Building History

[...] A writer for the Seattle Times described the high-tech machine: “The magnetic drum, operating on the principle of a magnetic tape recorder, can store or ‘memorize’ 20,000 digits. It is used to set up the problem. A problem that would take eight hours to do by hand on a desk calculator can be solved in five seconds with the new machine, university authorities said. A more difficult problem, which would require a week’s work by hand, takes a little more time—about five minutes.”

The UW Departments of Mathematics and Business Administration became the first on campus to access the Research Computing Center’s IBM Model 650’s “electronic brain." The Seattle Times writer noted that the computer would be used for two new classes: “Mathematics students will take a course known as ‘numerical analysis.’ The business students’ course will be called ‘data processing.’ Dr. Carl B. Allendoerfer [See https://en.wikipedia.org/wiki/Carl_B._Allendoerfer ], head of the Mathematics Department and chairman of an intercollegiate committee on high-speed computers, said university departments previously had been sending work requiring electronic calculation to California and New York. The I.B.M. Co. rents the machine to the university, donating 60 per cent of the basic rental fee of about $60,000, a year. The National Science Foundation is contributing $17,500 a year. University research and rental charges to departments using the machine will make up the rest of the cost.” (See “U.W. Opens Regions’s First Research Computing Center,” Seattle Times, 09/11/1956, p. 11.) ... [Basically, IBM and the NSF provided for what was essentially a free computer]

In an adjacent story in the same issue of the Seattle Times, the Boeing Airplane Company announced that it had begun operating a rented IBM Model 705 computer beginning on 09/10/1956. An official ceremony attended by William M. Allen, Boeing’s President and Thomas J. Watson, IBM’s President, launched the computer’s first day of work calculating “…payroll, stores, records, accounts payable, labor and material distribution, material records and other assignments. Savings of more than $500,000 are expected through use of the machine. Rental is $36,500 a month.”

In 2010, the University of Washington chose to rename the Academic Computing Center for John "Mike" Wallace, a renowned UW atmospheric sciences professor. A UW news web site noted two honors bestowed on Wallace in 2010, for his pioneering research on global climate change: "His impact on climate science, at the UW and globally, led to two recent noteworthy events on campus. In September, friends and colleagues from around the world gathered for a two-day Wallace Symposium that was billed as a celebration of his 70th birthday and his many contributions to science, students and the community. This summer, the former Academic Computing Center, which now houses the new College of the Environment, the Program on the Environment and a UW-based institute he headed for many years, was renamed John M. Wallace Hall." (See Vince Stricherz, UW News.edu, “Illustrious career brings singular honors for Mike Wallace,” published 10/07/2010, accessed 09/03/2020.)

The UW Program on the Environment continued to occupy Wallace Hall in 2020.

IBM 650 Computer and other UW computers

When I returned to the UW in January 1959, the hot topic was the IBM 650 computer. Many of us would have to learn how to program and operate this latest addition to the University. Dr. David Decker (1923?-2004) was the director of the Computer Laboratory and taught a class in programming. The Computer Lab was located on the fourth floor of Bagley Hall over the main entrance.

Because oceanographic data processing requires many repetitious mathematical operations, it was deemed necessary that many of us in oceanography learn how to use this device. When I was informed that I would be taking a class in computer programming, I jumped at the chance. The classes started one week before I was officially on the payroll.

The first time I addressed this monster was on, Thursday 8 January 1959. I was petrified as I stood in front of the control panel with its many dials and flashing lights. In about two weeks, we became very well acquainted and the fear level dropped. But the frustration level remained high and still does today with our “very advanced” computers.

The IBM 650 magnetic drum data processing machine consisted of two bays each being three feet wide, six feet tall and eight feet long. Attached to the computer was an IBM card reader/punch. The internal memory consisted of a drum, about the size of an oatmeal box, coated with ferric oxide and was turning at 1,800 RPM. It held a total of 2,000, ten character words in a bi-quinary format. In today’s concept of storage this would be the equivalent of 20 thousand bytes. The computer I am using to write this biography has 4 billion bytes of internal memory and this is considered to be small. Computers have come a long ways in 50+ years.

There were 3,750 vacuum tubes, mostly 12SN7’s, that did all the work. Operation speeds were measured in milliseconds. It could take as much time as 50 milliseconds to multiply two ten-digit numbers. A patch panel inserted into the end of the card reader/punch determined where the data would be placed into or retrieved from memory.

