COINCIDENCES LED CLOYD DAKE GULL TO BECOMING A PIONEER INFORMATION SCIENTIST


 Cloyd Dake Gull

    I was a traditional professional librarian when called to duty in January 1942 by the Navy. By the summer of 1944 I was the Lieutenant in charge of the Ship's Library at the Naval Training School at Cornell University, where we provided thousands of textbooks to undergraduates and officers. It was a warehousing job, and I adjusted by adopting minimal records. Then I was transferred to the Fleet Post Office in New York City for training as a postal officer. It was evident that the need for postal officers had already peaked, so I requested permission to seek another assignment. When I reached the office in the Bureau of Naval Personnel in Washington, the officer had one request to fill to finish his day. A professional librarian was needed, and in New York City, too! Coincidence number 1. The Commanding Officer of the Service Parts Agency of the Bureau of Ships suspected that his contractor, an engineering firm, was not doing the library part of the contract properly. My orders to join SPA arrived very soon.

SPA required all new Naval personnel to learn how to read engineering blueprints, which were included in manufacturers’ instruction books delivered with shipboard machinery. BuShips had a team at Mechanicsburg, PA recording data that would provide all part numbers for every replacement part to be stored. For example:

Jones’ Fresh Water Pump, Model R-9                       Smith’s Ventilating Fan, Model V-21
Main Axle Bearing 43-20                                           Front Axle Bearing 6-44-2
In this fictitious example the two manufacturers supplied New Departure Ball Bearing ND 773, and they supplied only their own part numbers on the blueprints. Later at a Congressional hearing, one officer admitted there were 1298 part numbers for one ND bearing. without saying how many unnecessary bearings were resting in Naval Supply Depots. In the spring of 1945 it became clear that the team would be unable to complete its work before the United Stats' began invding Japan. It was essential to locate parts already in forward depots and eliminate delays in getting parts from the mainland.  The CO of the SPA wondered what could be done to provide a short cut to achieve the desired cross-indexing.

Each new person was also required to learn how to operate the full line of IBM punched card accounting machines, which were used to uncover this growth of part numbers. I finished the two assignments and turned to the study of the library of manufacturers instruction manuals, some 7,000 titles for World War II.  The CO’s suspicions were correct; the contractor was providing three dictionary card catalogs where one was sufficient. So I threw two away and improved the third one. I also adopted the spirit duplicating process to make the catalog cards. The SPA library catalog was under control, but it was not available in the supply depots nor on board ships, where it was really needed. I was not asked to consider whether or how the dictionary catalog could be supplied to the depots to facilitate the finding of the parts in stock. We knew it could be published as a book.

I analyzed what the depot personnel needed to locate parts for shipboard use.. The sailor could bring the instruction manual and the damaged part(s) to a depot, where the parts man could look on the shelf for the part number. If in stock, everyone was happy. If not, we would try someone’s memory. If memory failed, then we would search other instruction books for the same machine or component. Since the library catalog supplied access by the following categories -- Old and New Library accession numbers, Ship Type and Number, Machinery and Components, Manufacturers names -- this information would be very useful in every supply depot in book form.

I then responded to the CO’s invitation to suggest a short cut, using IBM cards and machines to print copy for a book form index instead of 3 x 5 cards, typewriters and spirit duplication. There would be one 80-column punched card for each manual (7,000 punched cards in all). The IBM tabulator was built to print 150 lines per minute, remarkably fast then for an electromechanical machine, but limited to 39 columns to print the English alphabet and 10 Arabic numerals and 3 punctuation marks, and 41columns for Arabic numerals only. The tabulator forced the use of brief numbers and abbreviations in the several columns of the punched cards. The 7,000 cards would be sorted numerically by new accession number and then printed by the tabulator. The cards would be resorted, for example by ship type, and tabulated by that column. When all sorting and tabulating were finished, the Government Printing Office would print the index by photo offset.

My suggestion was accepted, and, during the summer of 1945, some 12 persons worked on the master set of manuals in Washington. We were ordered to finish our task rapidly so GPO could deliver the indexes on September 1. GPO met its deadline, but Japan had already surrendered on August 14.

The Navy released me from active duty early in December 1945, and I started working for John W. Cronin, newly promoted to Assistant Director of the Processing Department of the Library of Congress. The Department held an internal conference on the future of the card catalog to which I was invited. This is where I learned (coincidence number 2) that LC had a full complement of IBM machines, used mostly for billing about 10,000 libraries for their purchases of LC printed catalog cards and for payroll work. I also learned that these men were anxious to learn about new kinds of equipment and procedures: Librarian of Congress Luther H. Evans; in the Processing Dept. Herman H. Henkle, Director and Cronin; Seymour Lubetzky, cataloging principles; and John Meehan, Manager for IBM Operations. I started to draft at home a system using punched cards and machines to supplement hot metal composition and flatbed printing of LC catalog cards. When I finished the draft in July 1947, it demonstrated that punched cards and machines could be used to prepare a dictionary card catalog, but at the price of degrading all the procedures already using hot metal and flatbed printing, with one exception.

In 1947 most users of card catalogs were forced to copy references from them by hand onto blank cards. The draft showed that mechanical copying could replace manual copying if punched cards were used. But the size of the task of preparing a punched card catalog was just too great for serious consideration.

