Iterations – Norberg –
Software Development at EMCC
3
Very early in the history of EMCC, John Mauchly assumed responsibility
for programming, coding, and applications for the company’s planned
computer systems. His early interaction with representatives of the Census
Bureau in 1944 and 1945, and discussion with people interested in
statistics, weather prediction, and various business problems in 1945 and
1946 focused his attention on the need to provide new users with the
software to accomplish their objectives. He knew it would be difficult to
sell computers without application materials, and without training on how
to use the systems. As others in the company designed the hardware and
logic of the EDVAC II, as it was then called, Mauchly focused his
attention on software questions.
2
There was no organized software and
applications department during 1947; Mauchly designed the organization
as time transpired and machine design determined needs. Nevertheless, in
early 1947, he began to recruit a staff of mathematicians interested in
coding for a new department he intended to organize as president of the
company.
The early members came with experience in coding for various computer
systems including ENIAC. Frances Elizabeth Snyder (later Holberton)
joined EMCC in 1947. She had already had an illustrious career in
computing, as one of the original members of the ENIAC computing
group. In 1942, shortly after graduation from the University of
Pennsylvania with a degree in journalism, she joined the Philadelphia
Computing Unit at the Moore School. This group worked on problems
associated with the tables being produced by the Aberdeen Proving
Ground. Snyder, and another member of the group, Betty Jean Jennings
(later Bartik), developed a trajectory program used to control the operation
of the ENIAC during the public demonstration in February 1946. In 1947,
Snyder transferred to Aberdeen when the ENIAC was moved there. She
served EMCC as a part time consultant in February through April 1947,
and in July of 1947, she left the civil service and became an employee of
EMCC.
3
She stated later that she had actually asked Mauchly if she could
join EMCC.
4
Jean Jennings Bartik graduated from Missouri State Teachers College
(now University) in 1945, with a mathematics major. She studied analytic
geometry, trigonometry, and physics. In summer 1944, she worked at
Pratt-Whitney Aircraft in Kansas City on engine work, but did not want to
engage in that type of work after graduation. She applied for a job at the
Aberdeen Proving Ground for their location at the University of
Pennsylvania. After two months, she received a letter offering her the job
and she left for Philadelphia the next evening. Not long after, Jennings
applied for a position as an ENIAC coder, and she, Snyder, and several
others were chosen and sent to Aberdeen in June 1945 for training. After
several months at Penn coding for the ENIAC, Jennings was selected to
Iterations – Norberg – Software Development at EMCC
4
head a group to generate programs to turn ENIAC into a stored-program
system. She decided not to move to Aberdeen when ENIAC moved there,
and after completing this programming work she accepted a job at EMCC
in early 1948.
5
With these accomplished mathematicians, the department immediately
engaged in a number of projects while they acquired knowledge of the
needs of users to whom they expected to sell UNIVAC systems. Many
conversations occurred between Mauchly and Snyder, on the one hand,
and prospective customers on the other. Snyder remembered early trips to
the Census Bureau, NBS, and Martin Marietta. Later, she visited many
UNIVAC I sites for consultation on programming problems.
6
As she
pointed out in her 1983 interview for the Charles Babbage Institute (CBI),
she had no training in computing, no courses in formal logic. Indeed,
when she interviewed potential programmers for EMCC and elsewhere,
she asked such questions as “Did you like plane geometry?” in an effort to
establish whether the person had a puzzle-solving ability, which might
make them good programmers.
7
But this line of questioning did not
always produce the desired result. In an interview in 1990, Jean Jennings
reported that in EMCC
We used to argue about what kind of a person [made the best programmer]. We
had this little test that Art Katz worked out. We used to give these people this
little test… I don’t know what anybody else used, but their enthusiasm for doing
something new was what always impressed me personally. …Hildegard
Nidecker came along, who had much experience in doing calculations for the
Army. And she flunked his test, so nobody wanted to hire her. Then he [Katz]
said, ‘this is ridiculous. This just proves to me that the test is ridiculous. We
know that she is going to do a good job,’ and in fact [we hired her] and she
retired from UNIVAC [in the 1980s]. So the truth is we did it by the seat of our
pants, and I personally did it if I liked the person.
8
Whatever the evaluation scheme, an effective group was assembled at
EMCC.
The applications group grew slowly between 1947 and 1950. M. Jacoby
joined the firm in December 1947. Dr. Arthur Katz came to EMCC shortly
thereafter in February 1948 and, as mentioned, Jean Jennings became an
employee at the end of March 1948. Four more people joined in the
second half of 1948 and early 1949—M. League, V. Hovsepian, Hubert
M. Livingston, and Arthur J. Gehring. Besides the special studies for
prospective customers noted above, Snyder worked closely with Mauchly
on the early codes (instruction sets) for EDVAC II and BINAC, codes C-1
through C-5. More special studies involved work for the Army Map
Service, Oak Ridge, and Glenn L. Martin Company, all completed by
early 1949. Snyder worked on a floating decimal routine (at one point
along with Katz), a reciprocal routine, double precision operations, and
reciprocal square roots. Mauchly, while working with Snyder on the code,