Dick,

I've compared some of the PDFs you sent
me (evolved from the set called S0 in
the MSU paper) to the PDFs I get using the
CTEQ5 evolution program.  I've compared the
Uquark distributions and the Gluon distributions.

The numbers agree pretty well, but there are
noticeable differences. The largest differences
occur at small $x$ and intermediate $Q^2$.
For example, at $x=0.000161$ and $Q^2=16$ GeV2,
the Uq and Gl differ by -12% and -10% respectively.
(Your PDFs are lower.)

I've put a postscript figure on a web page that you
should be able to access. The location is
	www.pa.msu.edu/people/stump/roberts/
There are two files there:
compare.ps  -- postscript
compare.bps -- bitmapped postscript, in case you
  have any problem with the fonts for the other figure.
These are plots of $x*Uq(x)$ and $x*Gl(x)\10$
versus $x$ for 4 values of $Q^2$.
The blue curve is the CTEQ5 evolution;
the black dots are the values you sent me.

Do the differences at small $x$ make any difference?
In other words, is there significant data that is
sensitive to these PDFs?
I don't know how to answer this question except perhaps
by comparing the chi-squares for, say, the H1 data with
the two evolution programs.

Best regards,
Dan Stump