The primary objective of the GAIA mission is to observe the physical characteristics, kinematics and distribution of stars over a large fraction of the volume of our Galaxy, with the goal of achieving a full understanding of its dynamics and structure, and consequently its formation and history (see, e.g., Gilmore et al. 1989; Majewski 1993; Ibata et al. 1997; Wyse et al. 1997; de Zeeuw 1999; as well as extensive details of the scientific case given in ESA 2000). An overview of the main Galaxy components and sub-populations is given in Table 1, together with requirements on astrometric accuracy and limiting magnitude.

Table 1: Some Galactic kinematic tracers, with corresponding limiting magnitudes and required astrometric accuracy. For various tracers (Col. 1), Cols. 2-6 indicate relevant values of parameters leading to the typical range of V magnitudes over which the populations must be sampled (Cols. 7-8). These results demonstrate that the faint magnitude limit of GAIA is essential for probing the different Galaxy populations, while the astrometry accuracies in Cols. 9-12 demonstrate that GAIA will meet the scientific goals (based on Gilmore & Høg 1995)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

Tracer M_V $\ell$ b d A_V V₁ V₂ $\epsilon_{\rm T}$ $\sigma_{\mu_1}$ $\sigma^{\prime}_{\mu_1}$ $\sigma^{\prime}_{\pi_1}$

mag deg deg kpc mag mag mag km s^-1 $\mu$ as/yr - -

Bulge:

gM -1 0 <20 8 2-10 15 20 100 10 0.01 0.10

HB +0.5 0 <20 8 2-10 17 20 100 20 0.01 0.20

MS turnoff +4.5 1 -4 8 0-2 19 21 100 60 0.02 0.6

Spiral arms:

Cepheids -4 all <10 10 3-7 14 18 7 5 0.03 0.06

B-M Supergiants -5 all <10 10 3-7 13 17 7 4 0.03 0.05

Perseus arm (B) -2 140 <10 2 2-6 12 16 10 3 0.01 0.01

Thin disk:

gK -1 0 <15 8 1-5 14 18 40 6 0.01 0.06

gK -1 180 <15 10 1-5 15 19 10 8 0.04 0.10

Disk warp (gM) -1 all <20 10 1-5 15 19 10 8 0.04 0.10

Disk asymmetry (gM) -1 all <20 20 1-5 16 20 10 15 0.14 0.4

Thick disk:

Miras, gK -1 0 <30 8 2 15 19 50 10 0.01 0.10

HB +0.5 0 <30 8 2 15 19 50 20 0.02 0.20

Miras, gK -1 180 <30 20 2 15 21 30 25 0.08 0.65

HB +0.5 180 <30 20 2 15 19 30 60 0.20 1.5

Halo:

gG -1 all <20 8 2-3 13 21 100 10 0.01 0.10

HB +0.5 all >20 30 0 13 21 100 35 0.05 1.4

Gravity, $\cal {K}_Z$ :

dK +7-8 all all 2 0 12 20 20 60 0.01 0.16

dF8-dG2 +5-6 all all 2 0 12 20 20 20 0.01 0.05

Globular clusters (gK) +1 all all 50 0 12 21 100 10 0.01 0.10

internal kinematics (gK) +1 all all 8 0 13 17 15 10 0.02 0.10

Satellite orbits (gM) -1 all all 100 0 13 20 100 60 0.3 8

**Table 1:** Some Galactic kinematic tracers, with corresponding limiting magnitudes and required astrometric accuracy. For various tracers (Col. 1), Cols. 2-6 indicate relevant values of parameters leading to the typical range of V magnitudes over which the populations must be sampled (Cols. 7-8). These results demonstrate that the faint magnitude limit of GAIA is essential for probing the different Galaxy populations, while the astrometry accuracies in Cols. 9-12 demonstrate that GAIA will meet the scientific goals (based on Gilmore & Høg 1995)
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
Tracer	M_V	$\ell$	b	d	A_V	V₁	V₂	$\epsilon_{\rm T}$	$\sigma_{\mu_1}$	$\sigma^{\prime}_{\mu_1}$	$\sigma^{\prime}_{\pi_1}$
	mag	deg	deg	kpc	mag	mag	mag	km s^-1	$\mu$ as/yr	-	-
Bulge:
gM	-1	0	<20	8	2-10	15	20	100	10	0.01	0.10
HB	+0.5	0	<20	8	2-10	17	20	100	20	0.01	0.20
MS turnoff	+4.5	1	-4	8	0-2	19	21	100	60	0.02	0.6
Spiral arms:
Cepheids	-4	all	<10	10	3-7	14	18	7	5	0.03	0.06
B-M Supergiants	-5	all	<10	10	3-7	13	17	7	4	0.03	0.05
Perseus arm (B)	-2	140	<10	2	2-6	12	16	10	3	0.01	0.01
Thin disk:
gK	-1	0	<15	8	1-5	14	18	40	6	0.01	0.06
gK	-1	180	<15	10	1-5	15	19	10	8	0.04	0.10
Disk warp (gM)	-1	all	<20	10	1-5	15	19	10	8	0.04	0.10
Disk asymmetry (gM)	-1	all	<20	20	1-5	16	20	10	15	0.14	0.4
Thick disk:
Miras, gK	-1	0	<30	8	2	15	19	50	10	0.01	0.10
HB	+0.5	0	<30	8	2	15	19	50	20	0.02	0.20
Miras, gK	-1	180	<30	20	2	15	21	30	25	0.08	0.65
HB	+0.5	180	<30	20	2	15	19	30	60	0.20	1.5
Halo:
gG	-1	all	<20	8	2-3	13	21	100	10	0.01	0.10
HB	+0.5	all	>20	30	0	13	21	100	35	0.05	1.4
Gravity, $\cal {K}_Z$ :
dK	+7-8	all	all	2	0	12	20	20	60	0.01	0.16
dF8-dG2	+5-6	all	all	2	0	12	20	20	20	0.01	0.05
Globular clusters (gK)	+1	all	all	50	0	12	21	100	10	0.01	0.10
internal kinematics (gK)	+1	all	all	8	0	13	17	15	10	0.02	0.10
Satellite orbits (gM)	-1	all	all	100	0	13	20	100	60	0.3	8

