the National
Bureau of Standards (NBS), was the first computer to be designed with an
internally stored program. EDVAC was installed at BRL in 1949, but design
problems delayed acceptance and practical operation until 1952.
BRL was already at work on a new system. ORDVAC (Ordnance Variable Automatic
Computer) belonged to the group of computers whose basic logic was developed by
the IAS. It was built by the University of Illinois and brought to Aberdeen in
1952. Thus, for a brief time in 1952, with ENIAC, EDVAC, and ORDVAC, BRL was
the world's largest computer center.
By 1955, available time on each computer had been pushed to 145 hours per week
of error-free production to support ballistic research and compute firing
tables and other ballistic data for artillery, rockets, and missiles. The
computers also performed calculations in other fi elds, including weather
prediction, atomic energy research, thermal ignition, cosmic ray studies, and
wind tunnel design.
In 1956, engineers and scientists in the BRL computing laboratory began to
develop a new computer, to be called BRLESC (BRL Electronic Scientific
Computer). At the same time, the Ordnance Department transferred money to NBS
to develop logic modules-the arithmetic, logical, and control units for the new
system. BRLESC went on line in 1962, and tests indicated that it was two to
eight times faster than commercial systems.
Planning for BRLESC II began in 1965. BRLESC II was a solid-state digital
computer designed to be 200 times faster than the ORDVAC, which it replaced in
November 1967. The integrated circuits for BRLESC were produced under an
industrial contract, but BRL employees did all the logic design, back-panel
wiring, and assembly. To facilitate use of these increasingly powerful
machines, BRL personnel pioneered software. FORAST (Formula and Assembly
Translator) was developed for ORDVAC in 1960 and implemented on BRLESC by 1961.
During 1962-63, BRL wrote a FORTRAN (Formula Translation) compiler for BRLESC,
one of the first uses of the FORTRAN language on other than IBM computers.
DoD directed BRL not to design and build any more computers inhouse, primarily
because of the extensive capabilities achieved by the commercial computer
industry. Even so, lab personnel continued to experiment with computer
hardware, software, and operations.
For example, during the 1970s, the lab worked with Denelcor, Inc., on its
Heterogeneous Element Processor (HEP), the world's first massively parallel
supercomputer. A 64-bit, floating-point digital machine with considerable
multi-user and multi-task possibilities, the HEP was the first supercomputer
to run the Unix operating system. In the late 1980s, BRL dedicated two of the
Army's first supercomputers, a Cray X-MP/48 followed by a Cray-2.
From the early 1970s, BRL employees had a prototype BRLNET up and running, and
by the later 1970s, they were pursuing the tools to permit localarea
networking. In the early 1980s, the lab played a major role in working with
the Defense Advanced Research Projects Agency (DARPA) to develop the ARPANET.
By this time, BRL had adopted the Unix operating system, implementing many
modifi cations along the way. For example, an engineer at the lab contributed
substantially to the specifi cations for the ANSI version of the C programming
These tremendous tools greatly enhanced research capabilities in such areas as
simulation, virtual reality, and scientifi c visualization. By the early
1980s, lab personnel were developing three-dimensional graphics display
hardware to assist in the development of combinatorial-solid-geometry
descriptions ofmilitary vehicles. In 1991, the lab began shipping BRL-CAD
release 1.0, which included a network-distributed image-processing capability.
The scientific visualization program was started in 1984 to provide tools and
expertise to help researchers graphically interpret the voluminous results of
scientific supercomputer calculations. Visualization techniques provide
threedimensional, color representations of calculated variables that
characterize ballistic phenomena such as density, pressure, temperature, and
Now, under the Secretary of Defense's High-Performance Computing Modernization
Program, ARL has been designated a Major Shared Resource Center (MSRC),3 one of
four being funded by the Modernization Program. ARL already operates an
8-node, 96-processor Silicon Graphics array system that became operational in
April 1995, and a classified Cray-2 system that was made available to the DoD
user community in June 1995. DoD awarded the contract for the ARL MSRC to
Raytheon E-Systems-an eight-year integration effort with a total life-cycle
value of more than $150M. These computational assets will greatly enhance
Defense research and development capability, especially in the technology areas
of structural mechanics, fluid dynamics, chemistry and materials, forces
modeling, nanoelectronics, electromagneticsrt, and acoustics, signal image
processing, and simulation and modeling.
On November 14, 1996, special awards were presented to honor computer pioneers
Herman Goldstine, Paul Gillon, and John von Neumann. Herman Goldstine was
presented with the U.S. Army Distinguished Service Medal in a military
ceremony (see pp 102-103); later he was also awarded the Decoration for
Distinguished Civilian Service at a Recognition and Dedication Ceremony, at
which the computer pioneers were recognized and the new Major Shared Resource
Center was dedicated (pp 104-113). Paul Gillon and John von Neumann,
represented by their families, were also posthumously recognized at these
ceremonies for their significant achievements.
Herman Heine Goldstine.
During World War II, as an Ordnance officer assigned to BRL, Goldstine played a
major role in the development of ENIAC. Initially, Goldstine was put in charge
of the BRL section at the Moore School of Electrical Engineering, University of
Pennsylvania, which operated a Bush differential analyzer and produced firing
tables for the Army. At the Moore School, he met John W. Mauchly and J.
Presper Eckert, Jr., and helped them develop their plan for an electronic
computer. After the Ordnance Department signed the contract with Moore School
to build the ENIAC, Captain Goldstine served as the Army's on-site technical
representative overseeing the project.
While at the Moore School, Goldstine helped develop the operating routines for
the ENIAC, and he also helped develop the original plans for the Army's second
computer, the EDVAC.
When released from the service in 1946, Goldstine joined von Neumann at the
Institute for Advanced Study, working on the IAS computer project (partly
funded by the Ordnance Department). During the late 1940s, Goldstine, von
Neumann, and others wrote a series of reports on the logic and operation of a
stored-program computer, very largely defi ning the structure of modern
The IAS group fathered a generation of machines, including the ORDVAC,
installed at BRL in 1952. ORDVAC was one of several direct descendants of the
IAS machine, but there were also several collateral descendants, perhaps the
most important being the 700 and 7000 series of IBM machines and the UNIVAC
1100 series from Sperry Rand.
In 1985, President Reagan presented Goldstine the National Medal of Science,
recognizing "his fundamental contributions to the development of the digital
computer, computer programming, and numerical analysis."
Paul Nelson Gillon.
A 1933 graduate of West Point, Gillon earned his master of science degree from
MIT in 1938. In 1939, he was assigned to BRL to work with R. H. Kent.4 In
1940, he was named Executive Offi cer of BRL, and the next year, he was named
Assistant Director, following Lieutenant Colonel Leslie E. Simon. In 1942, he
was assigned to the Office of the Chief of Ordnance as Deputy Chief of the
Service Branch, Technical Division. At BRL, Gillon pushed Simon and Colonel
Herman H. Zornig to approach IBM to obtain a set of punch card machines to
support the BRL Bush differential analyzer. In 1942, he contracted with Harold
Pender (Dean of the Moore School) to take over operation of the Penn Bush
differential analyzer. He put Goldstine in charge of the BRL operations at
Penn, and he supervised the ENIAC project, visiting Thomas J. Watson (then CEO
of IBM) and Oliver E. Buckley (then President of Bell Telephone Laboratories)
to request help. In 1944, he was awarded the Legion of Merit for his R&D
management accomplishments. Gillon was later the director of research at the
Watertown Arsenal and the commander of the Ordnance Research Office (1954-56).
John von Neumann.
In August 1944, Goldstine told von Neumann about the ENIAC project and showed
him the machine. That same month, von Neumann attended a meeting of the BRL
Scientific Advisory Committee when it was decided to proceed with a second
machine, the EDVAC. In the spring of 1945, Goldstine circulated copies of von
Neumann's "First Draft of a Report on the EDVAC," a seminal document in the
design of computers. Later that year, at von Neumann's instigation, Nicholas
Metropolis and Stanley Frankel journeyed to Philadelphia to run calculations
for the Manhattan Project on ENIAC.
Back at the Institute for Advanced Study, von Neumann launched his Electronic
Computer Project. In the fi rst conceptual paper on an internally programmed
computer, "Preliminary discussion of the logical design of an electronic
computing instrument" (June 1946), Arthur W. Burks, Goldstine, and von Neumann
issued a classic report that profoundly infl uenced all subsequent computer
developments. Among the family of machines built following the IAS concepts
was ORDVAC, installed at BRL in 1952. Meanwhile, in 1948, after ENIAC had been
reconstructed at BRL, von Neumann helped reprogram the
[...for some reason the narrative stops in mid-sentence for this part of the
November 13, 1996
1. Opening Session: History of Early Computing
Good morning, ladies and gentlemen. I'd like to welcome all of you here on
this very wonderful two-day celebration. My name is Paul Deitz; I'm with the
Army Research Laboratory. For some of you who remember the old days, I used to
be with the Ballistic Research Laboratories.
It's a very exciting time for us. We're actually here in part because our
friends in Philadelphia had a wonderful celebration last February.1 They
reminded some of us younger folks about a bit of our heritage that we hadn't
really truly forgotten but weren't as well informed of as we might have been.
There are of course two major groups that sponsored the celebrations last
February: the University of Pennsylvania and the Association for Computing
Machinery. Dr. Greg Farrington of the University of Pennsylvania will be here
tomorrow, and his fine staff welcomed us into their midst.
A number of folks at the Association for Computing Machinery were really
instrumental in our being there. Among these were Dr. Bertram Herzog, of the
University of Michigan, who I think is going to be here later; Dr. Frank
Friedman, who is Chairman of the Computer Science Department at Temple
University; and Dr. Tim Bergin, of the American University, who is up on stage
with us and whom you'll meet in just a few minutes. Tim, who is a historian,
had already put together a history track for the ACM meeting and very kindly
extended the program so that we could have our own special session devoted to
the work at Aberdeen Proving Ground (APG).
