BIT—a child of the computer

The back-ground of the Scandinavian computer journal B I T will be outlined, in particular with respect to computational demands in science, technology, industry and defense. The history of B I T will be described and related to the evolution of computers, numerical mathematics and computer science. Some contributed papers which have had an impact on the general development will be discussed briefly.

The 19th century could perhaps be characterized as a period of preparation for the advent of the computer. It so happened that quite a few Swedish inventors played a role in this development. Scheutz, father and son, as well as Wiberg constructed mechanical devices for a somewhat automatized calculation for solving simple arithmetic problems by series of pre-determined operations. In fact, Wiberg was able to compute a logarithm table which even appeared in print. Later, Odhner built a robust mechanical, hand-driven calculator which around 1930 was followed by electromechanical calculators. All lengthy calculations had to be performed manually by this time. Let me mention a few examples from Sweden.

One such problem was to find periods of so-called internal waves in the sea. These waves are huge, up to 20-30 meters in size, but nevertheless invisible. They are generated by the moon and observed as sharp changes in the salinity. The method used was numerical autoanalysis, that is a kind of Fourier analysis of the function by itself.

During the war there was a great need for ballistie tables, and I belonged to a group involved in computing bomb tables for the Swedish Air Force. We used the classical Runga-Kutta method with air resistance represented graphically, and we had only electro-mechanical calculators at our disposal. After having computed a basic set of orbits we could produce the wanted tables by a suitable interpolation procedure. It is a sad fact that all our tables could probably have been computed in a couple of minutes on a fast modern computer. After the war I was involved in some rather lengthy computations on the deuteron concerning energy levels and quadrupole moment and also in problems on scattering.

However, in 1946 some people in the Swedish Navy and in the Academy for Engineering Sciences got interested in the progress in the United States and after having visited the key projects they reported back with great enthusiasm. It was quickly decided to offer scholarships to four young students; they were selected in the spring of 1947. They arrived already in August or September; two of them went to Harvard and MIT while two, including myself went to Princeton. As far as I am concerned I enjoyed a phantastic hospitality, particularly from Herman Goldstine with whom I established a lifelong friendship. Back home in 1948 some of us got involved in the construction of a relay computer (BARK), completed in 1950. However, it was soon understood that there was a need for more computer facilities, and the construction of BESK under Erik Stemme was initiated. It was completed in 1953, and during a short period of time it was considered as one of the most powerful computers in the world. Clearly its structure was very much the same as that of the Princeton computer.

In 1956 a simplified copy of BESK called SMIL was completed at Lund University, built with a minimum budget of some 20,000 $. This computer was used for a large variety of problems in nuclear physics (particularly eigenvalue problems), spectroscopy, mathematics (number theory, construction of tables), and also social sciences (geographical simulations). Several problems coming from industry and different research organisations were also treated.

The interest in and use of computers created a very natural demand for conferences since the literature on the subject for obvious reasons was very scarce by this time. The first Scandinavian conference on computers was held May 1959 in Karlskrona, later known as the place where a Soviet submarine ran aground in 1981. One reason for this choice of site was the fact that the Swedish Navy played an important role in initiating the computer development, another that, especially in spring, it was a lovely place, situated on the Baltic. Preliminary discussions were held informally on the need for a Nordic journal on computer related problems, and at the next conference in Copenhagen in 1960 a more formal meeting was arranged. Niels Ivar Bech acted as chairman, and further Peter Naur of Denmark, Jan Garwick, Norway, Olli Lokki, Finland, and myself from Sweden were present. It was unanimously decided to start a Nordic journal within the computer area, to appear quarterly. The journal was intended to be international with papers published in English, German, or the Scandinavian languages. As it turned out, only about 4-5 papers have been published in German, and very soon papers in the Scandinavian languages gradually disappeared. Nowadays it is required that all papers be written in English.

The name of the journal was a long one: Nordisk Tidskrift for Informationsbehandling, but playing around with the initials in a clever way we were able to form the name B I T. In fact, this name is most convenient because of its shortness which makes it very easy to quote papers printed in the journal. As is well known it is somewhat dangerous to suggest an activity including work since there is a great risk that the proposer is elected to carry through the project. This is exactly what happened in this case, and from the very beginning up to this time I have served as Editor of B I T. Naturally, we have also an Editorial Board with members from the Nordic countries. Peter Naur of Copenhagen has been a member right from the beginning in 1961 and Germund Dahlquist from 1962. We got financial support from the Danish company Regnecentralen under Niels Ivar Bech and from several official sources including the Nordic Research Organisations for Natural Sciences. Finally, just a few years ago we managed to become self-supporting, perhaps mostly through favorable exchange rates.

During the first decade B I T tried to @@@@ the public to get acquainted with new developments within the computer area. It is natural that the growing crowds of people working with computer applications of different kinds felt an increasing difficulty in keeping up with the fast progress, both in hardware and in software. That left a gap which B I T tried to fill. Simultaneously we also tried to accommodate scientific papers, particularly in numerical analysis and in computer languages. Very early we opened a special column for algorithms written in ALGOL 60. As a consequence of this policy our subscribers to a large extent were private Scandinavians during the first decade. Then the situation changed slowly. The need for general information decreased because this was treated in special new publications of type Datamation and also in ordinary and weekly newspapers. Simultaneously the number of scientific contributions to B I T increased strongly, first in numerical mathematics, later also in computer science. As a result of this development the number of Scandinavian subscribers decreased while the number of non-Scandinavian subscribers, mostly libraries of research organisations and universities increased, the net result being slightly positive. From 1980 it was clearly indicated that B I T was divided in two sections, one for Computer Science, and one for Numerical Mathematics. In spite of obvious difficulties we have been able to strike a reasonable balance between these two.

The first volume (1961) had 290 pages and was type-written and photographed. Already volume 2 was printed in an ordinary way. B I T had obviously been observed also abroad since two contributions, one from the US (Louis Fein) and one from the Netherlands (Peter Wynn) appeared already in the first volume, while several “foreign” papers (among them one by Gene Golub) were presented in volume 2. From the beginning there was a certain ambivalence with respect to papers on hardware: during the first 10 years we published a few of that kind, but finally they disappeared.

Turning to computer science there is an important subject which attracted considerable attention during the first 10-15 years, namely computer languages and compiler construction. The main interest was centered on ALGOL 60 since by that time FORTRAN was only available for users while the corresponding compilers were held secret. However, different aspects on other programming languages, e.g. COBOL, ALGOL 68, PASCAL and SIMULA, have also been treated.

It is of course hard to tell which papers have had an impact on the general development, but I think that papers by Dahlquist and others on stability problems, Enright-Hull-Lindberg on numerical methods for stiff systems and a whole bunch on Runge-Kutta methods have had a considerable influence. Finally I think it is fair to mention that we offered a special issue dedicated to Germund Dahlquist on his 60th birthday, followed by one dedicated to B I T on its 25th birthday, both with about 300 pages.

Concerning the geographical distribution of authors and subscribers we can say roughly that the Nordic countries, the rest of Europe, and USA plus Canada account for about 1/3 each in both respects. The most striking feature is the steep increase in offered contributions from Taiwan, and we have also had quite a few from mainland China. In both cases the quality has been rather good. Also some exotic countries are represented by authors: Nigeria, Singapore, Ecuador, Sudan and the Fiji Islands, just to mention a few. Even if some papers must be rejected we try to encourage the authors, and in many cases the papers can be published after a more or less thorough revision. As a mean value the time between reception of a paper and publication is nine months.