This table of line fluxes is generated from LWS Auto-Analysis-Pipeline
7.0 products using Sangeeta's automatic reduction routines.

The Auto-Analysis-Pipeline 7.0 does dark current subtraction and
absolute gain correction on the measured LWS fluxes. It does not
correct for a responsivity drift. At this point this is not found to
be a serious omission. The biggest uncertainty in the raw measurements
is due to variable dark currents. This affects the measurement of the
continuum levels to upto a factor of few. So the *CONTINUUM VALUES ARE
NOT TO BE TRUSTED AT THIS POINT*. The line fluxes are not hugely
affected by the uncertainties in the continuum; the uncertainties in
the line fluxes are dominated by the absolute gain corrections, and
scatter therein and are of the order 30-percent.

The procedure used to determine line fluxes is as follows.  We first
combine all scans for each line by medianing with rejection. We take
all the non-flagged measurements and do a 5-sigma rejection on the
flux points on all scans. Four measurement points immediately
following each rejected point are also thrown out because of possible
memory effects. Then a final profile is created by running a median
filter of width 0.03 microns. The program then looks at the known
redshift/velocity of the galaxy and fits a linear baseline to the
continuum, excluding the points falling within the line profile.  The
width of the line profile is the larger of the instrumental profile or
Delta-V=1500 km/s. Total line flux is determined by integrating under
the line-profile after a baseline subtraction. The rms of the
measurement is determined from the baseline.

There were some doubts expressed that the pipeline products may not
have the dark currents properly removed. Sangeeta and Jim Brauher
looked at and corrected for the dark currents and absolute
responsivity for all the galaxies individually using ISAP utilities.
Using these "hand-corrected" data and running the above procedure for
rejection and medianing we get line fluxes which are consistent with
the pipeline products to 4 percent (1-sigma scatter), with extreme
discrepancy upto 15 percent for lines that are detected at 5-sigma or
better. 

In the next stage of data-reduction is planned to have the combining
of individual data scans and rejection of bad scan/points done by
hand. That is expected to be the most time consuming step of all, we
(the royal we, i.e. Sangeeta) do not expect the line fluxes reported
here to change substantially. We hope that some of the fainter lines
will be detected.

questions? comments? Send email to san@ipac.caltech.edu