2.2 The star formation history of our Galaxy

A central element the GAIA mission is the determination of the star formation histories, as described by the temporal evolution of the star formation rate, and the cumulative numbers of stars formed, of the bulge, inner disk, Solar neighbourhood, outer disk and halo of our Galaxy (e.g. Hernandez et al. 2000). Given such information, together with the kinematic information from GAIA, and complementary chemical abundance information, again primarily from GAIA, the full evolutionary history of the Galaxy is determinable (e.g. Freeman 1993; Gilmore 1999).

Determination of the relative rates of formation of the stellar populations in a large spiral, typical of those galaxies which dominate the luminosity in the Universe, will provide for the first time quantitative tests of galaxy formation models. Do large galaxies form from accumulation of many smaller systems which have already initiated star formation? Does star formation begin in a gravitational potential well in which much of the gas is already accumulated? Does the bulge pre-date, postdate, or is it contemporaneous with, the halo and inner disk? Is the thick disk a mix of the early disk and a later major merger? Is there a radial age gradient in the older stars? Is the history of star formation relatively smooth, or highly episodic? Answers to such questions will also provide a template for analysis of data on unresolved stellar systems, where similar data cannot be obtained.

2.3 Stellar astrophysics

GAIA will provide distances of unprecedented accuracy for all types of stars of all stellar populations, even those in the most rapid evolutionary phases which are very sparsely represented in the Solar neighbourhood. All parts of the Hertzsprung-Russell diagram will be comprehensively calibrated, from pre-main sequence stars to white dwarfs and all transient phases; all possible masses, from brown dwarfs to the most massive O stars; all types of variable stars; all possible types of binary systems down to brown dwarf and planetary systems; all standard distance indicators, etc. This extensive amount of data of extreme accuracy will stimulate a revolution in the exploration of stellar and Galactic formation and evolution, and the determination of the cosmic distance scale (cf. Lebreton 2000).

2.4 Variability

2.5 Binaries and multiple stars

A key scientific issue regarding double and multiple star formation is the distribution of mass-ratios q. For wide pairs (>0.5 arcsec) this is indirectly given through the distribution of magnitude differences. GAIA will provide a photometric determination of the q-distribution down to $q\sim0.1$ , covering the expected maximum around $q\sim0.2$ . Furthermore, the large numbers of ("5-year'') astrometric orbits, will allow derivation of the important statistics of the very smallest (brown dwarf) masses as well as the detailed distribution of orbital eccentricities (Söderhjelm 1999).

GAIA is extremely sensitive to non-linear proper motions. A large fraction of all astrometric binaries with periods from 0.03-30 years will be immediately recognized by their poor fit to a standard single-star model. Most will be unresolved, with very unequal mass-ratios and/or magnitudes, but in many cases a photocentre orbit can be determined. For this period range, the absolute and relative binary frequency can be established, with the important possibility of exploring variations with age and place of formation in the Galaxy. Some 10 million binaries closer than 250 pc should be detected, with very much larger numbers still detectable out to 1 kpc and beyond.

2.6 Brown dwarfs and planetary systems

Sub-stellar companions can be divided in two classes: brown dwarfs and planets. There exist three major genesis indicators that can help classify sub-stellar objects as either brown dwarfs or planets: mass, shape and alignment of the orbit, and composition and thermal structure of the atmosphere. Mass alone is not decisive. The ability to simultaneously and systematically determine planetary frequency and distribution of orbital parameters for the stellar mix in the Solar neighbourhood is a fundamental contribution that GAIA will uniquely provide. Any changes in planetary frequency with age or metallicity will come from observations of stars of all ages.

