This is a brief historical review of the discovery of binary stars, starting in the 18th and 19th century with the work of W. Herschel, W. Struve, and their famous sons. I describe how Ch. Doppler in 1842 found the Doppler effect in connection with the colours of double stars. I also describe how the observation of the stationary Calcium H and K lines in the spectrum of the spectroscopic binary δ Ori led to the discovery of the interstellar medium by J. Hartmann in 1904 with the Potsdam refractor. Some steps in the discovery of young binary stars are recounted, starting with the paper of Joy and van Biesbroeck in 1944. The history of ideas regarding the origin of binary systems is also briefly reviewed, beginning with J. Michell's statistical argument in 1767 that many close visual pairs of stars must indeed form physical systems, and W. Herschel's definition of a binary star as “a real double star - the union of two stars that are formed together in one system, by the laws of attraction”.

Although observations strongly suggest that fragmentation during the protostar collapse is the primary formation mechanism for wide binaries, the theoretical calculations as yet do not well explain the statistical properties of such systems. The results of a number of numerical simulations are discussed, and it is pointed out that, although fragmentation is obtained in such calculations, in many cases either the initial conditions are not realistic, or the calculations are insufficiently resolved, or the calculations have not been taken far enough to account for the accretion of most of the initial core material onto the components of the forming system. Certain aspects of the numerical results are, however, consistent with the fragmentation hypothesis.

I review the possible formation mechanisms of close binary stars. The formation of close binary systems is problematic in that there is no theory that does not encounter significant difficulties or unknowns. Fission does not appear to occur in stars. Capture is unlikely to form many close binary systems except possibly amongst massive stars. Fragmentation can form close binary systems but these need to accrete the majority of their eventual mass. Furthermore, there appears to be a limited window in initial conditions that may preclude forming sufficient systems in this way. Possible alternatives include the orbital migration of a binary due to its circumbinary disk and the disintegration of a non-hierarchical multiple system.

We discuss the important roles which impulsive processes seem likely to play in the formation of binary star systems and higher multiples. On the basis of numerical simulations and theoretical considerations, we show (i) that when a dense layer is produced by a cloud/cloud (or clump/clump) collision, or a dense shell is swept up by an expanding nebula (HII Region, Stellar-Wind Bubble or Supernova Remnant), fragmentation leads to the formation of small-N subclusters of massive protostellar discs, including many wide binaries; (ii) that impulsive interactions between these protostellar discs are frequent and can lead to an hierarchical cascade, spawning new protostellar discs in closer binary systems, plus some escapers. This cascade is particularly effective if there is continuing infall replenishing the protostellar discs. The binaries formed have a wide range of separations and orbital eccentricities, and the Mass Function of the new protostars has an exponent ∼ −1.

During the 1980s, numerical simulations showed that dynamic growth of a barlike mode in initially axisymmetric, equilibrium protostars does not lead to prompt binary formation, i.e., fission. Instead, such evolutions usually produce a dynamically stable, spinning barlike configuration. In recent years, this result has been confirmed by numerous groups using a variety of different hydrodynamical tools, and stability analyses have convincingly shown that fission does not occur in such systems because gravitational torques cause nonlinear saturation of the mode amplitude. Other possible routes to fission have been much less well scrutinized because they rely upon a detailed understanding of the structure and stability of initially nonaxisymmetric structures and/or evolutions that are driven by secular, rather than dynamic processes. Efforts are underway to examine these other fission scenarios.

Long-term radial-velocity surveys of G, K and M dwarfs of the solar neighbourhood are presented. The inferred orbital elements are discussed, focusing on the (e, log P) diagram, the mass-ratio and secondary mass distributions, and on the binary frequency of the studied samples. The proportion of companions to M dwarfs is found to be not significantly different from the binary frequency among G- and K-dwarf primaries. The mass function of stellar and planetary companions to solar-type stars strongly suggests different formation and evolution mechanisms for the two populations. Finally, beautiful mass-luminosity relations are obtained from low-mass binaries with high-precision radial-velocity measurements and adaptive optics visual data.

Observations of ultra-cool (spectral type later than M6) binaries are summarized. Only a few systems are known, and all have been discovered in the past few years. We describe each of these discoveries. Despite their small numbers, some interesting trends among the binaries may be emerging. Ultra-cool binaries have a frequency similar to but perhaps a little less than stars (binary fraction ∼20%) and have mass ratios closer to unity, with respect to solar-type stars. Neither of these results can be considered firm, as there are far too few systems known, and observational biases would tend to move trends in these directions. There also seems to be a trend towards smaller separations in lower mass systems, which is less subject to observational biases. We discuss how these results fit into current ideas about binary formation, and favor a fragmentation scenario.

Hipparcos observations of double stars, mainly A and F primaries, are modelled in order to check the statistics of double-star solutions in the Hipparcos Catalogue and to set quantitative limits on binary distributions. The number of companions per primary is estimated to 0.25 ± 0.05 for semi-major axes from 1 to 10 AU, and to 0.43 ± 0.20 for the range 10–100 AU. The total multiplicity found is between 0.9 and 1.2.