Through that experience, some of us younger folks got to meet some of the
earlier workers and pretty much fi xed in our minds that we needed to plan
something like this commemoration, from our own somewhat parochial perspective
here at APG. So here we are, and I welcome you as guests of the Army Research
Laboratory and the Ordnance Center and School. General Shadley has provided
wonderful support. Tomorrow we're going to see a wonderful event: there will
be a parade in review tomorrow morning and we'll all be going over there by
bus. After lunch we will have our civilian ceremony.
What are really the foci for this celebration? Some of us felt that it might
be useful to help the public remember that it was the Army that initiated the
computer revolution. I know that many of these ideas are arguable, but we'd
like to think that the Army had a lot to do with this. It's fairly clear that
very few inventions have had as large an impact on our civilization as the
computer. Modern computers are pretty much descended from the ENIAC, the
EDVAC, the ORDVAC, and the BRLESC machines, which were all a part of the early
history of the Army's Ballistic Research Laboratory here at APG. These
machines weren't built for some abstract need, but were conceived and built to
solve specifi c military problems. We're going to hear a lot more about what
those problems were, later on today.
1 The Association for Computing Machinery held its Computer Week in
Philadelphia in February 1996 to commemorate the 50th anniversary of computing.
The History Track, chaired by Tim Bergin of American University, had a panel on
"The Army, the National Need, and the ENIAC." The panel was chaired by Paul
Deitz, with Herman Goldstine, Harry Reed, and Barkley Fritz as panelists. A
transcript of this panel is provided on pp 142-158 (this volume). 9
The second focus is to attempt to give credit to the highly skilled and
dedicated military and civilian scientists and other workers who, along with
their counterparts in the private sector, solved a great national defense
problem. They had specifi c objectives and, ironically, the very fi rst
problem that the ENIAC was used to solve was not what the machine was built
for. As many of you know, it was fi rst used to perform hydrogen bomb
calculations for Los Alamos.2 Of course the machine was designed to compute fi
ring tables and bombing tables for BRL. So the ENIAC met a tremendous defense
The third focus is to capture this information and document it. That's part of
what this celebration is about, so we're fi lming and taping these sessions.
We are going to endeavor, later on, to put together a compilation of the
transcripts of what is said.
To my right are some wonderful exhibits.3 In addition, we have provided lots of
materials in your registration kit, so you can go home and at your leisure read
up on the many fascinating details.
Our guiding principle was to look backwards. However, another very happy event
was that the Army Research Laboratory was chosen as one of the four Department
of Defense sites for a Major Shared Resource Center (MSRC). So tomorrow
afternoon, there will be a panel on ARL efforts in supercomputing and the
dedication of the MSRC. Some folks around here that you'll meet later on,
including Charlie Nietubicz and Harold Breaux, have made major contributions
toward making this event happen and in helping to continue our proud ARL
Just a couple of related administrative issues. The principal focus today is
not on the people with the red badges-those are committee members. If you have
a problem, I suppose you would see one of us wearing the red badge. The people
that are really important are those wearing the blue badges; these are our ARL
computer pioneers. So if you're wearing a blue badge, get to know one of the
younger people; if you're not wearing a blue badge, get to know somebody who
We're trying to be as inclusive as possible. Two days seems like a long time
to do a program like this. However, when you start to look at the magnitude of
the set of events that happened here at APG, two days isn't really very much
time. So, we're inevitably going to miss important things and underestimate
the importance of various contributions. We really do want to try to be
inclusive, but that may only happen later on, as we put together our written
compilation of the day's activities.
So I'd encourage all of you who have another view, or who have photographs or
different ideas, to talk to somebody back there at the table and we'll do
everything we can to ensure that this additional material is included in the
final written record. Let's have a very good day.
I'd now like to introduce Dr. William Moye, who is the ARL historian. He has
been one of the really active people in seeing that this event took place.
I have just a few short administrative announcements. We're very glad to see
all of you here this morning. I'm told that the heat is on and it should be
warming up. [laughter] You're welcome to bring your coffee in here. As most of
you know, there's coffee and doughnuts in the room immediately behind us here,
so you're welcome to bring your coffee in.
Some of you took the buses from the motel this morning. There will be buses
this afternoon going to the APG Ordnance Museum for the social. There will
also be buses again in the morning. So if you didn't know about the buses, be
advised that there is transportation from the Sheraton coming over here to the
Top of the Bay [offi cer's club]. And if you didn't sign up for the social and
you want to come, we can still do that.
You probably found the rest rooms. They're basically down this hall and down
the outside hall. There's a pay phone in the coat room. There's also a pay
phone upstairs. We will be eating lunch upstairs. I hope you picked up your
bag at registration. Please remember that there's a coffee mug in there, so
please try not to break that. It has the little logo on it.
I want to thank Ray Aster, who is sitting here in the front row. Ray is one of
the APG retirees. He brought the large replicas of the computer technology
stamp that are over here by the registration table. [see sidebar] There will be
personnel from the Postal Service here at lunch time, both today and tomorrow,
selling the stamps.
Sharon McCullough is walking around here somewhere. She's the one from Expert
Events. They put together the big display on the ENIAC. This is the same
display that many of you saw up in Philadelphia at the ACM Conference in
February. Many nice pieces. I think a lot of you see yourselves in some of
those pictures in Sharon's display. There's also a nice display against the
back there on the Scientifi c Advisory Committee. Some of you will recognize a
lot of the BRL staff, as well as the scientifi c staff of the Scientifi c
Advisory Committee. There are also some video kiosks showing some of the later
computer developments here at ARL.
Tim Bergin, who Dr. Deitz referred to a minute ago, provided a lot of the
material and certainly provided all the little identifi cation pieces in the
display case over there. Some of the rest of you provided some of the booklets
and items. Tim had a different array of items up at the ACM Computer Science
Conference in Philadelphia. This is a very nice display of artifacts,
pamphlets, and manuals on some of the early machines. Tim is a professor at
American University, and he is our next speaker.
Keynote Address: A Short History of Computing or
How Did We Get to Aberdeen?
The history of technology in general, and of specifi c technology such as the
automobile, the television, and the computer, tells us that a given technology
springs not from a single idea, but emerges over time from the work of numerous
individuals- some of whom are known to us, and some whose contributions are
lost to history. Computing is a perfect example of this phenomenon.
Although we're here today to pause and refl ect on the role of the U.S. Army
Ordnance Department, and its successor organizations, in the development of the
electronic digital computer, the Program Committee asked me to set a context
for the day by identifying important milestones in the evolution of computing.
I will also try to relate this to the Ballistic Research Laboratory's use of
mechanical and electrical calculating equipment, punch card data processing
equipment, analog computers, and early digital machines.
Before beginning, we might all stop and ask the question, why? Why was it that
the Ordnance Department at Aberdeen Proving Ground got involved in computing?
I'll fi rst give you a short answer to the question. The longer answer is the
actual lecture itself.
I first met Harry Reed in Philadelphia, last February, at the Association for
Computing Machinery's annual conference. Harry was an early member of the team
at the Ballistic Research Laboratories. And in the course of a discussion
after the panel, Harry told me, "You don't fire a weapon unless you know where
the shell is going to land." Although that might seem self-evident to most of
you in this room, it seemed profound to me, as a college professor who did two
years, eight months, and 21 days in the Army Medical Corps in the early
sixties. So the important question is "how do we aim the weapon?"
Harry also shared some material with me, and after reading over some of it, I
realized that "interior and exterior ballistics" are topics that are simply
over my head. But I do know that the process of calculating firing tables is
complex and time consuming. The number of individual calculations needed to
prepare a firing table in the 1930s was staggering.4 And in 1941, the United
States entered the war.
The need for fi ring tables for existing weaponry, as well as newly developed
weaponry, was overwhelming. The men and women of the U.S. Army and Navy were
desperate for ways to improve the quality of fi ring tables and the timeliness
of their delivery. There is some data that has been fl oating around from
multiple sources, and it's actually on some of the display panels, that a
person with a desk calculator could compute a 60-second trajectory in about 20
hours. The differential analyzer, which I'll mention, could compute a
trajectory in about 15 minutes. The ENIAC could compute the trajectory in
about 30 seconds, or about half the time that the shell would be in the air.
Given these figures, it's easy to see that a firing table for a single weapon
could take a number of skilled workers a number of weeks to complete using desk
calculators. If you've never seen a firing table, there's one in the exhibit
According to Ballisticians in War and Peace, A History of The United States
Army Ballistic Research Laboratories (Volume 1, 1914-1956, p 38), the Aberdeen
differential analyzer computed 1560 production trajectories used in the
preparation of 10 firing tables, between February 1 and June 30, 1945.
Obviously the use of punched card machinery and hand-operated adding machines,
as in the 1930s, was much more labor intensive, and took longer periods of
In a nutshell, that is why it was the Ordnance Department that led the
way-because they had the need. Interestingly enough, the panel that Paul Deitz
put together for the ACM's February meeting was entitled "The Army, the
National Need, and ENIAC."
During the rest of this lecture I will try to identify some of the precursors
of the modern computer and highlight the important role played by the Ballistic
Research Laboratory, Aberdeen Proving Ground.5
Punched Card Technology
Although the punched card had been invented in the 1880s by Herman Hollerith
for the U.S. Census Bureau, the first really scientific use was by L. J.
Comrie in 1928, in the United Kingdom. Comrie used punched card technology to
calculate the motions of the moon, but he was not alone in his endeavors.
Wallace Eckert of Columbia University also experimented with punched cards. In
1929, Eckert convinced the IBM Corporation to fund the Columbia University
Statistical Bureau, which would use IBM punched card machines for processing.
Eckert not only performed essential scientifi c calculations using this
equipment, but he proved the value of this technology to science.
