The accuracy of the input coordinates, and of the pointing, have a large impact on the instrumental throughput, of the point source flux accuracy but only a minor impact on the wavelength accuracy for point sources. ISO's random pointing jitter (See section 3.3 of the ISO Satellite Data Users Manual, [Leech, K. et al.\1997b], for a discussion of pointing and improvements through the mission) can also introduce effects in the data if the target coordinates are incorrect or if the target is in a crowded field. While any wavelength error is of the order of the quoted accuracy, the flux error due to the pointing may well limit the final flux accuracy.
As an example, a star offset from the center of the aperture by 6'' in the
cross-dispersion direction (y-axis) loses approximately 40% in throughput
(example taken in the LW, 17 
m ). An offset in the dispersion direction
(z-axis) affects the wavelength calibration, with a 4'' mispointing in the
dispersion direction corresponding to about a 1/8 grating resolution element.
The maximum possible offset is therefore about 1/4 of a grating spectral
resolution element.
Position error problems may introduce what appears to be `noise' in data. The noise (in SPD or AAR) can be much higher than for other objects with similar flux levels. In the AAR this effect manifests itself as sudden flux jumps between adjacent wavelength regions and the up and down scans having different flux levels. Both these effects can be seen in figure 5.
Figure 5: Effect of pointing jitter as seen in AAR. Top plot is
the down scan (up in wavelength), bottom plot is the up scan. The difference is
due to pointing jitter.
Currently this effect is understood to be a function of incorrect positions
being given for the observation (or the object studied being in a crowded
field), the instrumental beam profile and the pointing jitter of ISO. At the
short wavelengths the instrumental beam profile has a triangular shape in the
y direction and a top-hat shape in the z direction. A preliminary reduction of
the beam profile data for band 1D (3.1 m ) shows that the beam profile goes
from 100% to 10% smoothly going from the centre to 
15'' away in
the y-direction. In the z-direction however, it goes from 90% to 10% going
from 3 to 9'' away - the beam profile is very steep in this direction.
If an object is misaligned from the slit centre in the y-direction then as the pointing jitters the object will move up and down a slowly varying response function. Hence the incident flux seen by the detectors will vary smoothly by small amounts. This will probably not be seen in the data.
If, however, an object is misaligned from the slit centre in the z-direction, slight shifts in the spacecrafts pointing will cause the object to be moved up and down the very steep beam profile. As an example, a misalignment of 5'' in the z-direction in band 1D will reduce the incident flux on the detectors by 50%. The 1 arcsec jitter will then move it from the 70% contour to the 30% one - the incident flux can change by more than a factor of 2 on short time intervals.
This also explains the increased noise. Each of the detectors 24Hz readouts (contained in the ERD) are from (slightly) different incident flux levels. This introduces extra noise into the system when the pipeline tries to fit a slope to the 24Hz readout data.
So, this effect, primarily noticeable at short wavelengths, can happen if: the
entered pointing is incorrect by more than 4 arcseconds; or there is
a nearby bright object, which can happen in crowded fields.
We urge users to examine all entered pointings to ensure no errors have been
made when entering coordinates, to go to catalogues to get the latest
coordinates and to ensure there are no bright objects within 10''
of the intended target.