An isolated brown dwarf is typically visible only at ages <1 Gyr because of their rapidly fading luminosity with time. However, in a binary system, the mass is conserved, and the gravitational effects on a main-sequence secondary remain observable over much longer intervals. GAIA will have the power to investigate the mass-distribution of brown-dwarf binaries with 1-30 year periods, of all ages, through analysis of the astrometric orbits.

There are a number of techniques which in principle allow the detection of extra-Solar planetary systems: these include pulsar timing, radial velocity measurements, astrometric techniques, transit measurements, microlensing, and direct methods based on high-angular resolution interferometric imaging. A better understanding of the conditions under which planetary systems form and of their general properties requires sensitivity to low mass planets (down to $\sim$ $10~M_\oplus$ ), characterization of known systems (mass, and orbital elements), and complete samples of planets, with useful upper limits on Jupiter-mass planets out to several AU from the central star (Marcy & Butler 1998; Perryman 2000).

Astrometric measurements good to 2-10 $\mu$ as will contribute substantially to these goals, and will complement the ongoing radial velocity measurement programmes. Although SIM will be able to study in detail targets detected by other methods, including microlensing, GAIA's strength will be its discovery potential, following from the astrometric monitoring of all of the several hundred thousand bright stars out to distances of $\sim$ 200 pc (Lattanzi et al. 2000).

2.7 Solar system

Solar System objects present a challenge to GAIA because of their significant proper motions, but they promise a rich scientific reward. The minor bodies provide a record of the conditions in the proto-Solar nebula, and their properties therefore shed light on the formation of planetary systems.

The relatively small bodies located in the main asteroid belt between Mars and Jupiter should have experienced limited thermal evolution since the early epochs of planetary accretion. Due to the radial extent of the main belt, minor planets provide important information about the gradient of mineralogical composition of the early planetesimals as a function of heliocentric distance. It is therefore important for any study of the origin and evolution of the Solar system to investigate the main physical properties of asteroids including masses, densities, sizes, shapes, and taxonomic classes, all as a function of location in the main belt and in the Trojan clouds.

The possibility of determining asteroid masses relies on the capability of measuring the tiny gravitational perturbations that asteroids experience in case of a mutual close approach. At present only about 10 asteroid masses are known, mostly with quite poor accuracy. Asteroid-asteroid encounters have been modelled, and show that GAIA will allow more than 100 asteroid masses to be determined accurately.

Albedo is a useful complement to spectrophotometric data for the definition of different taxonomic classes. The GAIA photometry will be much more reliable than most data presently available. The colour indices will provide a taxonomic classification for the whole sample of observed asteroids.

For direct orbit determinations of known asteroids, preliminary simulations have been performed in which the covariance matrices of the orbital elements of more than 6000 asteroids were computed using both the whole set of astrometric observations collected from ground-based telescopes since 1895 through 1995, as well as a set of simulated observations carried out by GAIA, computed by considering a 5 year lifetime of the mission, and present instrument performances. Another set of simulated ground-based observations covering the period 1996-2015 were also performed. For the known asteroids the predicted ephemeris errors based on the GAIA observations alone 100 years after the end of the mission are more than a factor 30 better than the predicted ephemeris errors corresponding to the whole set of past and future ground-based observations. In other words, after the collection of the GAIA data, all the results of more than one century of ground-based asteroid astrometry will be largely superseded.

In addition to known asteroids, GAIA will discover a very large number, of the order of 10⁵ or 10⁶ new objects, depending on the uncertainties on the extrapolations of the known population. It should be possible to derive precise orbits for many of the newly discovered objects, since each of them will be observed many times during the mission lifetime. These will include a large number of near-Earth asteroids. The combination of on-board detection, faint limiting magnitude, observations at small Sun-aspect angles, high accuracy in the instantaneous angular velocity (0.25 mas s^-1), and confirmation from successive field transits, means that GAIA will provide a detailed census of Atens, Apollos and Amors, extending as close as 0.5 AU to the Sun, and down to diameters of about 260-590 m at 1 AU, depending on albedo and observational geometry.

2.8 Galaxies, quasars, and the reference frame

GAIA will not only provide a representative census of the stars throughout the Galaxy, but it will also make unique contributions to extragalactic astronomy (Table 2). These include the structure, dynamics and stellar populations in the Local Group, especially the Magellanic Clouds, M31 and M33, the space motions of Local Group galaxies, a multi-colour survey of galaxies (Vaccari 2000), and studies of supernovae (Høg et al. 1999b), galactic nuclei, and quasars.

