From: Richard Thomas (Luisi) <rthomas@ifp.mat.ethz.ch>
  To  : rasmb@bbri.eri.harvard.edu
  Date: Wed, 26 Jul 95 11:33:37 +0200

Re: Baselines and externally-filling cells.

In the light of the recent correspondence on baselines etc (and in partcular point 3 of Jack
C.'s latest), I thought it might be a good idea to bring up again the subject of baselines in
the XLA.
On our machine (and having tested all cells available) I have found that there are always
baseline contributions. These are at their worst in 6 channel equilibrium cells where the
effects vary from sector to sector. The offsets are both wavelength and speed-dependent
but appear to be reasonably reproducible, and while not linear can be reasonably
approximated as so. These effects are not in themselves startlingly large (0.01-0.02A),
however they can obviously have a disasterous effect when working at low loading
concentration or in the upper part of the cell at high speeds (introducing errors of <10%).
As we are largely working with small peptides I always (a) use standard double sectors as
I am without externally-filling cells and (b) always run a baseline after the equilibrium
experiment.
Floating the baseline in fitting procedures produces worse statistics than setting it to the
known value and, in any case, it seems intrinsically more reasonable to subtract a 'real'
baseline when one exists.
Any ideas?
All the best
Richard