Let me bring this discussion closer to home. In 1937, Colonel H. H. Zornig
of APG (later head of BRL) became interested in the capabilities of punched
card equipment and its potential application to the creation of firing tables
and other ordnance processes. By 1941, BRL was using standard IBM tabulating
equipment, as well as equipment specially modifi ed to BRL's specifi cations.
This equipment was used in the preparation of fi ring tables, as well as other
problems such as the theory of breech rings, fuze-setting coefficients, shock
wave study, and probability integrals.6
Scientists and engineers have been using analog devices for many years.
Indeed, for a lot of us in this room, our introduction to higher mathematics
was accompanied by the acquisition of a slide rule. I still have the slide
rule I got in 1958, when I went to college. When I show it to students today,
they all say, "What's that?" There is a military slide rule in the display case
that I found in Boston about five or six years ago. I've not yet found anyone
that can tell me how to use it to calculate a trajectory. So if there's anyone
in the audience who could help with that, I will be most appreciative.
Vannevar Bush, who later served as President Truman's science advisor, had a
large analog machine known as the "differential analyzer" constructed at MIT in
1930.7 This machine solved differential equations by mechanical integration.
In 1934, the Moore School of the University of Pennsylvania constructed a
similar, but more powerful machine based on the MIT design. In the same year,
Bush suggested to BRL's predecessor organization that they use a differential
analyzer for calculating ballistic trajectories. In 1935, a differential
analyzer was built at Aberdeen Proving Ground and used in firing table
According to Ballisticians in War and Peace, "the differential analyzer was the
most important tool acquired before BRL was formed."8 The success of the
Aberdeen analyzer marked the beginning of the development of specialized
computing facilities for ballistic computations of various kinds. Since the
Aberdeen analyzer could not keep up with the Army's needs, BRL arranged to have
access to the larger and more powerful differential analyzer at the Moore
School of Electrical Engineering of the University of Pennsylvania. Staff from
BRL were sent to Philadelphia to operate this machine. During this period, a
number of improvements were made to increase the speed, mainly by substitution
of electrical for mechanical components.
Starting about 1937, a number of projects began working on improving the speed
of computation. In Germany, Konrad Zuse began experimenting with mechanical
computing devices; in the United States, George Stibitz began experimenting
with telephone relays, John Atanasoff started construction of an electronic
calculator, and Howard Aiken designed an electrical analog of Babbage's
analytical engine; and in Britain, Alan Turing and others began the
construction of machines to decode messages.
Although unknown until after the war, a young German engineer, Konrad Zuse,
began experimenting in 1934 with a mechanical device to do calculations. 9 In a
lecture that Zuse gave at the Computing Museum in the early 1980s, he explained
that he was "lazy," and he didn't want to spend the time necessary to do the
calculations by hand. By 1938, Zuse had completed his first machine in his
parents' living room in Berlin. The Z1 had a mechanical memory capable of
storing 16 binary numbers, each with 24 bits. Zuse's control mechanism used
holes punched in discarded 35-mm movie fi lm. By 1939, he had completed his
second machine (the Z2), which used relay technology. By 1941, Zuse's Z3 was
operational. This machine used relay technology and had a 64-word
floating-point binary memory. 10 Zuse went on to
design several other computers and after the war was the founder of a
successful computer firm in Germany.
This was the "first fully functional program-controlled electromechanical
digital computer in the world" (Lee, 1995a, p 759).
John V. Atanasoff
Across the ocean in Iowa, another young man was bothered by repetitive
calculations, the waste of time that they entailed, and the problems of error.
John Vincent Atanasoff was a professor of physics at Iowa State University.11
In 1937, he started thinking about a machine that would use capacitors as a
memory. From 1939 to 1942, he and his graduate assistant, Clifford Berry,
built a special-purpose electronic digital calculator that later became known
as the Atanasoff-Berry Computer, or ABC. This machine had a regenerative
memory that could store 30 (50-bit) binary numbers and used digital logic for
computation (addition and subtraction). Atanasoff left Iowa State in 1942 to
work at the Naval Research Laboratories, and did not return to work on the
12 Following a patent dispute between the Honeywell and Sperry-Rand
Corporations, Atanasoff was designated the inventor of the digital computer
(U.S. District Court, District of Minnesota, Fourth Division, October 19,
1973). For an excellent discussion of this dispute, see Rosen (1990).
John Mauchly met Atanasoff at a meeting of the American Association for the
Advancement of Science in Philadelphia in December of 1940. After an exchange
of correspondence during the spring, Mauchly visited Atanasoff in June of 1941.
Mauchly, a professor of physics at Ursinus College, was experimenting with
devices to assist in weather calculations. (I would be remiss if I did not
mention here that faculty at Iowa State have started a project to build a model
of the ABC.)
George Stibitz was a mathematician working at the Bell Telephone Laboratories.
13 In 1937, he too was interested in calculation. He took some wire relays out
of a scrap pile and took them home to experiment with. Stibitz had noticed the
similarity between binary numbers and the on/off states of the (telephone)
relay. He created a small (binary) adder, which was later called the Model
"K"-for the kitchen in which it was fabricated. There is a model of the
Model-K in the display case, built by Raymon Richardson, a student at American
Stibitz showed this model to colleagues at Bell Laboratories, and a project was
started in 1939 in which Stibitz and S B Williams built the "complex number
calculator." Although not a computer, it was capable of complex arithmetic
operations. On September 11, 1940, Stibitz demonstrated the machine to the
attendees at a conference of the American Mathematical Society at Dartmouth
College in Hanover, New Hampshire. Using a teletype connected over special
telephone lines, attendees were able to type in mathematical problems. The
answers were received in a matter of minutes. Mike Williams, in A History of
Computer Technology, mentions that John Mauchly and Norbert Wiener both spent a
great deal of time experimenting with the relay calculation system.14
When the United States entered the War, the National Defense Research Council
(NDRC) asked Stibitz to work on some projects. The first project was to build
a gun director. Stibitz suggested, of course, a relay calculator, which was
later called the "relay interpolator." The machine was operational in September
of 1943. Stibitz's third relay computer was designed to test the accuracy of
antiaircraft gun directors and became known as the "ballistic computer." A
second ballistic computer, known as the Bell Laboratories Relay Calculator
Model IV (or Error Detector Mark 22 by the Navy), was built for the Naval
Research Laboratories in Washington, D.C.
Finally, in 1944, the U.S. government gave Bell Labs a contract to build two
identical machines, known as the Bell Laboratories General Purpose Relay
Calculators or Model V. The first of these was for the National Advisory
Committee on Aeronautics at Langley Field, Virginia; the second was built for
the U.S. Army's Aberdeen Proving Ground. These computers were so reliable
that people would submit problems at the end of the day shift, the problems
would run overnight, and they would get the output the next morning.15
15 Ballisticians (vol. I), pp 39 and 40 (with photograph).
Howard Aiken of Harvard University's Computational Laboratory was also working
on computing devices.16 Unlike the others, Aiken borrowed from many
technologies, including mechanical devices based on IBM punched card equipment
and the relay technology used by Stibitz. Aiken was also saddled with complex
problems, and decided he needed a machine that was at least an order of
magnitude better than anything that existed at the time. Knowledgeable of the
work done at the Watson Astronomical Computing Bureau at Columbia (by Wallace
Eckert), and quite aware of Charles Babbage's efforts a century earlier, Aiken
set out to design such a machine.17
17 Williams (1997), pp 154-186; Lee (1995a), pp 51-64.
Ultimately, Aiken convinced IBM to build a machine with financial assistance
from the U.S. Navy. The Automatic Sequence Controlled Calculator was
operational at IBM's Endicott facility in January of 1943, and was later
relocated to Harvard. Commonly called the "Harvard Mark I," the machine was
dedicated in May 1944. It contained 72 mechanical registers, each capable of
storing 23 decimal digits and a sign, and it was controlled by punched paper
tape. During the war, Aiken served as a Commander in the U.S. Naval Reserve,
and the Mark I was used by the Navy's Bureau of Ships and later by the Navy's
Bureau of Ordnance. One of Aiken's young assistants was Lieutenant Grace
Murray Hopper, who was to play a major role in the development of standardized
programming languages and rise to the rank of Rear Admiral.
Another of Aiken's assistants was Herbert R. J. Grosch (an attendee at this
18 For Grosch's recollections on the "Aberdeen machines," see Grosch (1991), pp
British Code-Breaking Efforts
During World War II, the British government maintained its code-breaking
establishment at Bletchley Park, outside London. Alan
Turing was one of the people working there.19 By April of 1943, Dr CE
Wynn-Williams had constructed a machine using mechanical relays and electronic
components. The machine was called the "Heath Robinson," after a cartoonist at
the time, who was known for designing strange and wonderful machines-like those
designed by Rube Goldberg, the American cartoonist. Later this machine was
surpassed by the Colossus, which used 1500 vacuum tubes (or valves as they were
known in Britain at that time), which was more than any other device except for
the ENIAC.20 A museum has recently been built at Bletchley Park to commemorate
these activities, many of which are still protected under the Offi cial Secrets
19 Alan Mathison Turing was the creator of the concept of the "universal
machine," later called a "Turing machine" in his honor. See Lee (1995a), pp
670-678, and sidebar, p 33.
IBM and Mechanical Calculators
The last thing I want to mention is the mechanical calculators, which were used
for scientifi c as well as commercial calculations. In truth, their use at
Aberdeen preceded the other devices I just mentioned.
