We investigate the well‐known correlations between the dynamical mass‐to‐light ratio (M/L) and other global observables of elliptical (E) and lenticular (S0) galaxies. We construct two‐integral Jeans and three‐integral Schwarzschild dynamical models for a sample of 25 E/S0 galaxies with SAURON integral‐field stellar kinematics to about one effective (half‐light) radius Re. They have well‐calibrated I‐band Hubble Space Telescope WFPC2 and large‐field ground‐based photometry, accurate surface brightness fluctuation distances, and their observed kinematics is consistent with an axisymmetric intrinsic shape. All these factors result in an unprecedented accuracy in the M/L measurements. We find a tight correlation of the form (M/L) = (3.80 ± 0.14) × (σe/200 km s−1)0.84±0.07 between the M/L (in the I band) measured from the dynamical models and the luminosity‐weighted second moment σe of the LOSVD within Re. The observed rms scatter in M/L for our sample is 18 per cent, while the inferred intrinsic scatter is ∼13 per cent. The (M/L)–σe relation can be included in the remarkable series of tight correlations between σe and other galaxy global observables. The comparison of the observed correlations with the predictions of the Fundamental Plane (FP), and with simple virial estimates, shows that the ‘tilt’ of the FP of early‐type galaxies, describing the deviation of the FP from the virial relation, is almost exclusively due to a real M/L variation, while structural and orbital non‐homology have a negligible effect. When the photometric parameters are determined in the ‘classic’ way, using growth curves, and the σe is measured in a large aperture, the virial mass appears to be a reliable estimator of the mass in the central regions of galaxies, and can be safely used where more ‘expensive’ models are not feasible (e.g. in high‐redshift studies). In this case the best‐fitting virial relation has the form (M/L)vir= (5.0 ± 0.1) ×Reσ2e/(LG), in reasonable agreement with simple theoretical predictions. We find no difference between the M/L of the galaxies in clusters and in the field. The comparison of the dynamical M/L with the (M/L)pop inferred from the analysis of the stellar population, indicates a median dark matter fraction in early‐type galaxies of ∼30 per cent of the total mass inside one Re, in broad agreement with previous studies, and it also shows that the stellar initial mass function varies little among different galaxies. Our results suggest a variation in M/L at constant (M/L)pop, which seems to be linked to the galaxy dynamics. We speculate that fast‐rotating galaxies have lower dark matter fractions than the slow‐rotating and generally more‐massive ones. If correct, this would suggest a connection between the galaxy assembly history and the dark matter halo structure. The tightness of our correlation provides some evidence against cuspy nuclear dark matter profiles in galaxies.

We present the emission‐line fluxes and kinematics of 48 representative elliptical and lenticular galaxies obtained with our custom‐built integral‐field spectrograph, SAURON, operating on the William Herschel Telescope. Hβ, [O iii]λλ4959,5007 and [N i]λλ5198,5200 emission lines were measured using a new procedure that simultaneously fits both the stellar spectrum and the emission lines. Using this technique we can detect emission lines down to an equivalent width of 0.1 Å  set by the current limitations in describing galaxy spectra with synthetic and real stellar templates, rather than by the quality of our spectra. Gas velocities and velocity dispersions are typically accurate to within 14 and 20 km s−1, respectively, and at worse to within 25 and 40 km s−1. The errors on the flux of the [O iii] and Hβ lines are on average 10 and 20 per cent, respectively, and never exceed 30 per cent. Emission is clearly detected in 75 per cent of our sample galaxies, and comes in a variety of resolved spatial distributions and kinematic behaviours. A mild dependence on the Hubble type and galactic environment is observed, with higher detection rates in lenticular galaxies and field objects. More significant is the fact that only 55 per cent of the galaxies in the Virgo cluster exhibit clearly detected emission. The ionized‐gas kinematics is rarely consistent with simple coplanar circular motions. However, the gas almost never displays completely irregular kinematics, generally showing coherent motions with smooth variations in angular momentum. In the majority of the cases, the gas kinematics is decoupled from the stellar kinematics, and in half of the objects this decoupling implies a recent acquisition of gaseous material. Over the entire sample however, the distribution of the mean misalignment values between stellar and gaseous angular momenta is inconsistent with a purely external origin. The distribution of kinematic misalignment values is found to be strongly dependent on the apparent flattening and the level of rotational support of galaxies, with flatter, fast rotating objects hosting preferentially corotating gaseous and stellar systems. In a third of the cases, the distribution and kinematics of the gas underscore the presence of non‐axisymmetric perturbations of the gravitational potential. Consistent with previous studies, the presence of dust features is always accompanied by gas emission while the converse is not always true. A considerable range of values for the [O iii]/Hβ ratio is found both across the sample and within single galaxies. Despite the limitations of this ratio as an emission‐line diagnostic, this finding suggests either that a variety of mechanisms is responsible for the gas excitation in E and S0 galaxies or that the metallicity of the interstellar material is quite heterogeneous.

