5 Discussion

The evaluation of the 2D-spectra and the simultaneously taken broadband images show that the commonly called G-band bright points are characterized spectroscopically by an increased line core intensity of the CH absorption lines. They are embedded in regions that still show enhanced line core intensity of the CH-lines in comparison to the quiet Sun.

Our observations suggest to classify G-band bright structures by their spectroscopic signature. The defined BPI is a suitable tool to quantify the spectroscopic signature of G-band bright structures. A high value of the BPI, corresponding to a decreased line depression of the CH in comparison to the Fe II-line, characterizes G-band bright structures with an enhanced G-band contrast and is related to the downflow regions of the intergranular lanes. Their brightness is caused by significant weakening of the CH-absorption line. The remaining G-band bright structures show a low BPI and a lower CH line core contrast, but an increased contrast in the narrow band continuum. They are related to granules and do not to show a downflow. Their brightness is caused by enhanced continuum intensity.

The presented results confirm and complement former observations (Langhans et al. 2001). The BPI provides an identification method for G-band bright structures from the spectroscopic viewpoint. The statistics of our data set shows that the BPI is a reliable criterion for the identification.

To our knowledge no simultaneously observed magnetograms are available for the data presented here. But it seems reasonable to suppose that our spectroscopic distinction of G-band bright structures, by analysis of the BPI and the flow, is identical to the classification in magnetic and non-magnetic sub-arcsecond structures done by Berger & Title (2001). Berger & Title (2001) find that bright points that are co-spatial with magnetic fields, are located in intergranular lanes and the non-magnetic ones at the edges of certain bright rapidly expanding granules. Concerning the location, this distinction matches with the measured flows. Berger & Title (2001) find that within the best spatial resolution of their magnetograms ($\sim$0.3 arcsec) all G-band bright points, properly distinguished from granulation brightening, are magnetic in nature.

Assuming LTE and a similar formation height for the Fe II- and the CH-line, we can identify the brightness temperature, obtained from the measured absolute intensity in the Fe II-line core, with the temperature at $\tau=\mu=1$. An unchanged Fe II-line core intensity within a G-band brightening implies that the enhanced CH-line intensity is most likely not due to a direct temperature effect - be it through the line source function or through enhanced dissociation of CH. Consequently the CH dissociation must be caused by different mechanisms, like hot wall radiation. Our observations do not contradict this explanation, but further observations at higher spatial resolution and under constant seeing conditions are required to confirm this hypothesis.

Acknowledgements

J. Bruls provided very useful hints and comments. One of us (K.L.) is supported by the Deutsche Forschungsgemeinschaft.