When the 650 would act up, Chet Castle would be called in for the repair. He was the IBM engineer assigned to this 650. One time it took him over three weeks to determine what was wrong with the computer. When he opened a door to a bay, the computer would act correctly. He eventually located a chip out of a capacitor. Dust would accumulate on the exposed wire causing a temporary short. When the bay door was opened the dust would be cleared and normalcy would return. He gave me the module. I kept it on my desk for many years.

A keypunch (IBM model 24 or 26) was used to prepare the data and program on Hollerith (IBM) cards. Each Hollerith (IBM) card had 80 columns and 12 rows where information could be punched in a total of 960 locations. The information on the cards could be in either alphanumeric (A-Z upper case only or 0-9) or binary format. Data sets used alphameric characters while programs were usually in binary format. During my tenure at the UW, I used over one million IBM cards.

In addition, to the IMB 650, the computer lab housed the necessary peripheral devices necessary to feed data and programs into the compute. These devices included: key punches, a tabulating machine, card sorters, interpreters, and a collator. The IBM 650 computer, with its necessary peripheral equipment, required over 30,000 watts of electrical power. We had 4.5 tons of air conditioning to keep the room cool.

The IBM 650 laboratory was a do-it-yourself type with some assistance from the “experts” when needed. When you wanted to use the IBM 650, you signed up for a time slot usually in 30 minute increments. A large blackboard was used to write your name in the appropriate time slot. Frequently there were no Anglo- Saxon names on the board.

If you were going to use the machine after regular hours, there was a last-man- out routine for shutting down the entire system; computer, air conditioning, and any other peripheral equipment. Thus, it was always desirable to be the next-to- last person when using the computer on any given day because shutting down the system and securing the laboratory took about 30 minutes.

[...]

The first programming language we learned for the 650 was the native machine language, the lowest level of programming. Every operation was contained in a ten digit word. A typical instruction may look like 70 1950 0200. The first two digits were the operation code (70 for read a card into memory). The next four digits (1950) was the location of the data to be stored and the last four digits (0200) were the location of the next instruction. Every operation in a program required a separate IBM card so even a simple program such as computing density would take a hundred or more cards . Both program and data were entered into and out of 650 by means of Hollerith Cards.

A short time later, David Decker introduced us to an assembler language that used an alphameric code instead of numbers only to describe the process. An assembler program decoded the symbols and created the machine language code. The finished program was punched on to IBM cards in a binary format. The assembler’s name was SOAP for Symbolic Optimal Assembler Program. I used this assembler for about three years until the 650 was replaced with a larger IBM 704. This computer used a 32-bit word, was much faster, had 16,000 words of memory and a large hard drive. The many individual disks in this hard drive were about 30” in diameter were kept in a refrigerated cabinet.

The computer lab was moved from Bagley Hall to the basement of Roberts Hall which was a much larger room that housed both the 704 and the 650. The 650 was soon eased out forcing us to learn a new program language. The recommended language for scientific computing was FORTRAN. (FORmula TRANslation). At this point in time, I learned FORTRAN and used it for the rest of my career at the UW.

Several more computers followed. A building on Brooklyn Avenue south of Pacific was constructed to house the academic computer. When the building was first opened, there was a room on the south side of the first floor that had over 50 keypunches. Two high speed card readers were available to feed your data into the computer. The last main frame computer to occupy this space was a CDC 6700. This machine used 64 bit words and had a memory of about 64,000 words. There were several magnetic tape drives attached to computer and several large hard drives. This computer supported many peripheral stations that permitted a person to communicate with the main frame from a distance.

Oceanography had such a station in the first floor of the Showboat Apartment. This building was torn down in about 1998 to make room for the new Oceanography Building. Because of the large quantity of data I generated, I had an IBM Model 129 Interpreting Key Punch on the Oceanography Barge.

[...]

[...]

1963. The United Nations Treaty Series Project (UNTS Project) started as a vague hunch in a class discussion. The question was how to prove the growth of international institutions in conventional international law. The answer was simple in theory but difficult in practice. We should survey the texts of treaties. If references to international institutions increased, the point was proved; otherwise, disproved. The more treaties we surveyed the more reliable the results would be. There was no money to pay for anything, no special equipment, no extra work space, nothing but the idea and the library and a group of students volunteering to do the job as part of their classwork in international law.

