From: Dr A.J. Rowe <ajr@leicester.ac.uk>
  To  : rasmb@bbri.harvard.edu
  Date: Tue, 9 Sep 1997 15:14:56 +0100 (BST)

Gaussian fits in proFit - warning

Hello RASMBERs !

No doubts quite a few of you do Gaussian fits from time to time (or even for
lots of the time) and some of you may (like us) use the package proFit. It is in
general a very good package, and we have made lots of use of it for data
analysis, and indeed all the interpretative routines in our AUC Imager are
proFit based.

However, we have found some results obtained very hard to believe, when using
the built it Gaussian function. The reason for this can be discerned - as we now
realise - by looking at the non-standard form of the basic equation given in the
parameter window. Not a job for the visually challenged, it must be said.

Leaving aside their re-parameterisation which means that symbol A now stands for
a *normalised* peak height parameter, NOT the area (it does tell you that - but
why stick with A?) the more serious problem is that notwithstanding an explicit
statement in the parameter window that *SD* is the *standard deviation* (defined
in the conventional sense in the manual), *SD* is in fact the VARIANCE of the
curve (= SD^2 for the less statistically minded). The software company accept
that there is a mistake, and no doubt future versions will have this one sorted.

In the meantime, if any of you have (like us) used this function for any purpose
- maybe fiting g(s*) profiles or whatever - then you need to check back all your
notebooks !

Of course, it is not difficult to write your own function in, with the
parameters defined in the conventional manner. Once you know you have to.

Arthur Rowe

***************************************************
Dr Arthur J Rowe
Director
UK National Centre for Macromolecular Hydrodynamics
Leicester Laboratory
Adrian Building
University of Leicester
Leicester LE1 7RH    UK

Tel: +44 (0)116 252 3448
Fax: +44 (0)116 252 5602
ajr@leicester.ac.uk
***************************************************