Theoretically, we can understand spiral arms as compression regions triggered by density waves, which may be either self-excited (Lin and Shu 1964, Lin et al 1969) or due to interaction with a companion galaxy, e.g. by swing amplification of an external perturbation (Toomre 1981). Spiral arms are visually distinctive since they are the sites of the most active star formation. Thus, they host very young stars, belonging to the extreme Population I. In general, metal-rich disk stars are members of Population I, while older, metal-poor halo stars are said to belong to Population II. The existence of a Population III of first-generation stars remains speculative.

Object types certainly belonging to extreme Population I are stars with the earliest spectral types, O and B. These are bright, hot, massive and blue stars which have a lifespan of only a few million years. Therefore, they have no time to disperse, but have to stay close to their birth region within spiral arms. It is their light that lets a spiral arm appear bright and bluish. The youngest open star clusters (age < 10 Myr) also belong to the extreme Population I bright O and B stars can be part of such clusters or less well-defined ‘OB associations’. H II regions, gas clouds ionized by newly born massive and hot stars, are of course also signposts of ongoing star formation, as are the stellar nurseries, the molecular clouds, themselves.

All these objects can, in principle, be used to trace spiral structure observationally. In external galaxies, spiral arms are readily visible and the contrast of dark, filamentary molecular clouds, reddish H II regions and the blue light of massive young stars in close vicinity to each other is particularly impressive. In our own Milky Way the task of locating spiral arms is more challenging. Due to interstellar extinction, optical tracers are only useful for local spiral structure, within a distance from the Sun of ∼5 kpc. Even within this range, accurate distances are not always easy to determine. Still, an analysis of the distribution of O stars and young open clusters yields several spiral features: the local Orion arm or spur, the inner Sagittarius arm, the outer Perseus arm and a hint of another arm outside the Perseus arm, named Perseus+1 or, simply, ‘Outer Arm’. The names of these spiral arms are derived from the constellations towards which they are most clearly seen.

To trace the spiral structure throughout the Galaxy, we need extinction-free tracers. The first such tracer which arrived with the advent of radio astronomy was the 21 cm line of H I. Surveys covering almost the entire Milky Way show a distribution that vaguely resembles a spiral-like structure, but it is not easy to pin down actual spiral arms (e.g. Kerr 1969). One reason for this is, of course, that H I clouds are not extreme Population I objects they also appear off spiral arms.

Molecular clouds, and especially the 2.6 mm CO line, are better suited to delineating spiral arms, since they combine the advantages of the 21 cm transition unaffected by extinction, velocity information and, assuming a rotation curve, at least easy-to-obtain kinematic distance information with their nature as extreme Population I objects highly concentrated in spiral arms. Studies of the large-scale distribution of molecular clouds indeed gave convincing evidence of long, continuous spiral arms (Grabelsky et al 1988, Solomon and Rivolo 1989, see figure 1.1). The Sagittarius arm can be shown to connect with an arm in the constellation Carina, extending over an angle of almost 270^o. This

Figure 1.1. The spiral arm structure of the Milky Way derived from the distribution of molecular clouds. Continuous arms become visible. (Image: Huttemeister/Janssen).

shows that the Milky Way is not a ‘flocculent’ spiral, an object like M 63 which has a large number of arm fragments which give the overall appearance of a spiral galaxy, but no distinct arms. However, we find too many arm segments for the Milky Way to be (convincingly) classifiable as a ‘Grand Design’ galaxy likeM51, an object with two very high-contrast spiral arms.

The ‘classic’ tracer of spiral structure is the distribution of H II regions. This is made possible by the fact that H II regions are not only visible in the optical, but also in radio recombination lines, which are extinction-free tracers allowing the observation of objects on the far side of the Galaxy. A model of the spiral structure of the Milky Way based on the location of H II regions was first constructed by Georgelin and Georgelin in 1976. Data by Downes et al (1980) and Caswell and Heynes (1987) were added and collected by, e.g., Taylor and Cordes (1993). In addition to H II region data, tangents to the spiral arms are well defined since the line-of-sight through the spiral arm is especially long and the density of the relevant tracers very high. Models show the Perseus arm, the Sagittarius–Carina arm, the Scutum–Crux–Centaurus arm and an ‘Inner’ or 3 kpc arm as continuous features.

Based on such data, most authors envision the Milky Way as a four-armed spiral (Vallee 1995). However, more complex models like the superposition of a 2 + 4 arm pattern (Lepine et al 2001), also based on analysis of the H II region, i.e. essentially the same data set, supported by stellar kinematics and N-particle simulations, are also still discussed. The four-armed nature of the spiral is most certain in young Population I tracers, which tend to be the most luminous objects, which are as we have seen those commonly used to define spiral arms. The picture may be different when we examine the distribution of older stars. Drimmel (2000) argues that K-band data, mostly originating from older stars, are well fitted by a two-armed spiral. This may indicate that the Milky Way has a different spiral pattern in the optical and the NIR (or, more physically, in its young and old populations), a phenomenon also seen in a number of external galaxies. How simulations based on the triaxial structure of the bulge also contribute to our understanding of the large-scale spiral structure of the Milky Way, which, however, remains far from complete and perfect.