> MY understanding of UNICODE_FSS was that it is a 'wide' string format,
> and that 24bits per character were used, so 3 bytes.

No, one character is 1 or 2 or 3 bytes.
It is the reason why it is not a good idea to use CHAR datatype with Unicode_fss,
because CHAR by definition uses fixed buffer length, e.g. if you have
CHAR(2) CHARACTER SET UNICODE_FSS
you can store 'ABCDEF' into it because each ASCII char is one byte only.

> So when you compare
> strings you do not have to do anything other than a simple byte by byte
> comparison.

In fact you _can_ compare Unicode_fss strings byte by byte
(if what you want is binary collation).

Ivan
http://www.volny.cz/iprenosil/interbase/