We discuss the observed multiplicity of massive stars and implications on theories of massive star formation. After a short summary of the literature on massive star multiplicity, we focus on the O-and B-type stars in the Orion Nebula Cluster, which constitute a homogenous sample of very young massive stars. 13 of these stars have recently been the targets of a bispectrum speckle interferometry survey for companions. Considering the visual and also the known spectroscopic companions of these stars, the total number of companions is at least 14. Extrapolation with correction for the unresolved systems suggests that there are at least 1.5 and perhaps as much as 4 companions per primary star on average. This number is clearly higher than the mean number of ∼ 0.5 companions per primary star found for the low-mass stars in the general field population and also in the Orion Nebula cluster. This suggests that a different mechanism is at work in the formation of high-mass multiple systems in the dense Orion Nebula cluster than for low-mass stars.

A WR binary frequency of 37% is found in the VIIth Catalogue of Galactic Wolf-Rayet Stars. Indications are that perhaps all WR stars are binaries and thus that all WR stars point to massive binary formation regions.

The available information on the statistics of high multiplicity (3–6 components) systems is reviewed. The ratio of triple to binary systems is f3 ≍ 0.11, while fn ≍ 0.25 for higher n. Despite selection effects in the multiple star catalogue, the signatures of formation mechanisms are found in the distributions of period ratios and mass ratios. For example, the frequent occurrence of close sub-systems with periods less than 6 days can be explained by tidal dissipation in a 3-body system. In triple stars the angular momentum vectors of inner orbits are inclined to those of outer orbits by an average angle of 50°, hence the orbital spins are neither co-aligned nor completely random. Close binaries have a tendency to be found in higher-multiplicity systems, showing that close and wide binarity is statistically related. Future theoretical and observational studies are outlined.

The overall frequency and other statistical properties of binary systems suggest that star formation is intrinsically a complex and chaotic process, and that most binaries and single stars actually originate from the decay of multiple systems. Interactions between stars forming in close proximity to each other may play an important role in the star formation process itself, for example via tidally induced accretion from disks. Some of the energetic activity of newly formed stars could be due to bursts of rapid accretion triggered by interactions with close companions.

A small fraction of the mass of a molecular cloud is in dense, gravitationally bound structures many of which eventually form stars. This paper reviews some of the important properties of these dense cores. Ideas about the formation of these cores are summarised and recent observational results on the structure and evolution of cores reviewed.

The Bok globule CB230 (L1177) contains an active, low-mass star-forming core which is associated with a double NIR reflection nebula, a collimated bipolar molecular outflow, and strong mm continuum emission. The morphology of the NIR nebula suggests the presence of a deeply embedded, wide binary protostellar system. High-angular resolution observations now reveal the presence of two sub-cores, two distinct outflow centers, and an embedded accretion disk associated with the western bipolar NIR nebula. Judging from the separation and specific angular momentum, the CB230 double protostar system probably results from core fragmentation and will end up at the upper end of the pre-main sequence binary separation distribution.

Numerical models of molecular cloud cores are obtained, with assumed Gaussian random velocity fields that are consistent with moderate turbulence. If the velocity power spectrum is taken to be P(k) ∼ kn, with n = − 4 or −3, then the constructed line-of-sight velocity maps on the plane of the sky show gradients that can be interpreted as rotation. Deduced values of angular velocity, angular momentum, and dimensionless rotational parameter β are consistent with observations. The velocity gradient Ω, of an individual core is not a good indicator of its intrinsic angular momentum J/M. However, the distribution of deduced angular momenta from a large sample of cores with different random velocity fields is close to the distribution of actual angular momenta of these model cores if one assumes J/M = pΩR2 where R is the core radius and p must be determined from a Monte-Carlo study. For centrally condensed cores the standard value of p = 0.4 overestimates the mean intrinsic angular momentum by a factor of 3.

VLA 7 mm continuum observations carried out by Rodríguez et al. (1998) with 0.06″ resolution toward L1551 IRS5 have shown two resolved dust sources. Both sources are elongated with almost the same orientation, similar flux densities, and similar angular sizes, being interpreted by these authors as a system of twin, aligned protoplanetary disks of 10 AU radius around the components of a young binary of 50 AU separation. These protoplanetary disks, the smaller ones ever imaged at mm wavelengths, are about a factor of ten smaller than disks around isolated stars, but have enough mass to form solar systems similar to our own. The fact that the disks are aligned with each other argues in favor for long-lived planetary systems in this binary. We all look forward to the future development of interferometers such as the EVLA, SMA, and ALMA in order to characterize the physical conditions of these associated protoplanetary disks by detecting spatially resolved molecular line emission. With these upcoming interferometers it will be possible to make great strides forward in our knowledge of the star formation processes.

High resolution images at millimeter wavelengths are providing new insights into the formation of binary and multiple star systems. These wavelengths are particularly useful in studying the earliest stages of multiplicity because they trace the bulk material distribution in the circumstellar environment and can penetrate 1000's of magnitudes of visual extinction. Current millimeter wavelength observations are finding a high incidence of multiplicity among young systems and that multiplicity begins at birth, or before. While the statistics are poor, the types of systems found (independent envelope, common envelope, and common disk systems) follow theoretical ideas about binary and multiple star formation. Systems can be identified which exhibit the characteristics of prompt initial collapse, central fragmentation during collapse, and fragmentation in high angular momentum scenarios. Expansion of this work to more systems and to more detailed studies of the structure and kinematics of individual systems will provide valuable insights into the formation of multiple systems.