To look further into this problem I opened up a saved PDF file in a text
editor and confirmed that the kerning values between a given character pair
are not constant. This is after I factored out character spacing, horizontal
scaling effects, and kerning pair data from the font metrics file. My usage
of framemaker is highly sensitive to very small changes in these values so I
need to know why the distance between a given character pair should change
from one line of my document to the next. I conducted my experiments using
Times-New Roman 12 point and the word 'encoding'. I noticed that if I create
a paragraph with many instances of this word and then I save that file as
PDF, the distances between any two characters as given in the operand of the
TJ operator in the PDF stream will vary quite a bit from one instance of the
word to the next. In one instance the distance between the 'e' character and
the 'n' character is -4 while in the next instance of the word this value is
+16. I can see absolutely no typographic explanation for such variation.
Further, when I run statistics on a large file containing many instances of
a given pair of characters, the distribution of values appears completely
unpredictable. Again, to be perfectly clear, I have removed any possible
effect of kerning pair values, justification, font, font size, horizontal
scaling or character spacing. I would very much appreciate a good
explanation for why this variation is happening and what type of algorithms
may be being applied by Framemaker to come up with the character spacing
values seen in the PDF file.
thank you,
Peter Lepeska
ICT, Inc.
617.773.3369
X205