Atlas Computer

Jump To Main Content

Jump Over Banner

Home

ACLTechnologyATLAS

Jump Over Left Menu

London Atlas, Additional Material

Dik Leatherdale

In addition to the University and British Petroleum, there were two other parties which had a major interest in the London Atlas.

The first of these was Professor Richard Buckingham's Institute of Computer Science (ICS) which was a university centre of excellence, separate from the various colleges, running a masters degree course. The Institute was located on the same premises as Atlas occupying part of a Georgian terrace in Gordon Square, Bloomsbury part of which had once been the home of Lytton Strachey and, indeed, Christopher Strachey! The Institute contributed numerous items of software to Atlas prominent amongst which were :

Atlas Commercial Language (ACL)

A programming language devised by David Hendry bearing a superficial resemblance to COBOL. For various reasons, Atlas would have struggled with COBOL so the flavour of the language was adapted to make is word-orientated rather than character-based. Various other changes were made to the language which became a major workhorse of the bureau operation (see below). ACL supported sterling variables in which money was held as (old) pence but conversion from and to £sd took place on input (from cards) and output to the printer. In 1971, towards the end of Atlas's life, ACL users were given the choice of having their programs re-written and data converted to meet the requirement for decimalisation, or to continue to calculate in old pence with updated I/O routines. Despite the inherent rounding error, all opted for the latter!

David Hendry went on to devise BCL - a completely unrelated language with some concepts in common with the Brooker-Morris Compiler Compiler. Sadly it was not a success.

Multi-Variate Counter (MVC)

Andrew Colin devised two linked programming languages which were aimed at the analysis of survey information, both sociological and commercial. Information from questionnaires would be transcribed onto punched cards potentially using all 960 holes. This data was then read by a program written in the first MVC language in which the programmer specified a procedure for reading a single questionnaire. this was largely, but not exclusively, a matter of defining variables and associating them with particular fields in the card(s). There were some unusual variable types. In addition to the usual numeric and binary (Boolean) types, there was polylog (multiple choice question - single answer) and polybin (multiple answers possible). Facilities were provided for deriving new "questions" from the original data, for iteration and validation.

The second MVC language was concerned with specifying the output that was required. This was in the form of tables usually or one or two dimensions which counted the number of respondents in each category or made simple statistical calculations. Tables whose width exceeded the 120 characters available on the printer were split so that they could be displayed side by side, as necessary. the variable names previously defined were used as row and column headers to which end, variable names were unusually long and were allowed to contain spaces.

CPL

The final interesting software item to emerge from ICS was CPL. CPL (Combined programming Language, Cambridge + London or Christopher's (Strachey) Private Language, according to taste) was largely a Cambridge, Atlas 2 effort, but London University also contributed in the persons of Eric Nixon and John Buxton. CPL was devised as yet another set of extensions and changes to Algol 60 in an attempt to become a "universal language". Although CPL was not successful - it saw little use in either place - it's progeny were rather more successful. In particular CPL was the great-grandfather to C one of today's most used programming languages.

Co-located with ICS was Atlas Computing Service (ACS), a straightforward service bureau operation owned by the University and charged with meeting the day to day expenses of running Atlas as well as repaying the loans which had been taken out to buy the machine in the first place. Since Atlas was by far the most powerful computer in London, sales of machine time were fairly straightforward. The bureau was managed by Alec Robinson who could trace his involvement with computers back to the early days with the Manchester Mark I. The chairman was Peter Parker who would later be famous as one of the more successful chairmen of British Railways.

ACS were the major users of both ACL and MVC, building up a large client base for each. In addition, Fortran V was developed by Mssrs Zell, Schofield and Coopersmith. Fortran V was essentially a version of Fortran IV with a considerable superset of extra language facilities taken from the many enhanced versions of the language. Unusually (uniquely?) dynamic arrays were implemented in which the array dimensions could be decided at run time, rather than specified by the programmer as constants.

Towards the end of the 1960, the various colleges started to acquire their own computers and the university's use of Atlas began to tail off. In 1969, a Regional Centre was set up nearby based on a CDC 6600 - a rather more powerful machine than Atlas. The bureau operation took on a life of its own acquiring a CDC6500 in 1971 but only lasted until 1974. Imperial College - one of the major users of Atlas - bought at CDC 6400 at the same time. The London Atlas was switched off in 1972 being the only Atlas 1 never to acquire discs.