' f ; a view of ij the history of I the ISO character code FROM THE EDITOR R. W. Berner Honeywell Information Systems This is another Phoenix, AZ, documented personal recollection of the US development of basic data processing tools. The char- acter code is perhaps the most fundamental of all such tools. We can program without COBOL, but not with- out coded representations of our most common sym- bols for calculation and language. Even the Japanese have found it necessary to represent their most elemen- tary alphabet, Kata Kana, in code suitable for computer processing. ORIGINS PRESSURES AND MOTIVATIONS Encoding of character sets has taken many forms — sema- Some early phore, pressures for code expansion Morse code for telegraphs, and mechanical came from pro-! I linkages gramming languages. or typesetting Univac I, the first computer machines. It did with al- i not become permanent and phabetic input, had a 1 5 -character set on its interchangeable until the advent of printer. The IBM perforated tape and set had mechanical punch keypunch limitations to 48 cards, representing the continuous and characters jfi discrete me- mostly business symbols. Starting in 1 the dia. Except for looms, punch cards were 954, need for FOR-!} used first for count- TRAN (mathematical ing and and arithmetic) symbols to statistical applications, take the ? requiring at first only a same encodings as these business 1 2 -character symbols led to con-i set - 1 0 digits and 2 signs. much || Perforated tape was fusion in used installations that did both types of work. first for message communication, Pressures! fj and the 5-track tape arose for unique encodings 1 [1,3,4,126], In 1 offered 60 characters - 30 encodings 956 September ft had double meaning " was decided «hal the ' following Figure IBM Stretch computer would have Shift and Letter Shift, the other two combi- a 64-bit word, and that the nations. characters would I This code was due to Baudot, be represented ! and the assignment by 8 bits of codes [19,70], This decision had much influence to characters was made (for on sub- alphabet only) on the sequent computer basis architecture. The Los of letter usage Alamos group de- f frequency, to conserve electrical energy. vised its own input character Both perforated set, and IBM built a special tape and cards, i being available, replaced entry typewriter for it 10 . manual [ ], switches as input devices for computers (cards The advent of ALGOL in 1 958, with (in theory) predominating in the US, 1 1 6 single!' I perforated tape in Europe and symbols, was the impetus for many elsewhere). Most early computing work was proposals. H. S Bright: i computational, proposed one such and the set to the IBM SHARE organization punch card set did not ont ! receive much impetus for 8 September 11, 1 enlargement as a “point of departure for " until application to business problems. hardware implementation of In the IAL“ [4], Often one the late 1 it to be ahead of his 950s became apparent that code expansion | j time, his tableau showed the European had grown uncontrolled. There were “one" and the letter 3 more than ’ 20 varia- °h Wlth a curlicue. tions of the Yet he objected to SHARE asking punch card code [1 29,13 IBM i! { 1J, and more than 60 for more than different 128 characters presentable in internal representations media and 1) ! in computers [12], equipment unless someone was able to The Baudot prove the need. ! Code was not without its variations, J although Others foresaw the the need very well, and there 1 CCITT (Consultative Committee was much f International for Tele- work going on graph within IBM at the time, following and Telephone) Working Alphabet on the f }j No. 2 was used for Stretch decision. interchange The 1403 chain printer was being throughout much of the world. Even devel- so, the oped, the 240 positions around binary the chain permitting f positional 2 sets i l l notation was perturbed by different assign- of 1 20 or 4 sets of 60, in addition to the planned ments of bits to the tracks. 5 sets of l’ j 48 for the traditional IBM set. $ ! 