Table 2: Local Group galaxies potentially accessible to GAIA. E(B-V) indicates the foreground reddening, and (m-M)₀ is the true distance modulus. $V_{\rm lim}$ is the brightest star in the galaxy. $\mu _{v_{\rm t}-v_{\rm r}}$ is the estimated proper motion, assuming the transverse velocity equals the observed radial velocity. * denotes observed values
Galaxy l b E(B-V) (m-M)₀ Distance $V_{\rm lim}$ N(stars) V_r $\mu _{v_{\rm t}-v_{\rm r}}$

( $^\circ$ ) ( $^\circ$ ) (mag) (mag) (kpc) (mag) (V<20) (helio) ( $\mu$ as/yr)

WLM 75.9 -73.6 $0.02\pm0.01$ $24.83\pm0.08$ $925\pm40$ 16.5 $\sim$ 500 -116 26

NGC 55 332.7 -75.7 $0.03\pm0.02$ $25.85\pm0.20$ $1480\pm150$ 15.0 10's 129 18

IC 10 119.0 -3.3 $0.87\pm0.12$ $24.58\pm0.12$ $825\pm50$ 15.0 10's -344 83

NGC 147 119.8 -14.3 $0.18\pm0.03$ $24.30\pm0.12$ $725\pm45$ 18.5 10's -193 56

And III 119.3 -26.2 $0.05\pm0.02$ $24.40\pm0.10$ $760\pm40$ 20 60

NGC 185 120.8 -14.5 $0.19\pm0.02$ $23.96\pm0.08$ $620\pm25$ 20 -202 69

NGC 205 120.7 -21.7 $0.04\pm0.02$ $24.56\pm0.08$ $815\pm35$ 20 -241 62

M 32 121.2 -22.0 $0.08\pm0.03$ $24.53\pm0.08$ $805\pm35$ 16 $\sim10^4$ -205 54

M 31 121.2 -21.6 0.08 24.43 770 15 $\gg10^4$ -297 81

And I 121.7 -24.9 $0.04\pm0.02$ $24.53\pm0.10$ $805\pm40$ 21.7

SMC 302.8 -44.3 0.08 18.82 58 12 >10⁶ 158 900^*

Sculptor 287.5 -83.2 $0.02\pm0.02$ $19.54\pm0.08$ $79\pm4$ 16.0 100's 110 360^*

LGS 3 126.8 -40.9 $0.08\pm0.03$ $24.54\pm0.15$ $810\pm60$ -277 72

IC 1613 129.8 -60.6 $0.03\pm0.02$ $24.22\pm0.10$ $700\pm35$ 17.1 100's -234 71

And II 128.9 -29.2 $0.08\pm0.02$ $23.6\pm0.4$ $525\pm110$ 20

M 33 133.6 -31.3 0.08 24.62 840 15 >10⁴ -181 46

Phoenix 272.2 -68.9 $0.02\pm0.01$ $23.24\pm0.12$ $445\pm30$ 17.9 $\sim10^2$ 56 27

Fornax 237.1 -65.7 $0.03\pm0.01$ $20.70\pm0.12$ $138\pm8$ 14 100's 53 81

EGB 0427+63 144.7 -10.5 $0.30\pm0.15$ $25.6\pm0.7$ $1300\pm700$ -99 16

LMC 280.5 -32.9 0.06 18.45 49 12 >10⁷ 278 1150^*

Carina 260.1 -22.2 $0.04\pm0.02$ $20.03\pm0.09$ $101\pm5$ 18 $\sim 10^3$ 229 478

Leo A 196.9 +52.4 $0.01\pm0.01$ $24.2\pm0.3$ $690\pm100$ 20 6

Sextans B 233.2 +43.8 $0.01\pm0.02$ $25.64\pm0.15$ $1345\pm100$ 19.0 10's 301 47

NGC 3109 262.1 +23.1 $0.04\pm0.02$ $25.48\pm0.25$ $1250\pm165$ 403 68

Antlia 263.1 +22.3 $0.05\pm0.03$ $25.46\pm0.10$ $1235\pm65$ 361 62

Leo I 226.0 +49.1 $0.01\pm0.01$ $21.99\pm0.20$ $250\pm30$ 19 10's 168 142

Sextans A 246.2 +39.9 $0.03\pm0.02$ $25.75\pm0.15$ $1440\pm110$ 17.5 10's 324 48

Sextans 243.5 +42.3 $0.03\pm0.01$ $19.67\pm0.08$ $86\pm4$ 230 564

Leo II 220.2 +67.2 $0.02\pm0.01$ $21.63\pm0.09$ $205\pm12$ 18.6 100's 90 95

GR 8 310.7 +77.0 $0.02\pm0.02$ $25.9\pm0.4$ $1510\pm330$ 18.7 10's 214 28

Ursa Minor 105.0 +44.8 $0.03\pm0.02$ $19.11\pm0.10$ $66\pm3$ 16.9 100's -209 1000^*

