Hi All, I have, thanks to a long plane ride, come up with what I hope is a clear exposition of the point that I was trying to make earlier this week. We have three relationships among 4 sets of code points. The code points are: - Latin Small Letters A and E - Latin Small Ligature Æ - Latin Small Ligature Œ - Latin Small Letters O and E There are potential variant relationships between each adjacent pair. Clearly these cannot all be made variant pairs, as transitivity would then require two ASCII letters (A and E) to be variants, which is not allowed. So the question becomes, when we look at the various pairs, which is the strongest case for variant and which is the weakest? If we look at the ligature Æ and the ligature Œ in one of the fonts where A is rendered as a we see æ. Comparing this to œ we see that visually they are at least a 2, and quite possible a 1. Which means this pair does need to be included as an in-script variant. In contrast, a comparison between AE and Æ (ae vs æ) is at best a 3 and perhaps even a 4. Likewise with OE and Œ (oe vs œ). But the resemblance is not, in my opinion, equal. Because of the vertical line on the right-hand side of the small letter A, there is about half as much white space at the top and bottom of the non-ligature A and E as in the case of O and E. Which to me says that, since we must drop one of these from consideration as a variant, OE and ligature Œ is the relationship which gets downgraded to Confusable. I hope that clarifies where I was coming from. (Certainly it is clearer than just saying "gut feel". Which, alas, was all I could come up with on the fly for an explanation.) Bill Jouris Inside Products bill.jouris@insidethestack.com 831-659-8360 925-855-9512 (direct)