During this period, the International Business Machines Corporation (IBM)
continued to design more powerful accounting machines, including a
Card-Programmed Electronic Calculator (CPC). Of more importance to us, Herb
Grosch, in his memoirs, mentions that the Ballistic Research Laboratories,
under Major Leslie Simon, let a contract to IBM "to develop two high-speed
relay calculators with plug board sequencing."21 Constructed during 1944, these
machines were called the IBM Pluggable Sequence Relay Calculators, but they
were better known as the "Aberdeen machines" and were used for preparing firing tables.22 Five additional copies of this machine
were built, three for the Naval Proving Ground at Dahlgren, and two for the
Watson Scientifi c Computing Laboratory. Ultimately IBM built its largest
calculator, the IBM Selective Sequence Electronic Calculator (SSEC), which
contained about 13,000 vacuum tubes and was dedicated in 1948.23
22 "The two IBM Relay Calculators were used for a short time but were not
successful." -Kempf (1961), p 17. (Copies of this monograph were distributed to
23 For more on the SSEC, see Williams (1997), pp 255-258.
The topic of mechanical calculators leads us to the fi rst use of the term
computer, which was originally used to identify individuals who did
calculations using mechanical and electrical calculators. These people were
typically women with a strong educational background in mathematics. There's
an excellent paper in your packet by Barkley Fritz from the Annals of the
History of Computing on the women computers.24 In the second panel this morning
we will all have the opportunity to hear some of these women "computers" speak
about their experiences here at Aberdeen.
The Moore School of the University of Pennsylvania was founded in 1923. By the
1930s, it had formed an arrangement with the US Army's Ballistic Research
Laboratory here at Aberdeen. As I mentioned earlier, a major result of this
collaboration was the construction of two differential analyzers, one at the
Moore School, and one here at Aberdeen.
By the 1940s, faculty at the Moore School were involved in radar and other
electronics research. John Mauchly, whose interest in high-speed computation I
mentioned earlier, was a professor of physics at Ursinus College outside
Philadelphia.25 Because of the war, he enrolled in a wartime electronics course
at the Moore School; John Presper Eckert, Jr., was a graduate student
overseeing the laboratory for that course.26 Mauchly and Eckert spent many
hours discussing electronics, especially Mauchly's fascination with weather
prediction, an effort retarded by the lack of high-speed computational
capacity. When the Moore School needed to replace faculty who were drafted
into military service, Mauchly agreed to join the Moore School Faculty.
In August 1942, Mauchly distilled his ideas into a
short paper, "The use of high-speed vacuum tube
devices for calculating," in which he compared the advantages of electronic
techniques to those of mechanical technology.27 Mauchly estimated that
calculations for ballistic trajectories would be in the 100-second range,
compared to the 15 to 30 minutes required using mechanical technology.
Mauchly's paper was not well received.
[Note: the graphical timeline presented in the original document shows
Mauchly's memo was dated August 1941, not 1942, which conforms with other
27 First printed in Randell (1973), pp 355-358, from original typescript.
By 1941, the production of firing tables was far behind. The officers of BRL
were searching for any opportunity to improve processing. Lieutenant Herman
Goldstine, an assistant professor of mathematics at the University of Michigan
before the war, was assigned to oversee the production of firing tables,
including the supervision of the women computers at the Moore School. Hearing
of Mauchly's ideas, he approached his former supervisor, Colonel Paul Gillon,
about pursuing the construction of an advanced machine. Colonel Gillon, who
now worked in the office of the Chief of Ordnance, recognized the potential
for success and convinced the Army to fund the project.
Accordingly, on June 5, 1943, the Army Ordnance Corps and the University of
Pennsylvania signed a contract for "research and development of an electronic
numerical integrator and computer and delivery of a report thereon."28 The
initial contract was in the amount of $61,700.
It should be noted that Colonel Gillon was responsible for the addition of the
words "and computer" to the name of the device in the contract. Gillon wanted
to forestall future problems if the machine was used for more general
problem-solving purposes later on.
The Electronic Numerical Integrator and Computer (ENIAC) was offi cially
dedicated on February 14, 1946. It had 40 units with 18,000 vacuum tubes,
1,500 relays, 70,000 resistors, 10,000 capacitors, and miles of wire. It was
eight feet high, three feet deep, and 100 feet long, weighed 30 tons and
consumed 130 kW of power.29 Although the Army had initially budgeted $150,000
for the project, the fi nal accounting showed a total expense of $486,804.22.30
So you can see that we got a real bargain for our tax dollars.
In March of 1946, after a dispute about patent rights, Eckert and Mauchly left the Moore School to found their
own fi rm, the Electronic Control Company.31
31 For their reflections, see Eckert (1980) and Mauchly (1980).
In the first panel this morning, we'll have an opportunity to hear from Herman
Goldstine, BRL's liaison to the Moore School; Harry Huskey, one of the ENIAC
engineers; and Harry Reed, an early member of the BRL team at Aberdeen. Your
packet also contains some reprints from the Annals
of the History of Computing devoted to the ENIAC and the ENIAC applications.32
32 Volume 18, Number 1, of the IEEE Annals of the History of Computing was a
special issue "Documenting ENIAC's 50th Anniversary." Two papers from this
issue were reprinted for the attendees: Winegrad (1996) and Goldstine and
Goldstine (1946). Reprints of Fritz's ENIAC history papers (Fritz, 1996, 1994)
were also included.
After the planning and construction were well under way, the project members
started discussing ways to improve the machine. Since it used cables to move
pulses from place to place, setting up the machine was time consuming and
subject to error. One idea was to create a memory that would store
instructions and data. Indeed, Pres Eckert envisioned using a modifi cation of
the mercury delay line that he had developed for radar use.
It was during this period that John von Neumann, a member of the BRL Scientifi
c Advisory Committee, met Herman Goldstine at the Aberdeen train station,
learned of the project, and joined the ENIAC team as an advisor. In June 1945,
von Neumann prepared a document called "First draft of a
report on the EDVAC," which, as its title suggests, was preliminary and
informal.33 Goldstine distributed this report to
members of the Moore School staff and interested outside scientists, and thus
it was the first widely distributed report on electronic digital computers.34
33 Copies of this report are contained in Stern (1981) and Randell (1973).
34 Herman Goldstine points out that this distribution of the report "placed its
contents in the public domain, and hence anything disclosed therein became
unpatentable." See Goldstine's Jayne Lecture (Goldstine, 1992) (provided as a
handout to attendees).
EDVAC stood for Electronic Discrete Variable Automatic Computer. It was
constructed by the Moore School for BRL and delivered in August of 1949. This
stored-program computer was put into "practical use" at Aberdeen in April
1952.35,36 The first panel this afternoon will provide insight into the EDVAC and its immediate successors, the Ordnance Variable
Automatic Computer, or ORDVAC; the BRL Electronic Scientific Computer, known as
BRLESC I; and its successor BRLESC II.
35 Ballisticians (vol. I), p 73.
36 For interesting first-person discussions of EDVAC and its capabilities, see
Mauchly (1973b) and Burks (1980)
Moore School Lectures
Although the ENIAC served BRL well until it was turned off in October of 1955,
perhaps the most important contribution of all these efforts was the Moore
School lectures during the summer of 1946. Actually the fi rst course on the
"Theory and Techniques for Design of Electronic Digital Computers," the series
included lectures by most of the members of the Moore School team, as well as
prominent members from the small but growing computer community. These
lectures spread the "good news" about computing to many of the scientifi c
organizations across the country and the world.
Evolving out of these projects, or inspired by the Moore School lectures, were
many other projects to design and build electronic computers in the forties and
early fifties. Three main lines of development can be traced:
37 Weik (1955) includes a three-branched "computer tree" of domestic computing
history/ evolution (included in the attendees' packets); see pp 52-53, this
(1) The EDVAC was the model for a National Bureau of Standards project that
became known as the Standards Eastern Automatic Computer (SEAC),38 as well as
for the BINAC and the UNIVAC machines.
(2) The Institute for Advanced Study at Princeton established a computer
project under the direction of John von Neumann. The team included Herman
Goldstine and Arthur Burks, both associated with the ENIAC project. The IAS
computer39 inspired a second NBS effort, the National Bureau of Standards
Western Automatic Computer (SWAC)40 (which was headed up by Harry Huskey, who
is one of our speakers this morning), as well as the ORDVAC,41 MANIAC,42 and
(3) And finally, Maurice Wilkes, who was one of a number of people inspired by
the Moore School lectures, returned to Cambridge to design and build the EDSAC,
the Electronic Delay Storage Automatic Calculator.43 Other projects were
started by Manchester University and the National Physical Laboratory in
Britain.44 Harry Huskey, one of our speakers, spent 1947 in the UK, working at
the National Physical Laboratory, and visited the other British computing
From these humble beginnings, an industry developed that has altered modern
society more than any other. Each semester I teach an undergraduate course,
"Introduction to Computing." Each semester I have to discuss new capabilities.
A few years ago, it was personal computers, then it was the Internet, and now
it's the World Wide Web. Such courses are only a barometer of the computer's
impact on society at large.
The Army too has had to keep up with technology, and the first panel tomorrow
morning will tell us about how the Army Research Laboratory has attempted to
stay ahead of the computing fi eld with leading edge research and applications
such as supercomputing and networking. And fi nally, after the awards
ceremonies tomorrow, a panel will explore the future of computing from ARL's
perspective. Certainly all of us will get insights into the future of
computing from that panel.
One of the metrics I use for measuring technological change is the extent to
which new technology is discussed in the popular media, such as newspapers and
news magazines. Fifteen years ago, computers were rarely mentioned in such
media, and certainly they were never advertised in them. Now there's rarely an
issue of the Washington Post, Time, or Newsweek without major stories on
computing topics and multiple advertisements for personal computers. Where
most people had not heard of the Internet just three years ago, a growing
number of people are using electronic mail on a daily basis, and public and
private organizations are planning for and exploring electronic commerce.
Finally, in the last year the world has jumped on the World Wide Web, so that
you rarely see an advertisement that does not have a Uniform Resource Locator
(URL) at the bottom. Indeed, the history of computing community, of which I am
a member, has a number of Web sites, including one by Mike Muuss here at ARL.
If you have some time, you can kill a couple of hours out on Mike's web site45
examining all the materials he has collected.