Draco 86.4 +34.7 $0.03\pm0.01$ $19.58\pm0.15$ $82\pm6$ 17 100's -281 1000^*

Sagittarius 5.6 -14.1 $0.15\pm0.03$ $16.90\pm0.15$ $24\pm2$ 14 >10⁴ 140 2100^*

SagDIG 21.1 -16.3 $0.22\pm0.06$ $25.2\pm0.3$ $1060\pm160$ -77 16

NGC 6822 25.3 -18.4 $0.26\pm0.04$ $23.45\pm0.15$ $490\pm40$ -57 25

DDO 210 34.0 -31.3 $0.06\pm0.02$ $24.6\pm0.5$ $800\pm250$ 18.9 10's -137 36

IC 5152 343.9 -50.2 $0.01\pm0.02$ $26.01\pm0.25$ $1590\pm200$ 124 16

Tucana 322.9 -47.4 $0.00\pm0.02$ $24.73\pm0.08$ $880\pm40$ 18.5 10's

Pegasus 94.8 -43.5 $0.02\pm0.01$ $24.90\pm0.10$ $955\pm50$ 20 -183 40

**Table 2:** Local Group galaxies potentially accessible to GAIA. E(B-V) indicates the foreground reddening, and (m-M)₀ is the true distance modulus. $V_{\rm lim}$ is the brightest star in the galaxy. $\mu _{v_{\rm t}-v_{\rm r}}$ is the estimated proper motion, assuming the transverse velocity equals the observed radial velocity. * denotes observed values
Galaxy	l	b	E(B-V)	(m-M)₀	Distance	$V_{\rm lim}$	N(stars)	V_r	$\mu _{v_{\rm t}-v_{\rm r}}$
	( $^\circ$ )	( $^\circ$ )	(mag)	(mag)	(kpc)	(mag)	(V<20)	(helio)	( $\mu$ as/yr)
WLM	75.9	-73.6	$0.02\pm0.01$	$24.83\pm0.08$	$925\pm40$	16.5	$\sim$ 500	-116	26
NGC 55	332.7	-75.7	$0.03\pm0.02$	$25.85\pm0.20$	$1480\pm150$	15.0	10's	129	18
IC 10	119.0	-3.3	$0.87\pm0.12$	$24.58\pm0.12$	$825\pm50$	15.0	10's	-344	83
NGC 147	119.8	-14.3	$0.18\pm0.03$	$24.30\pm0.12$	$725\pm45$	18.5	10's	-193	56
And III	119.3	-26.2	$0.05\pm0.02$	$24.40\pm0.10$	$760\pm40$	20			60
NGC 185	120.8	-14.5	$0.19\pm0.02$	$23.96\pm0.08$	$620\pm25$	20		-202	69
NGC 205	120.7	-21.7	$0.04\pm0.02$	$24.56\pm0.08$	$815\pm35$	20		-241	62
M 32	121.2	-22.0	$0.08\pm0.03$	$24.53\pm0.08$	$805\pm35$	16	$\sim10^4$	-205	54
M 31	121.2	-21.6	0.08	24.43	770	15	$\gg10^4$	-297	81
And I	121.7	-24.9	$0.04\pm0.02$	$24.53\pm0.10$	$805\pm40$	21.7
SMC	302.8	-44.3	0.08	18.82	58	12	>10⁶	158	900^*
Sculptor	287.5	-83.2	$0.02\pm0.02$	$19.54\pm0.08$	$79\pm4$	16.0	100's	110	360^*
LGS 3	126.8	-40.9	$0.08\pm0.03$	$24.54\pm0.15$	$810\pm60$			-277	72
IC 1613	129.8	-60.6	$0.03\pm0.02$	$24.22\pm0.10$	$700\pm35$	17.1	100's	-234	71
And II	128.9	-29.2	$0.08\pm0.02$	$23.6\pm0.4$	$525\pm110$	20
M 33	133.6	-31.3	0.08	24.62	840	15	>10⁴	-181	46
Phoenix	272.2	-68.9	$0.02\pm0.01$	$23.24\pm0.12$	$445\pm30$	17.9	$\sim10^2$	56	27
Fornax	237.1	-65.7	$0.03\pm0.01$	$20.70\pm0.12$	$138\pm8$	14	100's	53	81
EGB 0427+63	144.7	-10.5	$0.30\pm0.15$	$25.6\pm0.7$	$1300\pm700$			-99	16
LMC	280.5	-32.9	0.06	18.45	49	12	>10⁷	278	1150^*
Carina	260.1	-22.2	$0.04\pm0.02$	$20.03\pm0.09$	$101\pm5$	18	$\sim 10^3$	229	478
Leo A	196.9	+52.4	$0.01\pm0.01$	$24.2\pm0.3$	$690\pm100$			20	6
Sextans B	233.2	+43.8	$0.01\pm0.02$	$25.64\pm0.15$	$1345\pm100$	19.0	10's	301	47
NGC 3109	262.1	+23.1	$0.04\pm0.02$	$25.48\pm0.25$	$1250\pm165$			403	68
Antlia	263.1	+22.3	$0.05\pm0.03$	$25.46\pm0.10$	$1235\pm65$			361	62
Leo I	226.0	+49.1	$0.01\pm0.01$	$21.99\pm0.20$	$250\pm30$	19	10's	168	142
Sextans A	246.2	+39.9	$0.03\pm0.02$	$25.75\pm0.15$	$1440\pm110$	17.5	10's	324	48
Sextans	243.5	+42.3	$0.03\pm0.01$	$19.67\pm0.08$	$86\pm4$			230	564
Leo II	220.2	+67.2	$0.02\pm0.01$	$21.63\pm0.09$	$205\pm12$	18.6	100's	90	95
GR 8	310.7	+77.0	$0.02\pm0.02$	$25.9\pm0.4$	$1510\pm330$	18.7	10's	214	28
Ursa Minor	105.0	+44.8	$0.03\pm0.02$	$19.11\pm0.10$	$66\pm3$	16.9	100's	-209	1000^*
Draco	86.4	+34.7	$0.03\pm0.01$	$19.58\pm0.15$	$82\pm6$	17	100's	-281	1000^*
Sagittarius	5.6	-14.1	$0.15\pm0.03$	$16.90\pm0.15$	$24\pm2$	14	>10⁴	140	2100^*
SagDIG	21.1	-16.3	$0.22\pm0.06$	$25.2\pm0.3$	$1060\pm160$			-77	16
NGC 6822	25.3	-18.4	$0.26\pm0.04$	$23.45\pm0.15$	$490\pm40$			-57	25
DDO 210	34.0	-31.3	$0.06\pm0.02$	$24.6\pm0.5$	$800\pm250$	18.9	10's	-137	36
IC 5152	343.9	-50.2	$0.01\pm0.02$	$26.01\pm0.25$	$1590\pm200$			124	16
Tucana	322.9	-47.4	$0.00\pm0.02$	$24.73\pm0.08$	$880\pm40$	18.5	10's
Pegasus	94.8	-43.5	$0.02\pm0.01$	$24.90\pm0.10$	$955\pm50$	20		-183	40

