-2- 4xii50 An interesting point is that the little silver disks at Knossos, marked _ and _ , and weighing (so i understand) 7.4 g and 3.7 g respectively, fit into this binary system as 1/32 and 1/64 of _ . These symbols don't seem to fit into the system of true fractions, but they couldn't very well have used _ again for 1/32 All the extant examples of weights wont fit into any one system, and evidently several different systems can reasonably be assumed. I enclose a sheet of the log graph paper with a rough idea of my way of arranging the data. (iv) A guess at the absolute values of the 'liquid' and 'dry' measures is much more precarious, but it seems reasonable to assume that the smallest term _ has something like the value of the Egyptian h e n (470 cc), Greek k o t y l e (280 cc), Biblical l o g (540 cc) or Syrian k o t y l e (350 cc). If we take a lowish value of 300 cc , the value of the unit measure, 96 times _ , equals 28.5 litres, equivalent to the cubic capacity of 1 _ of water (?). Whether such a direct relationship is likely is hard to say, but it's perhaps reasonable to assume that the values of the system are somewhere between the following estimate and one twice as high: 'LIQUID' 'DRY' Fractional unit Metric U.S. Metric U.S. - - 71.2 1 2.02 bushels 28.5 1 7.5 gallons - - _ or _ 7.12 1 1.9 gallons 7.12 1 1.6 gals dry _ 1.19 1 2.5 pints 1.19 1 2.1 pints dry _ 300 cc 0.6 pint 300 cc 0.5 pint dry If one assumes that _ and _ have the same value in both systems, then the two symbols _ and _ must actually be identical, although dif- ferent kinds of actual measuring vessels might be reasonable from this size upwards. Obviously these tables are very conjectural, but might help to give a first rough idea of the amounts of various commodities recorded on the tablets. The only check I have made myself is to do some juggling with the kitchen scales and figure that drinking vessels of thin silver (?) of the types shown on Kn02 would probably weigh about 100 g for small ones about 8-10 cm high, and about 200 g for large ones about 16-20 cm high (in other words, about _ 3 and _ 6 respectively with the values suggested overleaf). This would agree reasonably with the entries on Kn01, if the purpose of the lumps of _ was indeed to make 1 or 2 cups out of each. (v) The use of fractional units at HT compounded with _ , which you mention, sounds like an interesting forerunner of the B use, but my Carratelli is in the attic, and I haven't been able to check up whether the actual ratios appear in any way analogous. I suppose that the Hieroglyphic signs _-_-_-_-_-_ are similarly forerunners of the A fractions, but I haven't looked into them either. RELATIONSHIP BETWEEN LINEAR A AND LINEAR B. (vi) I was rather alarmed by the figures you give in the AJA of only 10 2-letter groups and. 3 3-letter groups identical out of 500 A groups and 2700 B groups; and I sat down to check mathematically the probabi- lity that this number would be identical under various assumptions. First it was necessary to assume some sort of breakdown into words of different lengths; and to estimate what proportion of each length would be left, after one had eliminated groups containing signs peculiar to A and B, and therefore not comparable. Here I've made a pot-shot at the resulting figures, and not having the whole of both lists to work from, they may be badly out.