In the spring of 1946 the Veterans Administration asked Dr. Evans for help. He authorized a project named Surplus Books for Veterans and designated Dr. Leslie Dunlap and me to set it up. The VA owned some 5,000,000 college-level textbooks in colleges and universities where servicemen had been educated during World War II. These books were in short supply in the civilian market, and the VA wanted to give them to veterans who were signing up for courses paid for by the GI Bill. The VA had brought the books to Washington where they were stored in freight cars. We took over the basement parking space for 105 cars in the Annex and shelved the books in boxes to be used for shipment to schools. We produced an author catalog from punched cards, one line per title. We mailed the catalogs to the schools and asked them to mark the titles and quantities needed and return the marked catalogs. Our staff picked the required quantities from the boxes and marked each school’s identification number and quantity on every picking card. We made frequent reports to the Librarian and the schools by tabulating the punched picking cards.

It probably would have been easier and less costly to have left the books where they were, because the schools already had faculty members to teach the subjects.

In the spring of 1946 I received an invitation from Norman Hill to present a paper on punched cards to the Division of Chemical Literature of the American Chemical Society. It was titled A Punched Card Method for the Bibliography, Abstracting and Indexing of Chemical Literature and was published in October 1946 in the Journal of Chemical Education 23:500-507. It had been announced as "The Bibliographic Revolution," and, while it created quite a stir at the meeting, the procedures presented no useful improvement over the printing techniques already in use for Chemical Abstracts, but the presentation put me in touch with others who believed that improvements were absolutely essential. And history has vindicated the original title.

The Librarian asked for additional help from IBM. One request was for a 7.5 x 12.5 cm unpunched card mechanical distributor to reduce human labor, which had been used for nearly 50 years. As a contribution to world bibliography, LC gave complete sets of LC cards every year to some 206 libraries. In 1948 production was about 65,000 titles (which required some 70,000 cards to care for titles requiring more than one card per title) or about 14,420,000 cards per year. The GPO delivered 65,000 packages per year in random order as to author and title order. LC wanted to give away 206 sets of 70,000 cards in author-title order. The total was about 14,420,000 cards to rearrange per year. To achieve author-title order it sufficed to order the packages on shelves as they came from GPO each year. To convert 65,000 sets into 206 larger piles required a mechanical distributor.

IBM converted a regular punched card sorter to accept LC’s thicker and narrower card, eliminated the electric sensing feature, and opened the stacking pockets of the sorter, one title per pocket until all cards had been distributed into 206 piles. The successful conversion was used for years after 1948.

It was clear by 1948 that the optimum choice of one feature of IBM cards and machinery usually resulted in the degradation of performance by another feature. I noted that the most frequently requested improvement was the provision of a larger set of characters. After all, IBM manufactured the Electromatic typewriter with 88 characters on its manually operated keyboard, and it could be driven by punched paper tape at 10 characters per second. Why not drive the Electromatic from information stored on punched cards?

Consequently another request to IBM was for a larger character set in printing from punched cards. IBM chose to modify its standard verifier, used to read proof on punched cards column by column, which fitted the way a Card Controlled Typewriter would print, character by character, at 10 characters per second. IBM added the plugboard from the collator so that the signals could be read from the cards column by column and sent to the correct solenoids to activate each key. While familiarizing myself with the CCT, after IBM delivered it, I realized that I was using three columns every time to print an upper case character, one to shift the typebasket into upper case, one for the character, and one to return the typebasket to lower case. When the service representative came to work on the machine, I described the problem. He responded that the machine was so slow that we could wire the plugboard so that shift or unshift was ready before the next keystroke was called for. I wired the plugboard as he had directed, checked all 88 characters, recorded the plugboard wiring in color, and produced samples of LC cards. They were typographically pleasing, but as part of a system could not compete with letterpress printing.

We developed two experimental uses for the CCT in the Navy Research Section of LC. We put the Section’s List of Subject Headings into punched cards, filed the cards by hand, and printed them in several editions. This experiment was used at least until 1961. The second use was to prepare the subject index to the individual issues of the Technical Information Pilot, which was published in three parts to cover unclassified/restricted, confidential, and secret reports. Years later I had to retrieve the wiring diagram from my files for the service representative, for he refused to service the CCT without the diagram.

During my employment at the Library of Congress (1945-1952), I also held administrative posts in the Catalog Maintenance Division, the Navy Research Section, and the Union Catalog Division which had priority over working with punched cards and machines. The literature of punched cards and machines was also changing very rapidly to include a new device, the computer, which was emerging from its wartime security classified status. I began to study the literature of this new device. It was evident that remarkable progress was in prospect for handling all kinds of information, and I offered my views to the men listed earlier. I asked the Selection Officer to route all the literature of computers and punched cards to my attention.

He approved a visit by Dr. Mortimer Taube, then Chief of the Navy Research Section, and me to John W. Mauchly at the Moore School of Electrical Engineering of the University of Pennsylvania, where they were building ENIAC, one of the world’s first digital electronic computers. Toward the end of our visit we asked our basic question: "Can you work with alphabetic characters as well as with numerals and some mathematical symbols?" Mauchly quickly said, "No." When he read the amazement on our faces, he revised his answer: "Yes. I can code letters by using numerals." Although we recognized that coding would require more storage capacity per word, we retained the computer as the pre-eminent system to be watched for information handling in the future.

We did not see the control panels nor the racks of vacuum tubes, but we were shown a mercury delay line and a surveyor’s 1,000-foot steel tape as examples of their current work to develop storage devices. Neither one proved practical.

Last Page | Next Page | Index







Project Coordinator: Dr. Robert Williams     Site Design: Eric Chamberlin     Comments may be sent to: bobwill@sc.edu