274 HONEYWELL COMPUTER JOURNAL v/o\ q ISO (International Standardization Organization) EARLY STANDARDIZATION EFFORTS In the Sturen) that recom- it was Sweden (probably Olle • area, III processing standards. mended, in late 1 959, activities in data future for computers became clearer in 1958 and The real Geneva on 1961 outgrowing the period This led to a Round Table Conference in 1959. Data processing centers were of ISO/TC97, affair, May 16, which in turn led to the formation when information processing was primarily a local had Computers and Information Processing, for which the US some connection to communications networks and Working Croups were interchange, as for was assigned the secretariat. Several been made. First to emerge was data Character Sets. Service established, in particular WG B, on Coded ^ial Security Administration and Internal Revenue j formed in 1961 May, and had liaison program interchange, more difficult be- ECMA had just been ; records. Then came meeting. Its corresponding varied. representation at the Geneva I cause both symbols and encodings projects began at body for codes was Technical Committee 1. Several independent standardization about this time: US CONTRIBUTIONS Institution (BSD, starting with paper : British Standards j The punched cards, gradually moving into the gen- tape and The 1960 January meeting of ASA had led to the formation problem. eral of Sectional Committee X3, with the same name as TC97. ‘ SHARE, seeking to coordinate its IBM equipment. and X3.2 (Codes X3 held its first meeting on 1 960 August 4, which also started The Electronic Industries Association, by I.C. Lig- and Input-Output) its first on October 4, chaired aspect, partly due to impetus I first from the paper tape con- gett of IBM, who was directing IBM efforts (other work from Numerical Control. established tinued in the US until X3.2 jurisdiction was i;j* Department of Defense, for, whom the Signal The US authoritatively). Membership was from the Federal Govern- developed the Fieldata Code (which, despite Corps ment, data processing, and communications interests. drawbacks, was a great improvement on existing codes, The latter were represented primarily by the Bell System, ISO code). The and many of its features are seen in the to dial which was to convert its TWX service from manual in the Uni- Fieldata Code still exists as the internal code operation in 1962. Along with this was a new transmission vac 1107, 1108, and 1110. in arrangement that enabled 100 word-per-minute service that the addition to the existing 60 wpm service. This meant there IBM was also doing much work in codes, for in 1959 majority of teletypewriters (Model 15, limited to 75 wpm) were 9 different internal codes existing in IBM equipment, would have to be replaced. To Bell, and to John Auwaerter of and interchange on media was a problem. Development Teletype, this seemed an excellent opportunity to escape and of 8000 series (precursor of the 360) was under way, the from the limitations of 5-level Baudot Code and the 3-row between group (in IBM) was planning the coordination my keyboard to a 4-row keyboard more similar to conventional and media representation, following on the Stretch codes typewriters. The Bell System was not only an active partici- 1959 November memo to V.P. John McPherson work. A pant, but actually hastened development by adopting con- used by all outlined a plan for a "single character set to be them clusions as fast as they were reached and incorporating the external stor- future computer systems ... the code to be tape track in their new designs. It even chose the perforated machine code" (i.e., direct age code as well as internal it assignment in advance of that standard, and thus made 8-bit set. representation in punch cards, as well). It was an of rather easy to reach agreement because of the volume Cosden saw this work in the Fall of 1959 and ar- John Model 33 and 35 equipment in actual operation. be invited to make a presentation to the BSI ranged for me to X3.2 did extensive work on development of criteria for a in 1960 February. This codes committee, which occurred collating sequence and code. It considered aspects such as ' established a very important contact point with Hugh natural orderings of subsets, keyboard design consider- colloquially McGregor Ross (for a while the code was known of ations, and programming effects such as manipulation Association) in ECMA (European Computer Manufacturers graphics by classes, simpler language scans, and more circles as the Bemer-Ross code). graphics for clarity of printed output. This was all based upon the assumption of a dense 6-bit code, as used by almost all EXPANDING THE JURISDICTION computers of the time (Stretch was the exception, at 8 bits). No existing code was found to be satisfactory. The Fiel- final stages as Military Standard 1 88 [72], was in its rapid multinational growth, data Code, It was natural that IBM, with offered for consid- adoption in 1 959 December, and was would take a strong interest in the standardization process. of for communications, al- drew up eration. However, it was primarily Jim Birkenstock, V.P. of Commercial Development, the process. For Insti- though computer equipment was used in plans to revitalize the Office Equipment Manufacturers that control codes and Manufacturers data processing, it had the defect tute and convert it to the Business Equipment graphics were intermingled in 3 of the 4 columns.