4. Observed connections
The following subsections discuss the connections or absence of connections between various quantities referring either to the bar morphology, or to the star formation activity, or to abundances indices. Some conventional terminologies are useful. A bar with an axis ratio will be called strong in the text, whereas the term weak will be reserved to bars with . Similarly, a bar will be called long if its relative length and short in the opposite case. Finally, we will consider two classes of FIR colours ( or ) corresponding to more or less pronounced star formation activity. Clearly, these chosen limits are somewhat arbitrary. However, slight changes of these values do not affect the results presented below. They have been chosen as follows: The axis ratio of 0.6 corresponds to the middle of the interval of observed values, i.e 0.2 - 1.0. The length of 0.18 separates our sample in roughly half long bars and half short bars. The FIR colour of -1.2 approximately separates systems whose FIR emission is dominated either by star formation from those dominated by cirrus (see Sect. 2.1.3).
4.1. Star formation activity - bar strength
The link between the and the deprojected bar axis ratio (the bar "strength" parameter) is shown in Fig. 1. In our sample, all the galaxies having have strong bars (10 galaxies). On the contrary, all the weakly barred galaxies display low current star formation activity (14 galaxies). There are also strongly barred galaxies which do not actively form stars (8 galaxies). These non-active strongly barred galaxies could be either in a "pre-starburst" or in a "post-starburst" phase (see Sect. 5). Using other indicators of star formation activity mentioned in Sect. 2.1.3 (e.g. ) does not alter the tendency shown in Fig. 1. With only 32 objects, our statistics is still poor and one should remain cautious before drawing general conclusions. However, in Fig. 1 the different behaviour of weak and strong bars is striking and consistent with the results of numerical simulations presented in Sect. 5. Moreover, Martin (1995) had also noticed that the fraction of strong bars is higher in galaxies with nuclear activity than in quiescent galaxies.
4.2. Star formation activity - bar length
The link between the and the relative deprojected bar length is shown in Fig. 2. All the galaxies but one (NGC 1637) with have long bars (9 galaxies). This can be explained as follows: Bar-driven movement of gas towards the center takes place inside the co-rotation radius, which is generally close to the end of the bar. Thus in those systems with longer bars, a greater fraction of the total store of gas in the system can be swept up and driven towards the center. Similarly, all the galaxies with a short bar appear to be more quiescent (15 galaxies). However, there are also galaxies with low star formation and long bars (7 galaxies). Thus, the increase of the bar length generally seems to have a similar effect as the decrease of the bar axis ratio.
Surprisingly enough, these two quantities are strongly correlated in our sample (see Fig. 3). Strong bars are long (except NGC 1637 and NGC 6744) and weak bars are short. Whereas it is well-known that on average bars of early-type galaxies are longer than those of late-type galaxies (e.g. Martin 1995), so far no correlation between the length and strength of late-type bars seems to have been highlighted.
4.3. Abundance gradient - bar strength
Martin & Roy (1994) established a correlation between the radial O/H abundance gradient in the discs of SBs and the bar axis ratio in the sense that stronger bars have a rather flat gradient, whereas steeper gradients are observed in galaxies with weak or no bars. Taking into account the internal uncertainties and comparing with the data by other authors (e.g. Vila-Costa & Edmunds 1992; Zaritsky et al. 1994), a rather larger dispersion of points in the diagram versus is observed but the above general trend is clearly present.
This suggests that other parameters, such as the efficiency of star formation might play a role in the connection between the chemical and dynamical evolution of bars. According to Tinsley (1980), in a quasi-stationary state the radial chemical gradient essentially depends on the ratio of two timescales, i.e.
where is the characteristic timescale of gas inflow through the center, is the characteristic timescale for exhausting gas through star formation, and y is the yield. The extension of this model to the present context suggests that at first approximation since stronger bars have higher gas mass inflow (as shown e.g. by Friedli & Benz 1993). Furthermore . So, locally the dependence of the chemical gradient on bar axis ratio must be weighted by the star formation efficiency.
Due to the lack of data in our sample, we must restrict ourselves to register a qualitative agreement between relative observed and calculated gradients for NGC 3344, 4303, 4321, 5236, and 6946. But the real situation can even be more complicated. First, the simplified formula above also shows an R dependence. Second, in some galaxies two different slopes for the O/H abundance gradient have recently clearly been inferred, i.e. NGC 3359 (Martin & Roy 1995) and NGC 1365 (Roy & Walsh 1997). In these two galaxies, the abundance gradient is flat in the disc region, whereas a moderate negative gradient subsists in the bar region (see Table 1). Note that very few galaxies have at least 30 measured HII regions, a necessary condition to be in position to highlight this feature. The numerical simulations reported in Sect. 5 show this feature and indicate that the age of the bar is another factor influencing the chemical gradient.
4.4. Connections with Hubble type
The sample of Table 1 contains late-type galaxies with Hubble types between T=3 and 7. No link has been found between the Hubble type and either , or , or .
© European Southern Observatory (ESO) 1997
Online publication: June 5, 1998