I think the problem with the codes for Norwegian arises from the existence of three codes. If one is using codes "nob" and "nno," then what does "nor" represent? A colleague of mine pointed out that a similar situation exists in modern Greek with Demotic and Katharevusa usage. Will we have to deal with this situation also? When 639-2 was developed, the code "esk" for Eskimo was eliminated because the four added codes for Yupik, Greenlandic, Inuktitut, and Inupiaq covered all the Eskimo languages. If we have separate codes for the two forms of Mari or the two forms of Sorbian, what use would be made of the existing codes for Mari or Sorbian? Perhaps we should have a principle that the "collective" code is discontinued (withdrawn, deprecated, etc.) when all constituent elements are represented by other codes. Milicent Wewerka