From: dr. jack correia <jcorreia@fiona.umsmed.edu>
  To  : rasmb@BBRI.ERI.HARVARD.EDU
  Date: Sun, 8 May 94 16:41:56 -0500

Re: Bo Demeler's Wavelength Suggestion

	The nonflat plateau and baseline problem in the XL-A is a serious
problem when attemting the VanHolde-Weischet method of analysis.  It
exaggerates or creates nonideality effects.  (It is also an obvious problem
with the 2nd moment analysis method.)  Attemting to analyze only the central re
regions of a boundary is not sufficient since the sloping baseline is 
superimposed upon the entire data set.  I pointed this problem out to Bo Demeler
for the VanHolde-Weischet analysis and requested he add the ability to 
subtract out a straight line from the entire data set to his software.  It 
can be derived from a fit of a small region of the plateau but it should also
have an arbitrary selection option.  It must be used with great caution!

	I do not accept the wavelength drift or error suggestion as the 
cause of this artifact!  Yes we often use an absorption shoulder because of 
too little or too much light.  But we also observe this problem when at a
maxima.  Why is the plateau drift also positive with r?  The light source 
does not know where the photomultiplier slits are, does it?

	We observe the problem to be worse with low light intensity!  I do
not understand why?  It gets better after cleaning the light source, slits, etc
but it does not go away!

	It the radial shape of the velocity cell a potential source of this 
artifact?  More stray light?  We have readjusted out lamp position so light
intensity is within Beckman spec (+/- 20% at 280 nm across the cell radius)
and peaked at 6.5 cm.  But this is air vs air!  Check buffered solutions 
as well!  Will interference slits possibly improve the intensity linearity?

	No answers here, just questiopns!

Re: short column centerpieces.

	We did our first short column run! Using a 100,000 MW tubulin dimer
we got 109,000 +/- 10,000.  Did four channels with 2, 1.5, 1, 0.5 mg/ml.  Not
bad for a prtein prone to aggregate.  Used Nonlin and a joint fit!  We are 
using different delay values for each channel (180-158, channel A-D).  We scan
each empty channel as a function of delay and plot ref and sample intensity
from the central region vs. delay value.  The curves do not peak at the same
delay value, so we use the crossover point to maximize signal from both ref
and sample!  We are testing each centerpiece to be sure they are all the same!