2.9 The radio/optical reference frame

The International Celestial Reference System (ICRS) is realized by the International Celestial Reference Frame (ICRF) consisting of 212 extragalactic radio-sources with an rms uncertainty in position between 100 and 500 $\mu$ as. The extension of the ICRF to visible light is represented by the Hipparcos Catalogue. This has rms uncertainties estimated to be 0.25 mas yr^-1 in each component of the spin vector of the frame, and 0.6 mas in the components of the orientation vector at the catalogue epoch, J1991.25. The GAIA catalogue will permit a definition of the ICRS more accurate by one or two orders of magnitude than the present realizations (e.g. Feissel & Mignard 1998; Johnston & de Vegt 1999).

The spin vector can be determined very accurately by means of the many thousand faint quasars picked up by the astrometric and photometric survey. Simulations using realistic quasar counts, conservative estimates of intrinsic source photocentric instability, and realistic intervening gravitational lensing effects, show that an accuracy of better than $0.4~\mu$ as yr^-1 will be reached in all three components of the spin vector.

For the determination of the frame orientation, the only possible procedure is to compare the positions of the radio sources in ICRF (and its extensions) with the positions of their optical counterparts observed by GAIA. The number of such objects is currently less than 300 and the error budget is dominated by the uncertainties of the radio positions. Assuming current accuracies for the radio positions, simulations show that the GAIA frame orientation will be obtained with an uncertainty of $\sim$ $60~\mu$ as in each component of the orientation vector. The actual result by the time of GAIA may be significantly better, as the number and quality of radio positions for suitable objects are likely to increase with time.

The Sun's absolute velocity with respect to a cosmological reference frame causes the dipole anisotropy of the cosmic microwave background. The Sun's absolute acceleration can be measured astrometrically: it will result in the apparent proper motion of quasars. The acceleration of the Solar System towards the Galactic centre causes the aberration effect to change slowly. This leads to a slow change of the apparent positions of distant celestial objects, i.e., to an apparent proper motion. For a Solar velocity of 220 km s^-1 and a distance of 8.5 kpc to the Galactic centre, the orbital period of the Sun is $\sim$ 250 Myr, and the Galactocentric acceleration has the value 0.2 nm s^-2, or 6 mm s^-1 yr^-1. A change in velocity by 6 mm s^-1 causes a change in aberration of the order of 4 $\mu$ as. The apparent proper motion of a celestial object caused by this effect always points towards the direction of the Galactic centre. Thus, all quasars will exhibit a streaming motion towards the Galactic centre of this amplitude.