How It All Began
So, tomorrow morning when you pick up your newspaper and see an article on the
Internet or an advertisement for a personal computer, remember how it all
began: a small research project funded by the Ordnance Department, to meet the
needs of the Ballistic Research Laboratory, at the Moore School of the
University of Pennsylvania, and a dedicated cadre of engineers and (female)
This morning, we have the incredible privilege of listening to some of the men
and women who were there "at the beginning." Thank you.
Thank you very much, Tim. We're going to take a short break. I wanted
to mention to you that a few minutes ago, one of the returning pioneers
mentioned that a decade or two ago, somebody said that "It would be a cold
day at Aberdeen before they recognized Army computing." [laughter] Well, as
usual, our pioneers were prescient in their judgments.46 I do want to tell you
Timeline of Selected Early Computing Activities
[Note: this is a graphical representation that is available from many other
2. ENIAC: Development and Early Days
Harry Reed (chair):
I've been asked to make one announcement. As you all may notice, we've got
television coverage, at least video coverage, of all the events here, and we
are going out over a thing called the MBONE, which is the Internet slow TV
system. People around the world can log in on their PCs and watch what's going
on here today. This is one of the little marvels that Tim Bergin referred to
I'm Harry Reed. I came to BRL in 1950, and started working on the ENIAC and
the preparation of fi ring tables. For many years, I'd heard about some of the
people that are here today, some of whom I fi nally had a chance to meet at the
1996 ACM Conference in Philadelphia. It's my privilege today to have two real
pioneers on this panel: Herman Goldstine1 and Harry Huskey.2 Herman was, as
mentioned earlier, the project offi cer on the ENIAC project. Harry was at the
University of Pennsylvania and worked on the ENIAC.
We are going to have a rather informal format; we're going to have a chat. I'm
not sure how it's going to go, and I'm not sure what topics we're going to
cover. But we're going to try to refl ect on some of what happened in those
days, get a little bit of the fl avor of the times, and perhaps hear a few
anecdotes. We will try to give you some idea of some of the things that led up
to the development of the ENIAC and some feeling for the early days of the
Guys, I want to thank you both for being here. Herman, let's start with you.
It was one of those fortuitous things, I guess. In 1936, you were at the
University of Chicago, right?
And that was rather significant, wasn't it?
Yes, it was! I was fortunate enough to be able to teach a course on exterior
ballistics. I was very fortunate in being the assistant to a man named Gilbert
Bliss, who had been here during the First World War; he thought it would be a
good idea if we taught a course in exterior ballistics.3 Unfortunately, Bliss
developed heart trouble, and so I taught the course. This got me into a shape
where I knew a lot about the subject. We finished the book!
And also, I was fortunate in having taken a master's degree in mathematical
astronomy, and so I knew a fair amount about calculations. It was the natural
thing for the Army to transfer me immediately to Lowry Field, in Denver, and
then to Sacramento, California, as an Adjutant of a squadron. Fortunately
again, Bliss got hold of Oswald Veblen,4 who was the
Chief Scientist in both the First and the Second Wars here at BRL, and they got
me out of the Army Air Corps, and back here to Aberdeen.
4 Goldstine (1993) discusses this period in great detail (ch 9), "Ballistics
and the rise of the great mathematicians"; Oswald Veblen is introduced on page
77. According to Ballisticians, vol. I, p 1: "The Range Firing Section, under
Major Oswald Veblen, prepared all firing tables (at that time called Range
Tables), made mathematical analyses of ballistics problems, and conducted
experiments to obtain information for increasing the range and accuracy of the
That was a little unusual, actually, for someone to teach a course in exterior
ballistics in a university, in the pre-war days, wasn't it?
Yes, it was.
I guess Bliss was really kind of enthusiastic about exterior ballistics.
He was. There was a group of men that came here including Oswald Veblen and
Gilbert Bliss. Norbert Wiener was an enlisted man
here in that period.5 And there was a great interest by mathematicians in the
work at BRL. Mr. Bliss was a young patriot, who decided that he really ought
to equip young people to help in the war effort. That was my good fortune.
I'll tell you, when I got my travel orders at Sacramento to come to Aberdeen
Proving Ground, they were signed by the Adjutant General of the United States
Army, but I also had orders to be the Adjutant of a squadron in Sacramento. So
I called the Commanding Officer in Sacramento for advice, and he said, "Well,
who signed the orders to come to Aberdeen?"
I said, "It's the Adjutant General of the Army."
He said, "Well, he takes precedence over anybody here." [laughter]
So I said, "What should I do?"
He said, "Get in your automobile and start driving and don't ask any
So I did.
Along the way, I picked the route and got here, where I met Colonel [Paul]
Gillon,6 who was to be my Commanding Officer. Under him, talking about
mathematicians from the First War, there was a Major Albert
A. Bennett, who had been an officer here during the First War also.7
So there was a group that knew a lot about computing here [at APG].
5 Norbert Wiener, one of the most eminent mathematicians of this century,
served as a "computer" at APG in 1918 and 1919. His most well-known efforts
are Cybernetics (1948) and The Human Use of Human Beings (1950). He served on
the MIT faculty from 1919 to 1960, and died in 1964.
7 "Another man, active in ballistics both in World War I and II and a longtime
professor of mathematics at Brown University, Albert A. Bennett, described
what went on as `wrenching the equations into a form that could be easily
solved by very simple means.'" Goldstine (1993), p 74.
So then they sent you up to Philadelphia, right?
Yes, I was very fortunate in that Colonel Gillon and I somehow just clicked
together. There was an empathy that developed, which persisted through the
remainder of his lifetime. Even after the war ended, we would see the Gillons-
not frequently, but once a year or so. At that time, he took me to
Philadelphia, and it was clear that something needed to be done. Fortunately,
he said, "You do it."
You had something like a hundred and some people pushing hand calculators.
Is that right?
Yes. One of the important things about Philadelphia was that there were a
number of colleges and universities in the area from which we could hire
The word "people" in those days meant women, because the men were all taken
into the draft. I had, I believe, maybe one man, John
Holberton, who was my civilian aide.8 He did the civilian things for
the unit. For the rest, we hired women. We were very fortunate. We managed
to get very competent young women from all the schools in the area, and even
older women. We had some remarkable people. As you know, it all worked out
8 John Holberton attended the Commemoration; he is married to Frances Elizabeth
(Betty) Snyder Holberton, one of the original six programmers. Betty
participated in the Women Pioneers session (pp 38-49, this volume).
The next session, in fact, is going to focus on some of those first women
programmers. [to Harry Huskey] Let's see, we have Herman at the University of
Pennsylvania, Harry, so what was going on with a guy by the name of Mauchly,
for instance, who was starting to have some ideas about putting computers
together. Can you tell us something about the environment at Penn up to that
They were well under way with the ENIAC project by the time I arrived on the
scene. I came to Penn as an instructor in mathematics, and as an instructor my
salary wasn't all that high, so I looked for extra activities. I heard that
there were projects going on at the Moore School, where one might get a job.
So I went over and applied. The ENIAC, of course, was a classifi ed project,
so I didn't even know what I was applying for, or what it was about.
After clearance, I was admitted through the locked gate, and here was this
machine which was partly constructed. There was a lot of work going on. This
was after the initial test of the accumulators. This is when I came on the
You asked about Mauchly. [see sidebar] Certainly,
his interest in this sort of thing was because of his research in
meteorological computations. I think he looked upon the interest at Aberdeen
Proving Ground as something that complemented his interest in meteorology. Of
course, Eckert was the engineering expert. [see
sidebar] He had worked on radar and so on. So between them, they worked up a
proposal and submitted it to the Army, and in due course contracts were written
and the machine was built.
Mauchly and Eckert
John William Mauchly received his degree in physics in 1932 from Johns Hopkins
University. After a short tenure at Ursinus College, he joined the faculty of
the Moore School of Electrical Engineering at the University of Pennsylvania.
He is regarded as co-inventor of the electronic digital computer (with J.
Presper Eckert). He married Kathleen McNulty, who was one of the six original
ENIAC programmers, in 1948. Kay (now Kathleen McNulty Mauchly Antonelli) was a
speaker in the Women Pioneers session. John Mauchly died on January 8, 1980.
(For a retrospective on Mauchly's life, see Stern, 1980.) John Presper Eckert
was born in 1919 and took his BS and MS degrees from the Moore School,
University of Pennsylvania, in 1941 and 1943. "Pres," as he was known, worked
with John Mauchly on ideas for increasing the speed of computations. On his
24th birthday (April 9, 1943), the Moore School received the authority to begin
the ENIAC project, for which he served as chief engineer. Eckert died on June
3, 1995. (For retrospectives on Eckert's life, see Lee, 1995a, pp 271-275;
One of the interesting things is that these things all kind of came together
then, because for hundreds of years, people such as Charles Babbage were
talking about building some sort of a calculating device, but it never quite
culminated in anything until the wartime technology came along.
I think there are two requirements: first, you have to have the technology to
build appropriate components; and second, you have to have a need so that
people will support the effort, financially and otherwise. Both of these
things were certainly present at that time.
Do you mean radar development and things like that?
Yes, like digital circuitry, for example.
[to Herman Goldstine] Ok, so anyway, Herman, you're at Penn and your young
ladies are overloaded with trajectory calculations and so forth. And you bump
into this guy, Mauchly, right?
And you start talking .
It was a very interesting little group at the Moore School. We haven't talked
about John Grist Brainerd yet.9 He was the man who
was the liaison for the Moore School with the Army. He was a very important
person on the ENIAC project, and when we got to seriously talking, he became
the head of the project. Another thing I'd like to mention was the fact that
when we started, my first efforts had nothing to do with building a computer.
I was occupied with hiring young women and supervising the construction of
Brainerd formed a teaching team with my [first] wife Adele
[see sidebar], Mary Mauchly (who was John Mauchly's first wife), and a
splendid woman named Millie Kramer, whose husband was a great Assyriologist at
the University of Pennsylvania. She is still alive, and a remarkable person.