Two months later, some fifty students had combed through all volumes of the United Nations Treaty Series (UNTS). There were 426 volumes in the autumn of 1963. A few key characteristics were extracted from every one of the 6,734 treaties published up to that time in the UNTS. The students coded these key characteristics on dittoed code sheets. By Christmas I had on my desk a pile of several hundred code sheets filled out by 50 students in as many handwritings. I had no way of guessing the level of accuracy and completeness, let alone usefulness, of the whole exercise.

[...][ page 178 ][...]

Also in 1965, the UNTS Project became part of the regular graduate instruction in international law at the University of Washington. A system was developed (and is still in force) whereby seminar students must combine traditional and quantitative methods in international law. Standard procedures include the handling of UNTS Project printouts, code manuals, and raw data cards. Students also use auxiliary equipment in the computer center as well as desk calculators and other simple devices for minor individual tasks. Measurements and computer-oriented activities must be related to traditional research wherever possible.

1966. This year saw a gradual shift of emphasis from investment to production. Data enlargement continued by adding identifications of international organizations referred to in treaty texts and identity of registering party, as well as some refinements in the 1965 topic code. The second annual updating was completed in June by adding to the data bank UNTS Volumes 454-503 published between June 1965 and June 1966. Data enlargement and updating having become a routine function of an on-going operation, the main thrust of the work in 1966 concerned new programs for the machine-handling of UNTS data, and analytic work in selected aspects of the law of treaties. Completed were the study of the Canadian UNTS Gap," a special study on the role of military matters in the world's treaty pattern,[12] a general statement of theoretical under- pinnings for the UNTS Project, [13] a paper on information retrieval aspects of the UNTS Project, [14] and an outline of a more general scheme of computerizing international law. [15]

Two professional meetings on the use of computers in international relations research included the UNTS Project: (1) the special working group on electronic data processing set up by the American Society of International Law; [16] and (2) the Institute on "Computers and the Policy-Making Community" sponsored by the Endowment for International Peace and the U. S. Atomic Energy Commission. [17]

Also in 1966, two graduate student papers on UNTS Project material attracted professional interest: (1) one paper was selected for presentation to a West Coast regional conference [l8] and was then revised and published elsewhere; [19] and (2) the other paper was chosen for presentation to the University of Washington Student International Law Society and then also published elsewhere. [20] By the end of 1966, a total of 15 students found their work referred to by name in UNTS Project publications,[21] and 8 of them more than once. [22]

Finally, the UNTS Project was given a broader institutional base at the University of Washington through the appointment of a joint advisory committee by the Dean of the Law School and the Chairman of the Political Science Department.

1967. Three new sets of data were added in 1967: (1) date of effect, (2) date of registration and (3) official languages. New work began on multilateral treaties and on treaty history subsequent to registration, e.g. prolongations, terminations, accessions, withdrawals, amendments, and reservations. The annual updating proceeded concurrently with the addition of new data. UNTS Volumes 504-550 were coded. This brought our data bank up to date, covering all UNTS volumes published through December 1967.

The computer center started developing for the UNTS Project a teletype capacity for instant retrieval between the UNTS Project office and the computer center. This is being done by means of the B-5500 XTAB retrieval program adapted by the computer center of the University of Washington especially to the needs of the UNTS Project. It will be used in addition to the existing capacity of the IBM-7094 for both retrieval and computation. The B-5500 program plus teletype response will cut the waste and the waiting time involved in using the IBM-7094 (or later the IBM-360) for simple retrieval jobs. [23][Grateful recognition for imaginative help far beyond routine cooperation is made to the two co-directors of the computer center at the University of Washington, David Dekker and Carl Young, and to the following staff members in particular: Charlene Haug, Elton Hay, Larry Menninga and Monique Rona.] Hunches can then be tested in minutes rather than days. Retrieval also has become a somewhat larger part of UNTS Project operations as more faculty and graduate students use our data bank and as more requests for special data printouts are received from outside. [24]

[...]

The United Nations Treaty Series Project - By PETER H. ROHN UNIVERSITY OF WASHINGTON

[...]

