Oops - I missed the following text in the description of implementation
The characters in the subset collections COMBINING DIACRITICAL MARKS
(0300 to 036F), COMBINING DIACRITICAL MARKS FOR SYMBOLS (20D0 to
20FF), HANGUL JAMO (1100 to 11FF) and COMBINING HALF MARKS (FE20 to
FE2F) are not allowed in implementation level 2.
So the difference between Level 1 and Level 2 is that in Level 2
you get the vowel and point marks required for various languages,
but still can't use the diacritical marks for Western languages.