We started right away with the training of our new people. At some point, when
we ran short of trained women, we brought in a company of WACs [Women's Army
Corps]. Our team trained them. I think there was a close connection between
our bringing women into the forefront of the working community as normal
employees, not just as ancillary or support personnel. I think that was an
important accomplishment in its own right.
We, moreover, had a differential analyzer-a copy of the Bush machine-and a
Moore School professor named Cornelius Wygandt, who was in charge of the
analyzer for us. Pres Eckert and John Mauchly were around that machine a lot.
In fact, Pres made a great invention. He replaced the mechanical torque
amplifiers with some Polaroid sheets, and made an electronic device that
increased the reliability of those machines a great deal. It was things like
that which helped to build my feeling of confidence that these young men, Pres
Eckert and John Mauchly, were really people who understood computing.
9 John Grist Brainerd was a lifelong faculty member and served as dean of the
Moore School of Electrical Engineering, University of Pennsylvania, and was
co-principal investigator for the ENIAC project.
Adele Katz Goldstine and stored-program machines
Among other contributions, Adele Goldstine pioneered an improved programming
system for the ENIAC. "During 1947 von Neumann realized that the lack of a
centralized control organ for the ENIAC was not an incurable defi ciency. He
suggested that the whole machine could be programmed into a somewhat primitive
stored-program machine. He turned the task over to Adele Goldstine, who worked
out such a system and passed it along to Richard Clippinger, who was then head
of the computing Laboratory . and is also a mathematician of note . The system
. provided the ENIAC programmer with a 51-order vocabulary. This was modifi ed
to 60 orders by Clippinger and then later to 92 orders." Goldstine (1993), pp
Adele Goldstine wrote Report on the ENIAC (Electronic Numerical Integrator and
Computer), Technical Report 1 (2 volumes), Philadelphia, PA, 1 June 1946.
See also Clippinger (1948, 1949). Clippinger (1948) was provided to attendees.
And you started talking?
We started to talk and .
One thing led to another?
Yes, it did. If I may, I'd like to read you something my first e wrote in her diary.10 It was written in about 1962, and
was her summary, for our two children, of how we spent the war years. It is
simply told, so the children would have little diffi culty understanding the
complex story of our war years. The reader must therefore recognize that these
few pages that I have excerpted were written for young children and yet are
poignantly clear. It certainly is the most unbiased of the ENIAC records. It
has not seen the light of day for many years, and I only now open up this part
of my life for the benefi t of historians. It's a very simple and elegant
description of the relationships among the ENIAC designers as told by a woman
who was a key worker, yet who had no need or desire to bias facts.
This document was written after Adele knew she was stricken with a mortal
disease and would be unable to tell her story of the ENIAC design and
construction to her children when they grew up.
...by now Daddy was working at the Moore School at the University of
Pennsylvania to set up a computing station for Aberdeen and classes to train
more computers under the auspices of the Engineering Sciences Management War
Training Program. I got a teaching job in this program that fi t me like a
glove, since it involved teaching serious grown-up students who wanted to
During this period, Daddy had begun to speak to two engineers at the Moore
School, Pres Eckert and John Mauchly, about building an electronic computer
that could take over some of the calculating work of the Aberdeen lab. Daddy
was already using the differential analyzer at the Moore School, which was a
mechanical machine that could solve differential equations through representing
quantities by lengths of rods and turns of cams . the idea these three men had
was to build a digital machine that actually carried out arithmetic operations
by electronic means. For several hundred years there have been gropings toward
a digital machine, but these were mechanical in operation. At this point in
the war, there was a tremendous growth in electronic technology that made it
seem feasible to build a digital computer that could work with prodigious
...the engineers had begun work on the computer. Daddy had persuaded Army
Ordnance to sponsor this work, and he was transferred back to Philadelphia to
supervise the project for the Army, as well as to run the computation office at
the University. Then I got a marvelous job at the Moore School which consisted
of learning how the computer would work and writing a manual to instruct
operators in the use of the machine. At fi rst I thought I would never be able
to understand the workings of the machine, since this involved a knowledge of
electronics I didn't have at all. But gradually as I lived with the job, and
Daddy and the engineers helped to explain matters to me, I got the subject
under control. Then I began to understand the machine and had such masses of
facts in my head, I couldn't bring myself to start writing. But this, too, I
fi nally surmounted. I am very proud to have fi nished the job. It has even
been printed by the Government Printing Offi ce and listed by the Library of
...Daddy and I worked together days and often returned to work in the evening
or had dinner with some of the engineers and then went back to the Moore School
or one of our houses for more talk. Pres Eckert, the chief engineer, was a
very clever young man and stimulating company. Grist Brainerd, who supervised
the ENIAC project for the university and later became Dean of the Moore School,
was another friend of ours.
As the ENIAC neared completion, Daddy interested John von Neumann in the idea
of large-scale computers. When he told Pres Eckert that the great von Neumann
was coming, Pres, who was not a mathematician, was not strongly impressed.
Skeptically he told Daddy that he would see if V.N. asked a particular question
that he considered crucial; then he would be impressed. Johnny passed the
[laughter and applause]
One of the other people who worked on this set of manuals was Harry Huskey.11
10 Ellen Goldstine (Herman's second wife) sent a note to Paul Deitz in October
1996, saying, "While the ENIAC celebration was being planned in Philadelphia, I
had copied the attached from Adele Goldstine's notebook. Thought you and Kay
would like to read it. It's all so simply put."
Right! [to Harry Huskey] Harry, some of the notes you gave me suggest how the
ENIAC compared to what one thought of as the general-purpose computer, the
Babbage machine, and stored-program machines. Would you put the ENIAC in some
context with respect to this?
As it was initially designed at the University of Pennsylvania, the programming
was all by jumper connections from one unit to another. So changing a problem
was a matter of removing these wires from the prior configuration and putting
them back in a new configuration and testing to see if it worked. This might
take days if it was a complicated program. So you can think about it as being
a wired program machine. It was electrical in the sense that you could change
the wiring, but once you set up the wiring, then it was fi xed. For example,
if you programmed a ballistic problem on it, it would take two or three days to
set up, but then you could run trajectories in seconds, varying initial
conditions and parameters.
Of course, it was really just a collection of
accumulators, some arranged to do multiplication and some arranged to
do division and so on.12 You could hook the units up in any combination you
liked. So, it was a general-purpose computer. This was in contrast to the
British (Colossus) code breaking activities that were mentioned this morning.13
So I think it was the fi rst electronic general-purpose computer. Programming
was wired programming, and it's only in the machines that come later, and
particularly the EDVAC machine, that the stored-program concept was introduced.
And that's a very important milestone in computer development.
12 The ENIAC consisted of 40 separate units: initiating unit, cycling unit,
master programmer (two panels), three function tables (two panels each), 20
accumulators, divider and square rooter, three multipliers, constant
transmitter (three panels), and a printer (three panels). In addition, the
ENIAC used three portable function tables, an IBM card reader, and an IBM
summary punch (Stern, 1981). 13 Keynote address, pp 16, 22 (this volume).
The ENIAC had an intermediate version of programming
[see sidebar] where you set the programs on the function table switches and the
wiring was left alone. You didn't move the jumper cables anymore, but they
were designed to accept two-digit codes from the function tables and then run
the programs. So you almost had a stored-program machine, but you couldn't
modify the program when you were running. The switches were set and that was
"Intermediate programming"-converter code
The converter code is similar to the Load and Store machine code of most
present-day central processing units. The ENIAC function tables consisted of
about 3600 rotary switches. In the converter-code version of the ENIAC, these
switches could be used to store machine instructions as well as numerical
constants (often in tabular form). Pairs of switches were used to represent
the approximately 100 orders, and these pairs were sampled in order of their
position on the function tables, unless an instruction required the
computations to move to another section of the tables. A converter was added
to the ENIAC, and the computer was wired permanently so that these numbers
could be converted into the approximate string of pulses. One set of orders
moved numbers from accumulator 15 (the central register) to the other
accumulators; another set moved numbers from the other accumulators to
accumulator 15 (with addition). A variety of other orders carried out various
arithmetic operations such as shifting and printing. Whereas the original
ENIAC required a fortnight to be reprogrammed by rewiring, the new system
required only a couple of hours for a programmer to set up a new program. (See
Clippinger, 1949.) -Harry Reed
What was it like working on the ENIAC project?
I was teaching full-time in the mathematics department, so this was extra work
for me, and extra pay too! [laughter] It was very interesting learning about
the concept and capability of this machine. Actually, before that, when I was
in graduate school, I had thought about building a relay machine to do
calculating. I decided there wasn't any use for this thing, that it wasn't
practical, and so I never spent much effort on it. Here was this electronic
version of a full-scale calculating machine, occupying a large room. It was
exciting! In fact, one of the things that we worried a bit about was the
effect of the rounding errors. Professor Rademacher
in the Math Department had been asked to do a report.14 So the first
opportunity that I had, I asked if I could run a problem on the ENIAC. So I
set up the integration of a very simple differential equation system to see
what the rounding error would be in practice. It was an exciting time!
14 According to Goldstine (1993), p 232: "Profs. Hans Rademacher and Harry
Huskey did computations of tables of sines and cosines to study the way
round-off errors develop in numerical calculations (15-18 April 1946)."
[to Herman Goldstine] Anyway, Herman, you and Mauchly and Brainerd, I
guess, put together a proposal which you showed to the Army; is that right?
I think it was Eckert and Mauchly, with help from Brainerd, who put the
proposal together. What I did was to get the idea across to them that if they
could make a proposal, I felt confi dent I could sell it to Paul Gillon. And
with Paul's help, Oswald Veblen was going to be a cinch. Veblen believed in
people rather than in projects, and he had a lot of confi dence that we would
Do you have a little story about that one?
