From: John Philo <jphilo@mailway.com>
  To  : Rasmb <rasmb@alpha.bbri.org>
  Date: Mon, 29 Nov 1999 10:30:42 -0800

SEDNTERP bug and update

Hello all,

Unfortunately I recently uncovered a bug in SEDNTERP which has been present
all along but went undetected by Tom Laue, David Hayes and myself as well as
the beta testers. The good news is that the errors are small compared to the
accuracy of the calculated and experimental values.

There has been an error in calculating the density of pure water as a
function of temperature. This error propagates into many other things:
calculated buffer densities, conversions of raw s to s(20,w), sigma to M,
etc.

The error was uncovered because of small discrepancies (few parts in 10000)
between s(20,w) values calculated internally by my newer versions of DCDT+
and SVEDBERG, and those done by SEDNTERP. John Steckert at Genetics
Institute had noticed these differences, which caused me to look at them
more carefully and uncover the SEDNTERP bug.

The source of the error was a constant representing the density of water at
20 degrees C, which was erroneously defined as .998324 g/ml instead of
.998234.

It appears that the most serious consequence is up to ~0.04% errors in
buffer densities calculated based on composition, still within the accuracy
of most such calculations. Such an error would translate into about an 0.11%
error in true mass.

So thankfully the consequences are minimal for real experiments, but the
errors are obviously unacceptable and I apologize of behalf of Tom, David,
and myself for missing this.

A new release of SEDNTERP (1.02) is now available to fix this problem. It
also corrects a bug in s(20,w) conversions that occurred only with certain
entry sequences when the temperature for the velocity experiment differs
from that for the equilibrium experiment (this bug was found by David
Schiffmann), and a bug in printing sample composition reports for
non-integer sample compositions (e.g. half disulfides), found by Jie Wen.

The new release also takes care of our European friends who like to use a
"," to indicate decimal points. It will obey the preferences set via Control
Panel.

To upgrade you can either download a self-extracting archive file containing
only the altered components (SDNUPDAT.EXE, ~220 KB) or download and
re-install the entire package (SDTR9901.exe, 1.75 MB).

The former method is a much smaller download, but there is no full installer
program and you must manually supply the location of the folder where
SEDNTERP was installed (usually c:\sednterp).

If you do a full re-install, you must be careful not to overwrite your own
databases (make a backup of SEDNTERP.MDB and PHYCONST.MDB just in case, and
answer YES that you want to preserve your existing databases when you get
the prompts).

Once you have upgraded you will want to re-calculate the densities for any
buffer records you have previously stored in your database (obviously not
needed if you have experimental densities).

Here is how to do that:
1.) Retrieve the next buffer record via the 'Buffer Data Select' drop-down
list box in the lower panel on the main page.
2.) If the green bar appears under the 'Compute' button then press this
button to force a recomputation; otherwise go back to step 1.
3.) This will bring up the buffer composition form with a new value of the
density at 20C.
4.) Press the 'Save to Buffer Database' button at the lower right, which
will bring up the Saving Data in Experimental Database form.
5.) The name for the buffer will already appear, so press the 'Save New
Record' button (lower right) and answer 'YES' that you wish to overwrite the
existing record.
6.) Back to step 1 until you have updated all that require it.

John Philo