Ever since the decipherment of Linear B, a large amount has been written on Mycenaean spelling. The basic rules were established very quickly, although the debate is still continuing as to the theoretical basis of Mycenaean spelling; various models have been proposed here. One of the most serious problems for all existing models is that at the edges, there is spelling variation that simply cannot be captured by one and the same algorithm. Here, we shall look at two such cases:
• the graphic rendering of the subphonemic glides j and w as in e.g. i-je-re-u, a-re-ku-tu-ru-wo which can be contrasted with later Greek ἱερεύς and ἀλεκτρυών where the glides are not or not routinely indicated;
• and so-called plene writing; this term is used in a variety of meanings. Here, we shall deal with the graphic rendering of consonants in the coda of final syllables where the standard spelling rules would predict their omission, e.g. wa-na-ka rendering /wanaks/ where the spelling rules would predict a rendering wa-na.