An interferometric astrometric mission, aiming at accuracies at around
the 10 microarcsec level, was recommended as a high priority concept
within the new ESA Horizon 2000+ scientific programme. The original ou
tline concept for such a mission, GAIA, presented its general feasibil
ity but did not address many questions of implementation or optimisati
on. Another concept of an interferometer for a scanning astrometric sa
tellite is presented. It contains a simpler optical telescope and a mo
re efficient detector system. The design utilizes the full resolution
of all light in the dispersed fringes of a Fizeau interferometer. A pr
eliminary optimization of the satellite indicates that two telescope u
nits with a baseline of 100 cm will achieve a precision of 3, 8, 22, 6
8, 302 microarcsec for parallaxes of stars with V = 12, 14, 16, 18, 20
mag, respectively, from a 5 year mission. Simultaneous spectrophotome
try of the entire spectrum of each star will be obtained with a resolu
tion corresponding to intermediate band photometry. The expected preci
sion of this photometry is about 0.003 mag for V = 16. The performance
is good in crowded fields, at least up to one star per 5 arcsec(2). A
Hipparcostype beam combiner of 150 cm width is placed in front of a t
elescope with 4 square apertures of 50 cm. The assumed focal length is
f = 60 m and the field 0.5 degree diameter. The detector consists of
CCDs used for time delayed integration (drift-scan.).