-4- 4xii50 (ix) Comparing the length-frequency of HT sign--groups with the frequencies that seem to result from a count of B sign-groups (admittedly liable to considerable error) we find the following percentage distribution: A B 2-letter groups ˇˇˇˇˇˇˇˇˇˇˇˇˇˇ 33 % 16 % 3 49 38 4 17 39 5 1 6 6 ˝ 7 ˝ Average word-length: 2.86 3.36 If we assume that A and B represent parts of a strictly equivalent "vocabulary", the A sign-groups must be abbreviated by an average of ˝ a sign per word, on something like the following scheme: Words, actual Shortened Left Words, written 2-letterˇˇˇˇˇˇˇ 16 % 0 % 16 % 33 % 3 ˇˇˇˇˇˇˇˇˇˇˇˇˇˇ38 17 21 49 4 ˇˇˇˇˇˇˇˇˇˇˇˇˇˇ39 28 11 17 5 ˇˇˇˇˇˇˇˇˇˇˇˇˇˇ 6 6 0 1 6 ˇˇˇˇˇˇˇˇˇˇˇˇˇˇ ˝ ˝ 0 0 7 ˇˇˇˇˇˇˇˇˇˇˇˇˇˇ ˝ ˝ 0 0 This is not necessarily a fact, but if it was so it would have the effect of reducing the probable number of apparent correspondences re- sulting from a finite vocabulary to something little more than ˝ the normal. The resulting figures are entered in red in the table overleaf. There is still a recalcitrant discrepancy of about 50%, however, which seems to make the further assumption necessary that the HT and Knossos/Pylos vocabularies are in some significant way not comparable, that they deal, to a considerable extent, with different categories - with places or gods instead of people, or with words expressed with consistently different case-endings. Which is another way of saying, in the end, that I agree with you that the facts are odd and still need explanation. --------------------------------------------------------------------------- INFLICTION IN THE PYLOS INSCRIPTIONS. (x) Remembering your remarks at lunch the other day, I was very interes- ted to confirm for my own satisfaction the apparent inflections occur- ring in the Pylos tablets published in your AJA article. They look like a hopeful source of additions to what I call the 'GRID', that is to say, a table of signs arranged in columns down and across according to the evidence for shared vowels and shared consonants respectively. In its very rough state, my grid has 3 separate pages: 1) Cases where varying inflections, applied to more than one form of the same 'root', apparently cause a vowel-change in the syllable immediately preceding. 2) Cases where different inflections do not change the preceding syllable in pairs of words. 3) Cases of inflections apparently falling within the same series, but for which the evidence for more than one form of root isn't available. The columns are subdivided under: 1) Sign-group used as evidence. 2) Reference. 3) Form of the syllable preceding each apparent inflexion, including -_, -_, -_, -_, -_ -_, -_ -_, -_, -_ -_, -_, -_, -_, -_, -_, -_, -_-_ . There are probably others worth including which I haven't struck myself. My reason for ignoring - _ I'll explain below. I imagine that you agree that the following items out of Kn02 etc rate inclusion in the grid (in fact, I think it was one of them which you pointed out to me):