Yes. It's a nice story. I think it was on Pres Eckert's
birthday15 that we drove down from Philadelphia with a proposal, with
Pres and John Mauchly in the back seat, writing away to get this thing put
together. We got there, and I felt confi dent that it was to be a fait
accompli, no matter what the University personnel were going to do. One had to
go through these formalities, though, and there were Leslie Simon16 and Oswald
Veblen representing BRL and there was Brainerd with his two young men. The
presentations went on and finally Veblen, who was sitting with his feet up on
the table leaning back, bounced forward, stood up, and said, "Simon, give
Goldstine the money." [laughter]
Just like that, huh?
Yes, I think it was kind of a letdown to Brainerd, who expected that it would
be a big hard proposition. It took just a matter of a few weeks for Paul
Gillon somehow to get the Philadelphia Ordnance District involved, and they
wrote the contract. Away it went.
15 It was Eckert's 24th birthday, April 9, 1943.
16 At the time of this incident, Colonel Leslie Simon was Director of the
Ballistic Research Laboratory (notes Harry Reed).
Veblen was quite a fellow.17 He was from Princeton.
During the war, he was chief scientist at BRL, I think, and helped pull
together the scientific staff.18 If you go look at
some of the displays, you'll find Nobel Laureates
and so forth.19 A lot of the credit for gathering the group, I think, went to
17 "The other leader in the story of ballistics in the United States in World
War I was Oswald Veblen (1880-1960) . After many years as a professor at
Princeton, Veblen and Albert Einstein were appointed the first professors at
the newly-founded Institute for Advanced Study, where he remained until his
retirement." (Goldstine, 1993, p 77.)
18 "Dr. Oswald Veblen . came back to the Laboratory as a consultant in April
1942, and proved to be one of the most successful recruiters of scientific
talent." (Ballisticians (vol. I, pp 29-30). Ballisticians then provides
numerous examples of scientists who were recruited and their institutional
19 There were a number of photographic displays at the commemoration. In
addition, a handout on "Applying science and technology to military problems"
was included in the registration packet. There was also a one-page fl ier,
with a photo, identifying the members of BRL's Scientific Advisory Committee
Well, Veblen was actually a part of Army Ordnance when it was located at Sandy
Hook Proving Ground before it was moved here. I think he was a Major, and I
know that he was busy leaning out of an airplane dropping bombs from his hands
trying to see how this whole bombing thing would go during the First World War.
He was a tremendous organizer. He was one of the men who organized American
mathematics. There were three key leaders: Veblen at Princeton, George
Birkhoff at Harvard, and Gilbert Bliss at Chicago; they ran American
mathematics when it was a thing in its infancy. It's sort of remarkable; one
thinks of mathematics as having been in existence in the United States for some
time and that we were a totally civilized country. But in fact, just before my
time, everybody who was going to be a "mathematician" went to Europe and got
his Ph.D. abroad. So it was just in the 1920s and 1930s that American
mathematics began to mature.
Curiously enough, with respect to the ENIAC project, I don't think there was
anybody on that project who had ever heard of Babbage or any of the machines
that were talked of in earlier times. Now Harry Huskey may have heard of
Babbage, I don't know.
[to Harry Huskey] Did you know about Babbage before ENIAC?
Not until afterwards.
Babbage never actually made this thing you called an analytical machine, did
he? It was just a concept?
He worked on parts of a "difference engine" and got so excited about the
"analytical engine" that he stopped working on the difference engine.
But was there technology to do really what he wanted to do?
It's kind of an interesting question. He proposed to build something involving
a lot of gears operating in sequence. Now the clocks that were built at that
time were very elegant mechanical devices, but did not have chains of gears
like those envisioned by Babbage. So it would be interesting if somebody had
really tried to make the calculator. They'd probably run into diffi culties.
Anyway, Babbage failed because of disagreements with his chief mechanical
assistant, so it is hard to say. [see sidebar]
He had another difficulty: he had to get money from the government. The Prime
Minister said, "It's going to be a long, cold day before I get up before the
House of Commons and ask for thousands of pounds to build a `wooden man' just
to evaluate the formula N2 + N + 41." [laughter]
But Babbage, to give him credit, was one of the important founders of
operations research. In part, he was responsible for the introduction of the
penny post. Before his time, people thought that the expense of postage had to
do with the distance that the letter traveled. He did an analysis pointing out
that almost all the expense of postage was incurred at the two termini. That
resulted in a common post.
Charles Babbage was born in 1791 in Devonshire, England. A polymath, he is
known as the "Father of Computing" for his work on his difference and
analytical engines, neither of which he ever completed. (See Williams, 1997, ch
4, and Lee, 1995a, pp 51-64.)
According to the story, Babbage, who was a diffi cult person, argued with
Samuel Clement, his chief mechanic. Because of their differences, Clement kept
the machine tools that he had constructed to Babbage's designs (as he was
allowed to do, under British law); he did, however, return Babbage's drawings.
In 1840, after reading about Babbage's ideas in the Edinburgh Review, a young
Swede, Georg Scheutz, did build a difference engine. One of Scheutz's
difference engines is in the Smithsonian Institution's Information Age exhibit
in Washington, D.C. (See Lindgren, 1990.)
The Science Museum in South Kensington (London) sponsored a project to build a
difference engine according to Babbage's plans; this work was completed in 1992
to commemorate the 200th anniversary of Babbage's birth. (See Swade, 1991.)
[to Harry Huskey] I'm not quite sure of the answers, but you had the chance to
work with Alan Turing, legendary in the computing business. [see sidebar opposite] He too built a conceptual computing
machine. What sort of a fellow was Alan Turing?
A genius, I guess. John R. Womersley, who was superintendent of the Math
Division and next above Turing, talked about how difficult it was to work with
him, primarily because Turing was a peculiar mixture of impatience. If he felt
you did something stupid, he had nothing to do with you. Typically he would
never talk to newspaper people. [laughter]
On the other hand, if you had a serious problem that you were struggling with,
then he would be very helpful. I spent a year working with him, and it was
certainly a very pleasant experience.
The British government's Code and Cypher School was located
just outside London in Bletchley Park. It was here that an
electromechanical device, known as the Heath Robinson, was
constructed to assist in deciphering code; later a number of
electronic special-purpose computers, the Colossus machines, were
built. Since this work is still covered under the British Offi cial
Secrets Act, many of the details-such as Alan Turing's contributions-
are unknown. (See Williams, 1997, pp 291-294.)
Alan Mathison Turing (b. London, June 23, 1912) is recognized as the creator
of the concept of the "universal machine," described in his 1937 paper.
Turing's "theoretical computer" is known as a "Turing machine."
Alan Turing left his code-breaking efforts at Bletchley Park and joined the
staff at the National Physical Laboratory (analogous to the U.S. National
Bureau of Standards) in the fall of 1945. He immediately began a project to
design an Automatic Computing Engine (ACE).
Another project to build a computer, later called the Manchester Mark I, was
started at Manchester University in July 1946, and was in limited operation in
June 1948. Turing joined the Manchester computer project in September 1948.
After many design changes, and assistance from Harry Huskey, construction
started on a modifi ed version, known as Pilot ACE, in early 1949; this machine
became operational on May 10, 1950. (See Williams, 1997, pp 321-344.)
Turing died on June 7, 1954, in Manchester. (See Lee, 1995a, pp 670-678. )
He was a very stubborn man. Von Neumann and I had the idea that Gauss's method
of elimination-which is a way of solving a system of linear equations- was
probably the correct way to use on a computer. We had a mathematical proof,
and Turing came over for a few weeks and worked with us. He had a somewhat
different idea, and we never were able to get him to agree with what we said.
I should mention that as a graduate student, Turing got his Ph.D. at Princeton
under Princeton logician Alonzo Church. Turing's thesis was essentially on a
paper computer.20 Johnny von Neumann's offi ce was just a few doors away from
Turing's, and Johnny followed everything that Turing did. There was a meeting
of minds there from 1937 on. Von Neumann was so impressed with Turing that he
wanted him to stay as his assistant. But Turing wanted to go back because he
had a call from the Foreign Office to work at Bletchley.
You mentioned John von Neumann. There was a rather important historic
meeting at the Aberdeen railroad station, wasn't there?
Yes, there was. Actually, I knew von Neumann but he didn't know me before that
date. There had been a conference of the American Mathematical Society at the
University of Michigan and I was the rapporteur.21 Among the lecturers were
Norbert Wiener, Johnny von Neumann, and a bunch of other very important people.
I was a young man who had just recently got his Ph.D. So I had to take notes
and write up the material for one of the mathematics journals. I was awestruck
at von Neumann. He was just brilliant.
The other person I was awestruck by was Wiener because he gave such a lecture
that I couldn't follow it. [laughter] I thought I would never get that summary
for the journal done. At any rate, that's neither here nor there, but that's
how I knew von Neumann and he didn't know me. Because all I was doing was
sitting in the back of the room scribbling as fast as I could to try to take
One day during the war, after attending a meeting here at APG, I went to the
railroad station to go back to the Moore School. I saw Professor von Neumann
standing on the railroad platform, all alone. I was an egotist and decided
that I would go and talk to this famous man. Accordingly, I went over, and
found him to be a polite, European gentleman, who felt it appropriate to be
polite and to make conversation. But he was totally uninterested.
Then, gradually we began to talk about more things. Pretty soon he learned
that we were building a machine that would do 300 multiplications a second, and
suddenly he changed. [laughter]
He finally found the thing he had been looking for. I didn't know, of course,
that he was a consultant to Los Alamos and had been doing everything in the
world to get computing done. He had even worked at Harvard on the Mark I.22
None of these things had been satisfactory.
So when he heard that this machine was going forward, he asked me how he could
get involved, and I arranged it through Paul Gillon, so that he was immediately
cleared to come. Every time he was in the east, he would be at the Moore
School, at least once a week. That had great consequences, because by the
summer of 1944, the ENIAC consisted of two accumulators and some other
ancillary gear. The two accumulators were feeding each other and were
calculating a very simple sine function or something of that sort.23 It was
very clear that the success of the ENIAC was practically assured. The only
thing that remained was the manufacturing of additional units.
