Íslenska

Yfirlitsglærur um íslensk tölvustafróf (PDF 1400K). Oddur Benediktsson.
ISO 8859-1 stafatafla með íslensku (PDF 10K).

Samantekt á ensku vegna ráðstefnu um sögu tölvunnar á Norðurlöndum 2003:

Computers and the Icelandic alphabet

The basic character set for early data processing consisted of the 26 characters in the English alphabet together with the 10 decimal numerals. The Icelandic language has 10 letters that do not occur in English, which means 20 when both upper and lower case are considered. This, together with the tiny market, has cost the Icelanders many headaches and added workload and it took ingenuity to make printed alphanumeric data look reasonable on printed output and display devices. Throughout the Unit Record period (1952 – 1964) the repertoire was limited to 37 or 38 characters, which meant that compromises had to be made. Out of the 10 additional Icelandic letters, four were considered essential for printing understandable names, namely Ð, Þ, Æ and Ö. The remaining 6 are vowels written with acute accent. In the tight situation it was decided to accept the plain unaccented vowels as substitutes as it would normally not cause misunderstanding. It is clear that not everybody was happy, though, to have their names “mutilated” in official documents. By dropping Q, which is not used in the Icelandic alphabet, and using the same glyph for 1 (one) and I as well as the same modified glyph for 2 and Z, this was possible.

The IBM 1401 computer was capable of handling 48 characters, and so was the first generation of the IBM 1403 printers, which meant that it was no longer necessary to have the same glyph stand for two different letters/digits, but there was still only room for two of the accented Icelandic letters and the characters Á and É were selected. On hindsight, it was of course rather useless to take on only 2 of the 6 accented letters.

A revolution can be said to have taken place with the Extended Binary Coded Decimal Interchange Code (EBCDIC), introduced by IBM with the 360 computer line. This was an eight-bit code with 256 different code points and thus plenty of room for letters specific to other languages than English. So, with the IBM 360/30 we got the capacity for a full Icelandic alphabet without sacrificing anything else.

It is a remarkable coincidence that at almost the same time the American Standards Association (ASA, which later changed its name to the American National Standards Institute, ANSI) published the first version of it’s 7 bit ASCII code (American Standard Code for Information Interchange), which was adapted by all American computer manufacturers except IBM[1]. To adapt ASCII to languages using accented letters there were 10 open code points for national adaptation. In Iceland, a version using a so-called floating accent was used, where the accent was actually stored separately, ahead of the letter it was supposed to sit on top of. This solution meant difficulties when sorting. Also a special adaptation to most printers was needed in order to combine the two. This solution was incompatible with EBCDIC.

The 7-bit limitation was a general problem in Europe and many other countries. An extension to 8 bits was needed. In the seventies, ECMA, the European Computer Manufacturers Association, started working on an 8-bit code table that would address the needs of the various European and other languages for which ASCII was not sufficient. A code table was proposed by Wilhelm Friedrich Bohn, the IBM representative in the ECMA working group on code tables and keyboards[2]. This table, nicknamed the Bohn Code was released in 1985 as the ECMA-94 standard. It was adapted by ISO and is widely known as ISO 8859-1, or Latin -1. 

Mr. Bohn was familiar with the requirements of the Icelandic language, and since there were not at that time, contrary to what later happened[3], other serious contestants for the code points in question, the complete Icelandic alphabet was part of his proposal for what later became known as the Latin-1 code page. The Icelandic government has duly acknowledged Mr. Bohn’s contribution to the Icelandic information society.

As personal computers, IBM PC and clones, capable of handling 8-bit character sets appeared on the Icelandic market, each importer created its own version of an Icelandic code table. By around 1985 there were at least four versions in circulation, plus one for the Apple Macintosh, causing difficulties in exchanging data. Jóhann Gunnarsson, then manager of the University Computing Centre in Reykjavík, called some meetings between the importers and a number of key users. Eventually, he proposed a compromise code table, sometimes named the JG Code, which was adopted by all PC and PC clone importers. Jóhann’s attempts at persuading the Apple community to adopt the same code points for the Icelandic letters were not successful, however.

In due time, the JG Code was gradually replaced by IBM’s Code Page 850, which, as far as placement of the 20 extra Icelandic letters is concerned, is identical to ISO 8859-1.

[1]Steven J. Searle. A Brief History of Character Codes in North America, Europe, and East Asia http://tronweb.super-nova.co.jp/characcodehist.html

[2] TrueType Fonts, http://www.norasoft.de/tt-fonts.html

[3] Turkey, for instance, has tried to persuade the standards community to exchange at least two of the Icelandic letters in ISO 8859-1 with letters from the Turkish alphabet.

jg 2003
Jóhann Gunnarsson