It was decided at the Cambridge meeting last week to adjust the
NIS calibration based on a reanalysis of the original lab
calibration, and on the comparison with the Woods sounding rocket
data.  This has now been done, and has been put online.  The
following adjustments were made:

NIS-1:  
The wavelength dependence was adjusted to match the latest
results of the lab calibration.  The curve was then adjusted
upwards so that the efficiency at 368A was increased by a factor
of 2.3 over the original calibration curve, based on the
comparison with the Woods rocket.

NIS-2:  
A new calibration curve based on the comparison with the Woods
rocket was installed.



Two databases are affected, efficiency.db* and wave_eff.db*.
Both databases need to be refreshed in order to correctly pick up
the new calibration.

You can still obtain the old calibration curve by passing the
keyword DATE_USED to GET_EFFICIENCY and GET_WAVE_EFF.  Any value
of DATE_USED between 23-Nov-1996 and 23-Dec-1998 will return the
old calibration.  This allows you to calculate a correction
factor to be applied to data analyzed using the old calibration.
For example

       WAV1 = PIX2WAVE('N1', INDGEN(1024))
       GET_EFFICIENCY, date_obs, 'N1', EFF1
       GET_EFFICIENCY, date_obs, 'N1', OLD_EFF1, DATE_USED='1-Jan-1998'
       GET_WAVE_EFF, date_obs, 'N1', COEFF1
       GET_WAVE_EFF, date_obs, 'N1', OLD_COEFF1, DATE_USED='1-Jan-1998'
       CORRECTION = POLY(WAV1, EFF1*COEFF1) / POLY(WAV1, OLD_EFF1*OLD_COEFF1)

To simplify the process, the keyword WAVELENGTH has now been
added to the GET_EFFICIENCY routine, so that GET_WAVE_EFF and
POLY no longer have to be called by the user.  This simplifies
the above to

        WAV1 = PIX2WAVE('N1', INDGEN(1024))
        GET_EFFICIENCY, date_obs, 'N1', EFF1, WAVE=WAV1
        GET_EFFICIENCY, date_obs, 'N1', OLD_EFF1, DATE_USED='1-Jan-1998', $
                WAVE=WAV1
        CORRECTION = EFF1 / OLD_EFF1

Bill Thompson

----------------------------------------
PS

It's been pointed out to me that the business about deriving a
correction factor for already calibrated data is a little
confusing.  It should be pointed out that the correction factor
defined above needs to be *divided* into the data.