The question was, "How would the engineering group keep busy?" With everybody
being restless, we all said, "What can we do to make a better machine? Our
ENIAC has got 18,000 vacuum tubes in it, and it's going to be a monstrosity."
Its programming is practically zero, as Harry Huskey has pointed out. You
could change from one problem to another, but it was a matter of days of manual
labor. So we said, "We've got to get rid of that."
We began to talk about how we could build the computer of our dreams. We had
more or less weekly meetings. Johnny von Neumann joined into that little club.
It was a remarkable experience. The main characters were Johnny von Neumann,
Eckert, Mauchly, Art Burks, me, and maybe a few others. And the conversations
were heated and often irrelevant, back and forth on all sorts of topics: the
technologies that could be used and the ideas that should be embodied in the
machine. Out of all of this hammering and changing, and tugging back and
forth, people expressed their ideas. Sometimes, they were talked out of those
ideas, and later on it turned out that some of those ideas were accepted. It
was just a free exchange with much freedom for everybody to express themselves.
Johnny von Neumann went off to Los Alamos for the summer as he always did, and
he wrote me a series of letters. Those letters were essentially the thing
called the "First draft of a report on the EDVAC." I bashed those letters
together into a document without any footnote references to who was responsible
for what. It formed a blueprint for the engineers and people to use.
Then, it somehow took off in a peculiar way. People began to talk in spite of
classification; people began to talk about this new idea, and letters kept
pouring into the Ordnance Office asking for permission to have a copy of this
report. And pretty soon, we had distributed a few hundred. Everybody in
American and British scientifi c circles began to get copies and understand how
important this was. [see sidebar]
About this time, Paul Gillon and I decided that we ought to get some people
over from abroad to get the idea of the computer out to the world-as much as we
could. I picked a man whom I thought would be a very good person to bring
over; that was a man named Douglas Hartree, who was a mathematical physicist
then at Manchester University. He had built a differential analyzer out of a
Meccano set.24 He was interested in air flow over airplane wings. So Paul
Gillon got permission from the government to bring him over, and he programmed
a big problem. Kay Mauchly [then Kay McNulty] was assigned the task of being
his programmer. Hartree and McNulty, with perhaps some help from me, got that
thing going as an early example of how the computer should run. He received,
while he was in the States, an offer to be a professor at Cambridge University,
which he accepted. He took documents from us like the "First draft report" and
got Maurice Wilkes interested. Maurice attended the class, which Tim Bergin mentioned, in 1946.25 Out of that
grew that first machine called the EDSAC, which was
a successful copy, essentially, of the EDVAC.26
[to Harry Huskey] Harry, were you mixed up with the EDVAC at all, or were
you just an ENIAC person?
25 In the summer of 1946, a lecture series on the "Theory and Techniques for
Design of Electronic Computers" was held at the Moore School of Electrical
Engineering of the University of Pennsylvania (the "Moore School lectures").
The lectures, covering a wide variety of topics on computing, were delivered by
faculty members as well as other members of the small but growing computing
community. Maurice Wilkes attended some of the lectures.
26 There were three projects in the UK: Wilkes' efforts to build the
Electronic Delay Storage Automatic Calculator (EDSAC) at Cambridge University,
the Mark I at Manchester University, and the Pilot ACE at the National Physical
I worked on the EDVAC before I left in June of 1946. In fact, during that
spring, Eckert and Mauchly had quit because of disagreement on patent rights
and commercial development. Herman Goldstine and Art Burks went to Princeton.
So at that stage of the game, I was the senior person, although I was only part
time. In fact they offered me a job of running the EDVAC project but they
didn't clear it first with J. R. Kline, who was chairman of the Math
Department. He wasn't about to let the Moore School hire somebody from the
Mathematics Department. [laughter] He said no. So I was mad, and I resigned.
[laughter] I worked on it there awhile.
The "first draft report"
Herman Goldstine notes that "Johnny's idea and mine was to take his beautiful
analysis and synthesis of what we had discussed as a first step in reaching a
final text. I did not take the time to attribute . to each individual but
lumped it all into what I viewed as a first draft, and put his name on it. I
believed that only a dozen copies would appear; in fact, hundreds did." Harry
Reed points out that "there has been, and still is, controversy over the degree
to which von Neumann was responsible for the development of shared memory
machines. Dr. Goldstine credits von Neumann with the seminal role. Another
school of thought credits Eckert, who claims to have had the idea for internal
programming long before von Neumann." (See Eckert, 1988, interview.)
In the light of this controversy, it is interesting to note a passage from
Adele Goldstine's diary that follows the excerpt given on pp 28-29: "During the
last few months of work on the ENIAC, von Neumann, Daddy and Pres Eckert began
formulating ideas for a superior computer with central control, the EDVAC."
Kind of a cantankerous machine, wasn't it?
Not at the stage I worked on. I was just doing diagrams of how it might work.
Later on, there were problems as far as the memory in particular. In fact, it
was probably worth mentioning that Wilkes was successful in getting his machine
going before the EDVAC was completed, partly because he operated at half the
frequency that it was proposed to run the memory lines on the EDVAC. They were
supposed to run at a megacycle [1 MHz], a million pulses per second, and Wilkes
settled for 500,000, which makes it a lot easier.
Actually the ENIAC was always advertised as operating at 100 kHz. But we found
that if you turned it up much above 70 kHz, you were headed for trouble. In
fact, we only ran it at 100 kHz early in the week to see if everything was
working all right, because it would develop more errors if we kept operating it
that way. We'd sit and turn the clock down a little bit, and things worked a
lot better. Pretty finicky machine. People here, maybe a lot of people, in
the audience probably have never seen a vacuum tube.
There are a few vacuum tubes over there in the display case.
Tim Bergin brought some. You think about it, you used to have television sets
with a dozen vacuum tubes, and keeping them running for a couple of months was
a sensation. The fact that you had 18,000 of these cantankerous things in the
machine was a monument to engineering persistence, I think.
When there was a problem, it took maybe two days to discover which tube was not
functioning. It was actually good engineering and the fact that the tubes were
burned in for 100 hours before ever being used; that reduced the number of
failures. So, that's really one of the signifi cant points about the ENIAC,
that it showed that a very complex electronic device could be made to run
reliably enough to be worthwhile.
Eckert was really the driving force behind that accomplishment, wasn't he?
He noticed that if you left the filaments burning all the time, it made a
tremendous difference. The thing couldn't possibly have worked had it not been
for Pres Eckert's understanding of how to run equipment at below standard
ratings to increase reliability.
At the beginning, Paul Gillon and I decided we would try to get opinions from
the NDRC, which was mentioned earlier by Tim
Bergin.27 We went to the electrical engineering group at MIT, consisting
largely of proteges of Vannevar Bush. In the first place, they thought analog
computing was the way to go. They said digital computing is "for the birds,
everything is analog, and 18,000 vacuum tubes is preposterous. There ain't
such an animal ever been built and it will never work." So that was the learned
opinion of those people.
Be kind to MIT. [laughter]
The person who turned them around was Jay
Forrester.28 Under his leadership, they did a magnificent job from then
on. The next group was the Applied Mathematics Panel of the NDRC. They were
the people who said, "Well, we don't know whether 18,000 vacuum tubes is too
many or not; the thing we know is there ain't such a thing as a problem which
would take 300 multiplications per second." [laughter]
In fact-Tim Bergin mentioned George Stibitz in his lecture29 -it was Stibitz who was largely instrumental in
taking that point of view. He had the idea that if you took a couple of his
relay machines and put them together and let them run day and night, that would
do all the computing that Aberdeen needed. So somebody at BRL got cold feet
during the war while the ENIAC was being developed, and ordered two Stibitz
There was a nice young mathematician named Franz Alt, a Viennese, who was in
charge of running those two computers here. It turned out that they were more
unreliable than the ENIAC by a considerable amount. They didn't run day and
night, they were always in trouble, and in the end, it turned out that instead
of the vacuum tube being the unreliable element, it was the mechanical parts
which were the unreliable thing.
The IBM cards, for instance, which we used to feed in data would swell with
moisture because of excess humidity.
That was one of the great things, because we air-conditioned the computer
27 The National Defense Research Committee (NDRC) was created in 1940 to
provide advice on weapons research. Partly in response to criticism that its
focus was too narrow, President Roosevelt established the Office of Scientific
Research and Development (OSRD) by executive order, which then incorporated
the NDRC. The NDRC did a report on the potential of the proposed ENIAC
project, which was not favorable. (Stern, 1981, pp 16-23.)
28 Jay Forrester was the technical leader of Project Whirlwind at MIT. This
project started in 1943, when the U.S. Navy needed an airplane stability and
control analyzer. After initially considering using analog techniques, the
project focused on digital techniques once Forrester learned of the ENIAC
project. The Whirlwind project and the computer that resulted from it were
signifi cant developments in the history of computing. See Redmond and Smith
29 See Bergin lecture and timeline (pp 14, 21-23, this volume) and Lee (1995a),
Right! The ENIAC room and the room below it, which had IBM equipment,
were the only two rooms in the whole of BRL that had air-conditioning.
Yep! And anybody who was smart in the computing fi eld said that they had
to have equipment in a cool room, and therefore had to have air-conditioning.
Well guys, I think we're just about at the end of our time. I want to thank
both of you. Anybody who wants to, I'm sure, may catch up with you and ask you
for some of your reminiscences. Any last words before we adjourn this session?
I'd certainly like to thank both of you. It's been a great pleasure. I met
Herman finally after Jimmy Prevas30 and other
people kept telling me about you. I finally met you last January, I guess.
And I finally caught up with Harry Huskey last night. One of the great
pleasures of this whole celebration has been meeting you guys. Why don't we
adjourn then? Thank you both again. [applause]
30 Chief of the Firing Tables Branch, BRL, in the 1940s and 1950s.