whilst using Font2DTest to test some changes I noticed that on my
Solaris 7 system if you select the "dialog" font and "dingbats" unicode
range some of the dingbats are off by 1 from their unicode code points.
Look at code points 0x2776 -> 0x2795.
        
Its much easier to spot now that we have Lucida in the composite
font in merlin as we actually have TWO adjacent copies of the serif 1
on a black circular b/g because the Lucida one is going in at 0x2776
(where it should be) since dingbats is saying it doesn't have that code point
since it wants to display it at code point 0x2777
        
You can compare with Lucida Sans as well as the unicode 3 manual.
        
It looks like an error in the table in CharToByteX11Dingbats.java
that's been there since day one.