1963. The United Nations Treaty Series Project (UNTS Project) started as a vague hunch in a class discussion. The question was how to prove the growth of international institutions in conventional international law. The answer was simple in theory but difficult in practice. We should survey the texts of treaties. If references to international institutions increased, the point was proved; otherwise, disproved. The more treaties we surveyed the more reliable the results would be. There was no money to pay for anything, no special equipment, no extra work space, nothing but the idea and the library and a group of students volunteering to do the job as part of their classwork in international law.

Two months later, some fifty students had combed through all volumes of the United Nations Treaty Series (UNTS). There were 426 volumes in the autumn of 1963. A few key characteristics were extracted from every one of the 6,734 treaties published up to that time in the UNTS. The students coded these key characteristics on dittoed code sheets. By Christmas I had on my desk a pile of several hundred code sheets filled out by 50 students in as many handwritings. I had no way of guessing the level of accuracy and completeness, let alone usefulness, of the whole exercise.

[...][ page 178 ][...]

Also in 1965, the UNTS Project became part of the regular graduate instruction in international law at the University of Washington. A system was developed (and is still in force) whereby seminar students must combine traditional and quantitative methods in international law. Standard procedures include the handling of UNTS Project printouts, code manuals, and raw data cards. Students also use auxiliary equipment in the computer center as well as desk calculators and other simple devices for minor individual tasks. Measurements and computer-oriented activities must be related to traditional research wherever possible.

1966. This year saw a gradual shift of emphasis from investment to production. Data enlargement continued by adding identifications of international organizations referred to in treaty texts and identity of registering party, as well as some refinements in the 1965 topic code. The second annual updating was completed in June by adding to the data bank UNTS Volumes 454-503 published between June 1965 and June 1966. Data enlargement and updating having become a routine function of an on-going operation, the main thrust of the work in 1966 concerned new programs for the machine-handling of UNTS data, and analytic work in selected aspects of the law of treaties. Completed were the study of the Canadian UNTS Gap," a special study on the role of military matters in the world's treaty pattern,[12] a general statement of theoretical under- pinnings for the UNTS Project, [13] a paper on information retrieval aspects of the UNTS Project, [14] and an outline of a more general scheme of computerizing international law. [15]

Two professional meetings on the use of computers in international relations research included the UNTS Project: (1) the special working group on electronic data processing set up by the American Society of International Law; [16] and (2) the Institute on "Computers and the Policy-Making Community" sponsored by the Endowment for International Peace and the U. S. Atomic Energy Commission. [17]

Also in 1966, two graduate student papers on UNTS Project material attracted professional interest: (1) one paper was selected for presentation to a West Coast regional conference [l8] and was then revised and published elsewhere; [19] and (2) the other paper was chosen for presentation to the University of Washington Student International Law Society and then also published elsewhere. [20] By the end of 1966, a total of 15 students found their work referred to by name in UNTS Project publications,[21] and 8 of them more than once. [22]

Finally, the UNTS Project was given a broader institutional base at the University of Washington through the appointment of a joint advisory committee by the Dean of the Law School and the Chairman of the Political Science Department.

1967. Three new sets of data were added in 1967: (1) date of effect, (2) date of registration and (3) official languages. New work began on multilateral treaties and on treaty history subsequent to registration, e.g. prolongations, terminations, accessions, withdrawals, amendments, and reservations. The annual updating proceeded concurrently with the addition of new data. UNTS Volumes 504-550 were coded. This brought our data bank up to date, covering all UNTS volumes published through December 1967.

The computer center started developing for the UNTS Project a teletype capacity for instant retrieval between the UNTS Project office and the computer center. This is being done by means of the B-5500 XTAB retrieval program adapted by the computer center of the University of Washington especially to the needs of the UNTS Project. It will be used in addition to the existing capacity of the IBM-7094 for both retrieval and computation. The B-5500 program plus teletype response will cut the waste and the waiting time involved in using the IBM-7094 (or later the IBM-360) for simple retrieval jobs. [23][Grateful recognition for imaginative help far beyond routine cooperation is made to the two co-directors of the computer center at the University of Washington, David Dekker and Carl Young, and to the following staff members in particular: Charlene Haug, Elton Hay, Larry Menninga and Monique Rona.] Hunches can then be tested in minutes rather than days. Retrieval also has become a somewhat larger part of UNTS Project operations as more faculty and graduate students use our data bank and as more requests for special data printouts are received from outside. [24]

Early in 1967, the second meeting of the automation working group of the American Society of International Law gave the UNTS Project major coverage as one of two "mature projects" in the field.

[...]