The ATLAS3D project is a multiwavelength survey combined with a theoretical modelling effort. The observations span from the radio to the millimetre and optical, and provide multicolour imaging, two‐dimensional kinematics of the atomic (H i), molecular (CO) and ionized gas (Hβ, [O iii] and [N i]), together with the kinematics and population of the stars (Hβ, Fe5015 and Mg b), for a carefully selected, volume‐limited (1.16 × 105 Mpc3) sample of 260 early‐type (elliptical E and lenticular S0) galaxies (ETGs). The models include semi‐analytic, N‐body binary mergers and cosmological simulations of galaxy formation. Here we present the science goals for the project and introduce the galaxy sample and the selection criteria. The sample consists of nearby (D < 42 Mpc, |δ− 29°| < 35°, |b| > 15°) morphologically selected ETGs extracted from a parent sample of 871 galaxies (8 per cent E, 22 per cent S0 and 70 per cent spirals) brighter than MK < −21.5 mag (stellar mass M★≳ 6 ×109 M⊙). We analyse possible selection biases and we conclude that the parent sample is essentially complete and statistically representative of the nearby galaxy population. We present the size–luminosity relation for the spirals and ETGs and show that the ETGs in the ATLAS3D sample define a tight red sequence in a colour–magnitude diagram, with few objects in the transition from the blue cloud. We describe the strategy of the SAURON integral field observations and the extraction of the stellar kinematics with the ppxf method. We find typical 1σ errors of ΔV≈ 6 km s−1, Δσ≈ 7 km s−1, Δh3≈Δh4≈ 0.03 in the mean velocity, the velocity dispersion and Gauss–Hermite (GH) moments for galaxies with effective dispersion σe≳ 120 km s−1. For galaxies with lower σe (≈40 per cent of the sample) the GH moments are gradually penalized by ppxf towards zero to suppress the noise produced by the spectral undersampling and only V and σ can be measured. We give an overview of the characteristics of the other main data sets already available for our sample and of the ongoing modelling projects.

We present the stellar kinematics of 48 representative elliptical and lenticular galaxies obtained with our custom-built integral-field spectrograph SAURON operating on the William Herschel Telescope. The data were homogeneously processed through a dedicated reduction and analysis pipeline. All resulting SAURON data cubes were spatially binned to a constant minimum signal-to-noise ratio. We have measured the stellar kinematics with an optimized (penalized pixel-fitting) routine which fits the spectra in pixel space, via the use of optimal templates, and prevents the presence of emission lines to affect the measurements. We have thus generated maps of the mean stellar velocity V, the velocity dispersion σ , and the Gauss-Hermite moments h 3 and h 4 of the line-of-sight velocity distributions. The maps extend to approximately one effective radius. Many objects display kinematic twists, kinematically decoupled components, central stellar discs, and other peculiarities, the nature of which will be discussed in future papers of this series.

Two-dimensional stellar kinematics of 48 representative elliptical (E) and lenticular (S0) galaxies obtained with the SAURON integral-field spectrograph reveal that early-type galaxies appear in two broad flavours, depending on whether they exhibit clear large-scale rotation or not. We define a new parameter λ R ≡ R |V | / R √ V 2 + σ 2 , which involves luminosityweighted averages over the full two-dimensional kinematic field as a proxy to quantify the observed projected stellar angular momentum per unit mass. We use it as a basis for a new kinematic classification: early-type galaxies are separated into slow and fast rotators, depending on whether they have λ R values within their effective radius R e below or above 0.1, respectively. Slow and fast rotators are shown to be physically distinct classes of galaxies, a result which cannot simply be the consequence of a biased viewing angle. Fast rotators tend to be relatively low-luminosity galaxies with M B −20.5. Slow rotators tend to be brighter and more massive galaxies, but are still spread over a wide range of absolute magnitude. Three slow rotators of our sample, among the most massive ones, are consistent with zero rotation. Remarkably, all other slow rotators (besides the atypical case of NGC 4550) contain a large kpc-scale kinematically decoupled core (KDC). All fast rotators (except one galaxy with well-known irregular shells) show well-aligned photometric and kinemetric axes, and small velocity twists, in contrast with most slow rotators which exhibit significant misalignments and velocity twists. These results are supported by a supplement of 18 additional early-type galaxies observed with SAURON. In a companion paper (Paper X), we also show that fast and slow rotators are distinct classes in terms of their orbital distribution. We suggest that gas is a key ingredient in the formation and evolution of fast rotators, and that the slowest rotators are the extreme evolutionary end point reached deep in gravitational potential wells where dissipationless mergers had a major role in the evolution, and for which most of the baryonic angular momentum was expelled outwards. Detailed numerical simulations in a cosmological context are required to understand how to form large-scale KDCs within slow rotators, and Ground-based photometric MDM Observatory data (Falcón-Barroso et al., in preparation) were obtained for all galaxies of the SAURON sample. We also made use of additional Hubble Space Telescope Wide Field Planetary Camera 2 (HST/WPFC2) data which are available for 42 galaxies out of the 48 E/S0. SAURONreconstructed images were used as well to directly derive a global