In summary, the results above presented show an overall agreement which
exceeds in most cases pre-launch expectations. This applies
both to the internal (among the ISO
instruments) and external (between ISO instruments and other space missions,
like IRAS, MSX or COBE-DIRBE) comparisons made. The consistency found
is in general of the order or better than the 10-15% level for what we
have called `well-behaved' data sets, with just a few
exceptions. In particular the discrepancies between IRAS photometry
and ISO synthetic photometry at 60 and 100 
m are shown
to be probably
related to a problem in the absolute flux calibration of IRAS at
high flux levels.
ISO, along with several other missions, was so-well calibrated because they were all underpinned by a common, self-consistent, calibration basis that was already in place at the time of launch.
The clear lesson learned from this success is that one needs a self-consistent common basis as much to bind together the instruments on a given satellite, as to compare transparently with data from different spaceborne, ground-based and airborne facilites. Building on a solid foundation of work from the ground, airplanes and satellites it has been possible to establish a robust infrared calibration framework.
The ISO Ground Based Preparatory Programme (GBPP) was able to deliver a uniform calibration database (photometric data plus spectral energy distributions) of some 400 sources covering the whole sky with an accuracy typically of a few percent which was used mainly for the calibration of the ISO instruments at short wavelengths (Hammersley & Jourdain de Muizon 2001, [73]).
Founded originally on empirical spectra derived from
observations on NASA's Kuiper Airborne Observatory (KAO), the entire
calibration context is traceable to emergent spectra from a pair of
stellar atmospheric models for Vega and Sirius by Kurucz. The scheme
proved very successful in unifying
infrared photometry and spectroscopy in the 1-35 m range for
a variety of instruments onboard KAO, IRTS, ISO and MSX
and was successfully extended to 300 m in support of ISOPHOT
(Cohen 2001, [34]). Although stars
were too faint to calibrate LWS directly, its spectra of several K-giant
calibration stars show flux levels and spectral energy distributions to be
perfectly consistent (albeit with absolute uncertainties of order 15-20%)
with LWS formal calibration from Uranus.
Many connections to IRAS, KAO and between ISO sub-instruments have been established via observations of individual asteroids, which have also provided means to check the absolute photometric calibration and validate the RSRF of the ISO instruments in critical areas, e.g. in the case of late type stellar standards at the CO and SiO fundamental absorption bands (Müller & Lagerros 2001, [126]).
In the near future, this
all-sky network of absolute calibrators is expected to be expanded to
a far fainter network
suitable not only for SIRTF but also for ground-based 8-10 m class
telescopes, still self-consistently.