Table 3: Light deflection by masses in the Solar System. The monopole effect dominates, and is summarized in the left columns for grazing incidence and for typical values of the angular separation. Columns $\chi _{\rm min}$ and $\chi _{\rm max}$ give results for the minimum and maximum angles accessible to GAIA. J₂ is the quadrupole moment. The magnitude of the quadrupole effect is given for grazing incidence, and for an angle of 1 $^\circ$ . For GAIA this applies only to Jupiter and Saturn, as it will be located at L2, with minimum Sun/Earth avoidance angle of 35 $^\circ$
Object                 Monopole term Quadrupole term

Grazing $\chi _{\rm min}$ $\chi=45^\circ$ $\chi=90^\circ$ $\chi _{\rm max}$          J₂ Grazing $\chi=1^\circ$

( $\mu$ as) ( $\mu$ as) ( $\mu$ as) ( $\mu$ as) ( $\mu$ as) ( $\mu$ as) ( $\mu$ as)

Sun 1 750 000 13 000 10 000 4100 2100          $\leq 10^{-7}$ 0.3 -

Earth 500 3 2.5 1.1 0         0.001 1 -

Jupiter 16 000 16 000 2.0 0.7 0         0.015 500 $7\ 10^{-5}$

Saturn 6000 6000 0.3 0.1 0         0.016 200 $3\ 10^{-6}$

**Table 3:** Light deflection by masses in the Solar System. The monopole effect dominates, and is summarized in the left columns for grazing incidence and for typical values of the angular separation. Columns $\chi _{\rm min}$ and $\chi _{\rm max}$ give results for the minimum and maximum angles accessible to GAIA. J₂ is the quadrupole moment. The magnitude of the quadrupole effect is given for grazing incidence, and for an angle of 1 $^\circ$ . For GAIA this applies only to Jupiter and Saturn, as it will be located at L2, with minimum Sun/Earth avoidance angle of 35 $^\circ$
Object	Monopole term	Quadrupole term
	Grazing	$\chi _{\rm min}$	$\chi=45^\circ$	$\chi=90^\circ$	$\chi _{\rm max}$	J₂	Grazing	$\chi=1^\circ$
	( $\mu$ as)	( $\mu$ as)	( $\mu$ as)	( $\mu$ as)	( $\mu$ as)		( $\mu$ as)	( $\mu$ as)
Sun	1 750 000	13 000	10 000	4100	2100	$\leq 10^{-7}$	0.3	-
Earth	500	3	2.5	1.1	0	0.001	1	-
Jupiter	16 000	16 000	2.0	0.7	0	0.015	500	$7\ 10^{-5}$
Saturn	6000	6000	0.3	0.1	0	0.016	200	$3\ 10^{-6}$

2.10 Fundamental physics

The reduction of the Hipparcos data necessitated the inclusion of stellar aberration up to terms in (v/c)², and the general relativistic treatment of light bending due to the gravitational field of the Sun (and Earth). The GAIA data reduction requires a more accurate and comprehensive inclusion of relativistic effects, at the same time providing the opportunity to test a number of parameters of general relativity in new observational domains, and with much improved precision.

The dominant relativistic effect in the GAIA measurements is gravitational light bending, quantified by, and allowing accurate determination of, the parameter $\gamma$ of the Parametrized Post-Newtonian (PPN) formulation of gravitational theories. This is of key importance in fundamental physics. The Pound-Rebka experiment verified the relativistic prediction of a gravitational redshift for photons, an effect probing the time-time component of the metric tensor. Light deflection depends on both the time-space and space-space components. It has been observed on distance scales of 10⁹-10²¹ m, and on mass scales from $1{-}10^{13} M_\odot$ . GAIA will extend the domain of observations by two orders of magnitude in length, and six orders of magnitude in mass.

Table 3 gives the magnitude of the deflection for the Sun and the major planets, at different values of the angular separation $\chi$ , for both monopole and quadrupole terms. While $\chi$ is never smaller than $35^\circ$ for the Sun (a constraint from GAIA's orbit), grazing incidence is possible for the planets. With the astrometric accuracy of a few $\mu$ as, the magnitude of the expected effects is considerable for the Sun, and also for observations near planets. The GAIA astrometric residuals can be tested for any discrepancies with the prescriptions of general relativity. Detailed analyses indicate that the GAIA measurements will provide a precision of about 5 10^-7 for $\gamma$ , based on multiple observations of $\sim$ 10⁷ stars with V<13 mag at wide angles from the Sun, with individual measurement accuracies better than $10~\mu$ as.

Recent developments in cosmology (e.g. inflationary models) and elementary-particle physics (e.g. string theory and Kaluza-Klein theories), consider scalar-tensor theories as plausible alternatives to general relativity. A large class of such theories contain an attractor mechanism towards general relativity in a cosmological sense; if this is how the Universe is evolving, then today we can expect discrepancies of the order of $\vert\gamma-1\vert\sim10^{-7}{-}10^{-5}$ depending on the theory. This kind of argument provides a strong motivation for any experiments able to reach these accuracies.

GAIA will observe and discover several hundred thousand minor planets during its five year mission. Most of these will belong to the asteroidal main belt, with small orbital eccentricity and semi-major axes close to 3 AU. The members of the Apollo and Aten groups, which are all Earth-orbit crossers, will include objects with semi-major axes of the order of 1 AU and eccentricities as large as 0.9. The Amor group have perihelia between 1-1.3 AU, and approach the Earth but do not cross its orbit.

Relativistic effects and the Solar quadrupole cause the orbital perihelion of a main belt asteroid to precess at a rate about seven times smaller than for Mercury in rate per revolution, although more than a hundred times in absolute rate.

