Herbig-Haro objects are defined as optically visible, ``small-scale shock regions intimately associated with star formation'' (Reipurth 1999; for reviews see, e.g., Schwartz 1983; Mundt 1985; Dyson 1987; Mundt 1988; Reipurth 1989a; Edwards et al. 1993; Reipurth & Heathcote 1997). The first object of this class (Burnham's Nebula, now known as HH 255: Burnham 1890, 1894) was observed close to the prototype T Tauri star T Tau itself. In the late forties, George Herbig (1950, 1951, 1952) and Guillermo Haro (1952, 1953) independently discovered three semi-stellar objects close to the variable star V 380 Ori with peculiar emission line spectra resembling that of Burnham's Nebula. The objects, now known as HH 1, HH 2, and HH 3, show strong hydrogen recombination lines and a variety of atomic forbidden lines, in particular [SII] and [OII], and no detectable optical continuum emission. Their discovery by Herbig and Haro eventually led to the designation of this kind of objects as Herbig-Haro objects (HH-objects; Ambartsumian 1954). Their nature and origin remained a puzzle over quite some time, although it was clear from the beginning that they were somehow connected to star formation. In the years that followed, further HH-objects were found and studied. Herbig (1974) gives a compilation of HH-objects (more than 40) found up to that year; nowadays, several hundred HH-objects are known and catalogued by Reipurth (1999; see http://casa.colorado.edu/hhcat).
A major step towards an understanding of Herbig-Haro objects came with the suggestion that their spectral properties might arise in gas that is shock excited by supersonic winds from the young stars (Schwartz 1975). Several different possibilities of how a wind could produce shocks resembling Herbig-Haro objects were considered: small cloudlets exposed to the wind (Schwartz 1978), moving shock fronts (Böhm 1978), fragmentation of a stellar wind bubble into a number of fragments, ``bullets'' (Norman & Silk 1979), or refocussing shocks at the tip of ovoidal cavities created by initially spherical winds collimated by a density stratified ambient medium (Cantó 1980; Cantó & Rodríguez 1980).
The next crucial step was the discovery of the high proper motions in Herbig-Haro objects indicative of space motions of several hundred km/s (Cudworth & Herbig 1979; see Fig. 4). A particularly insightful finding was provided by the prototype Herbig-Haro objects HH 1 and HH 2: these two objects appeared to move in opposite directions, apparently away from a common source (Herbig & Jones 1981; Eislöffel et al. 1994b).
Another observation finally led to the still widely accepted basic picture of what the majority of Herbig-Haro objects are: Dopita et al. (1982) concluded that the HH 46/47 system is caused by a bipolar, very well-collimated flow, a ``jet'', from a young, embedded star. Just one year later, Mundt & Fried (1983) presented images of the areas around some young stars in the Taurus star forming region taken with new sensitive CCD array detectors. These images showed clear evidence for very well-collimated, very narrow jets from the T Tauri stars under study (see also Mundt et al. 1990, 1991). Based on this kind of observations, it was suggested that most Herbig-Haro objects were not independent entities (like bullets), but shock fronts in continuous, well collimated jets driven by young stellar objects (e.g., Mundt 1985; Mundt et al. 1987). The jet beams appear to be rather broad (of the order of 100 AU) even very close to the source. This suggests that there must be an initially wide angle wind, which is collimated into the jet beam not too far from the disk plane (Mundt et al. 1991; Ray et al. 1996).
One puzzle, however, remained: the apparent terminating working surfaces of some jets (with typical lengths of order a few tenths of a parsec) were found to run into gas which apparently was already moving away from the driving sources at high velocities (e.g., HH 34: Heathcote & Reipurth 1992; Morse et al. 1992; HH 46/47: e.g., Dopita 1978; Morse et al. 1994; HH 111: Morse et al. 1993a). Thus one had to assume that the flows were much longer than was known at that time. Indeed it is now known that many Herbig-Haro flows extend over distances of several parsecs, among them some of the finest, prototypical examples like HH 34 (Bally & Devine 1994; Devine et al. 1997; Eislöffel & Mundt 1997), HH 111 (Reipurth et al. 1997), and also the HH 46/47 system (Stanke et al. 1999). Many of the working surfaces initially thought to be the terminating working surfaces of the jets are now known to be only one of a series of internal working surfaces in a larger flow. This points to another important jet property: the ejection of matter into the jets has to be nonsteady, like a sequence of eruptions, thus creating internal working surfaces. The timescales of the periods between the ejection events of order 1000 to 2000 years are similar to those found for the FUOr outbursts. Thus it was suggested that the ejection events might have their cause in the episodic strong accretion phases of FUOr outbursts (e.g., Dopita 1978; Reipurth 1985b; 1989a; 1989b; Reipurth & Heathcote 1992).