From the previous discussion it is possible to see that there is a number of questions whose answers are still escaping from us. For instance:
It is evident that a method through which one could estimate the dynamical ages of bars would greatly improve tentative answers to these questions. One possibility, explored in , is to use the vertical thickening of bars as a clock. As pointed out above, this is predicted by theoretical work and it is in very good agreement with observations. The vertical extent of bars translates directly into the vertical velocity dispersion of its stars, z. Thus, one can take spectra of face-on barred galaxies from the bar and from the disc and determine and compare the corresponding values of z. Recently formed bars should have values of z similar to that of the disc, from which it just formed. Evolved bars should have z substantially higher than the disc. Two parameters have thus been defined in :
Furthermore, with measurements of the length LBar and the colour (B - I)Bar of these bars, presented in , it is found that young bars have, on average, LBar 5.4 kpc and (B - I)Bar 1.5, whereas evolved bars have, on average, LBar 7.5 kpc and (B - I)Bar 2.2. Evolved bars are both longer and redder than young bars. The difference in length also holds when it is normalised by the galaxy diameter. The bar colour was measured close to the bar ends, but outside star forming regions.
The fact that evolved bars are longer than young, recently formed bars is in agreement with theoretical results [7, 3, 66]. These works indicate that, while bars evolve, they capture stars from the inner disc, redistribute angular momentum along the disc and dark matter halo, and get longer and thinner in the process. NGC 4608 and NGC 5701 might represent cases where the capture of disc stars by the bar is substantial. Recent results suggest that the bar in NGC 4608 had an increase in mass of a factor of 1.7, through the capture of 13% of disc stars (; see also [38, 54]).
The difference in colour between young and evolved bars represents a difference in the age of their stars of 10 Gyr. As seen in the previous section, bars seem to follow two different patterns of star formation, which might be related to the dynamical age. A recently formed bar seems to form stars along its whole extent, whereas an evolved bar seems to form stars mostly at its centre and/or its ends. This indicates that, when one carefully measures the colour of stars in the middle of the bar, one is probing mainly the first generation of stars formed in the bar. That seems to be the reason why those bars which are dynamically old, as estimated from their stellar kinematics, are also redder than the dynamically young bars. Alltogether, these results also indicate that at least some bars are very old, and thus most likely not recurrent (unless the first generation of bars has very short life times).
Interestingly, dynamically young bars are found preferentially in gas-rich, late type spirals. This suggests that bar recurrence is restricted to this class of galaxies, as expected from such models (see discussion in previous section). In addition, it was also found that galaxies hosting AGN typically have young bars, which possibly means that the funnelling of gas to feed the black hole at the nucleus occurs on short time-scales. A similar conclusion is reached in  after the finding of a significantly higher bar fraction in narrow-line Seyfert 1 galaxies, which are supposedly in an early stage of black hole evolution.
These results come, however, from the analysis of a small sample of galaxies. It is highly desirable to have the dynamical ages of bars measured for a much larger sample, and assess the validity of these results. Furthermore, at the current stage, one can only discriminate between recently formed and evolved bars. It is now difficult to measure with more precision the dynamical age of the bar. Finally, one would like to be able to give more stringent numbers to this parameter without having to rely on the age of the bar stellar population. The accuracy in estimates of the dynamical age of a bar has to be improved. This might be accomplished by an approach involving both observations (e.g. with a large scale 2D mapping of z in barred galaxies) and theory (e.g. with a more detailed analysis of the vertical evolution of bars with time).