**Table 4:** Perihelion precession due to general relativity and the Solar quadrupole moment for a few representative objects. a = semi-major axis; e = eccentricity; GR = perihelion precession in mas/yr due to general relativity; J₂ = perihelion precession in mas/yr due to the Solar quadrupole moment (assuming J₂=10^-6)
Body	a	e	GR	J₂
	(AU)		(mas/yr)	(mas/yr)
Mercury	0.39	0.21	423	1.24
Asteroids	2.7	0.1	3.4	0.001
1566 Icarus	1.08	0.83	102	0.30
5786 Talos	1.08	0.83	102	0.30
3200 Phaeton	1.27	0.89	103	0.40

Three cases of Earth-crossing asteroids are considered in Table 4 giving perihelia precession larger than Mercury, due to a favorable combination of distance and eccentricity. The diameters are of the order of 1 km for Icarus and Talos and 4 km for Phaeton. Observed at a geocentric distance of 1 AU, these objects have a magnitude between V=15-17 mag and an angular diameter of 4 mas and 1 mas respectively. Thus the astrometric measurements will be of good quality, virtually unaffected by the finite size of the source. A determination of $\lambda$ with an accuracy of 10^-4 is a reasonable goal, with a value closer to 10^-5 probably attainable from the statistics on several tens of planets. An independent determination of the Solar quadrupole moment J₂requires good sampling in a(1-e²), and one can expect a result better than 10^-7.

Revival of interest in the Brans-Dicke-like theories, with a variable G, was partially motivated by the appearance of superstring theories where G is considered to be a dynamical quantity. Using the white dwarf luminosity function an upper bound of $\dot G/G\le -(1\pm 1)~10^{-11}$ yr^-1 has been derived, which is comparable to bounds derived from the binary pulsar PSR 1913+16. Since this is a statistical upper limit, any improvement in our knowledge of the white dwarf luminosity function of the Galactic disk will translate into a more stringent upper bound for $\dot G/G$ . Since GAIA will detect numerous white dwarfs at low luminosities, present errors can be reduced by a factor of roughly 5. If a reliable age of the Solar neighbourhood independent of the white dwarf luminosity function is determinable, the upper limit could be decreased to 10^-12-10^-13 yr^-1.

2.11 Summary

With a census of the accurate positions, distances, space motions (proper motions and radial velocities), and photometry of all approximately one billion objects complete to V=20 mag, GAIA's scientific goals are immense, and can be broadly classified as follows:

The Galaxy: origin and history of our Galaxy; tests of hierarchical structure formation theories; star formation history; chemical evolution; inner bulge/bar dynamics; disk/halo interactions; dynamical evolution; nature of the warp; star cluster disruption; dynamics of spiral structure; distribution of dust; distribution of invisible mass; detection of tidally disrupted debris; Galaxy rotation curve; disk mass profile.

Star formation and evolution: in situ luminosity function; dynamics of star forming regions; luminosity function for pre-main sequence stars; detection and categorization of rapid evolutionary phases; complete and detailed local census down to single brown dwarfs; identification/dating of oldest halo white dwarfs; age census; census of binaries and multiple stars.

Distance scale and reference frame: parallax calibration of all distance scale indicators; absolute luminosities of Cepheids; distance to the Magellanic Clouds; definition of the local, kinematically non-rotating metric.

Local Group and beyond: rotational parallaxes for Local Group galaxies; kinematical separation of stellar populations; galaxy orbits and cosmological history; zero proper motion quasar survey; cosmological acceleration of Solar System; photometry of galaxies; detection of supernovae.

Solar System: deep and uniform detection of minor planets; taxonomy and evolution; inner Trojans; Kuiper Belt Objects; near-Earth asteroids; disruption of Oort Cloud.

Extra-Solar planetary systems: complete census of large planets to 200-500 pc; masses; orbital characteristics of several thousand systems; relative orbital inclinations of multiple systems.

Fundamental physics: $\gamma$ to $\sim$ 5 10^-7; $\beta$ to 3 10^-4-3 10^-5; Solar J₂ to 10^-7-10^-8; $\dot G/G$ to 10^-12-10^-13 yr^-1; constraints on gravitational wave energy for 10^-12<f<4 10^-9 Hz; constraints on $\Omega_{\rm M}$ and $\Omega_\Lambda$ from quasar microlensing.

Examples of specific objects: 10⁶-10⁷ resolved galaxies; 10⁵ extragalactic supernovae; $500\,000$ quasars; 10⁵-10⁶ (new) Solar System objects; $$\mathrel{\mathchoice {\vcenter{\offinterlineskip\halign{\hfil $\displaystyle ...$ 50000 brown dwarfs; $30\,000$ extra-Solar planets; $200\,000$ disk white dwarfs; 200 microlensed events; 10⁷ resolved binaries within 250 pc.

2 Scientific goals

2.1 Structure and dynamics of the galaxy