2.1. Bars and starburst activity

There is a clearly observed trend for nuclear starbursts to occur preferentially in barred hosts (e.g., Hummel 1981; Hawarden et al. 1986; Devereux 1987; Dressel 1988; Puxley, Hawarden, & Mountain 1988; Arsenault 1989; Huang et al. 1996; Martinet & Friedli 1997; Hunt & Malkan 1999; Roussel et al. 2001). For example, Hummel (1981) reported that the central radio continuum component is typically twice as strong in barred as in non-barred galaxies; Hawarden et al. (1986) found that barred galaxies dominate the group of galaxies with a high 25 µm / 12 µm flux ratio; and Arsenault (1989) found an enhanced bar+ring fraction among starburst hosts. Huang et al. (1996) revisited IRAS data to confirm that starburst hosts are preferentially barred, but did point out that this result only holds for strong bars (SB class in the RC3 catalogue, de Vaucouleurs et al. 1991) and in early-type galaxies, results confirmed more recently by Roussel et al. (2001). In contrast, Isobe & Feigelson (1992) did not find an enhanced far-IR to blue flux ratio among barred galaxies, and Ho, Filippenko & Sargent (1997) found only a very marginal increase in the detection rate of HII nuclei (indicative of starburst activity) among the barred as compared to non-barred galaxies in their sample, only among the late-type spirals (Sc-Sm), and most likely resulting from selection effects rather than bar-induced inflow (Ho et al. 1997). All results mentioned above rely on optical catalogues such as the RC3 to derive the morphological classifications, whereas it is now well-known that the presence of a bar can be deduced more reliably from near-IR imaging (e.g., Scoville et al. 1988; Knapen et al. 1995b). Although near-IR imaging leads to enhanced bar fractions as compared to optical imaging (e.g., Knapen, Shlosman & Peletier 2000; Eskridge et al. 2000), it is not clear how it would affect the results on bars and starbursts.

The statistical studies referred to above thus seem to show that bars and starbursts are connected, but that the results are subject to important caveats and exclusions. Further study is needed, determining bar parameters from near-IR imaging, using carefully defined samples, and exploring more direct starburst indicators than the IRAS fluxes which have often been used. Higher-resolution imaging of the starburst galaxies is also needed, to confirm the possible circumnuclear nature of the starburst, already suggested back in 1986 by Hawarden et al.