Astron. Astrophys. 322, 19-28 (1997)

3. Results of the separation

Fig. 2 shows the distribution of the derived quantities. The diagram at the top presents the distribution of , and the distribution of is plotted at the bottom. In case of galaxies with upper limits in the value of used is the mean of the range in given in Col. 4. The morphological types of the galaxies are coded with different grey scales. The mean values of the distributions with defined values are and . The calculation of the mean values of the subsamples with defined and and of the subsamples with upper limits in the two quantities does not yield any significant differences in the statistical moments of the subsamples. In the further analysis, the upper limits were treated in the same way as the defined values. Hence, calculation of the mean values of and of all 74 galaxies yields and . The standard deviations of the distributions of and are and , respectively. The extrapolation of the thermal fraction to 10 GHz yields a distribution which has a peak between 20 % and 30 % (Fig. 3). Only 15 % of the galaxies in our sample have a thermal fraction larger than , so that the non-thermal emission dominates the radio emission of galaxies at 10 GHz. The mean value of the distribution of is . The distributions of and seem to be independent of morphological type.

Fig. 2. The distribution of and . The upper diagram shows the distribution of , and the lower one the distribution of . The different morphological types in the galaxy sample are coded with different grey scales

Fig. 3. The distribution of . The different grey scales represent the morphological types of the galaxies

On the other hand, the morphological type and the non-thermal spectral index are not completely uncoupled. The range of is mainly covered by irregulars and early-type spirals. Spiral galaxies of type Sc are found in the range of . A subdivision of the sample into three subsamples consisting of Sa/Sab, Sb/later and Irr/Am galaxies yields the following mean values for the non-thermal spectral indices:


.
In case of early-type spirals the flat non-thermal spectral index may be influenced by emission from compact cores which may be present in them. In Paper I we have shown that the core emission dominates the total radio emission of spirals of type Sa and Sab. The emission of galactic nuclei often shows inverted radio spectra. This has been shown for e.g. NGC 4594 by de Bruyn et al. (1976). In case of core-dominated radio emission these inverted spectra can significantly flatten the integrated spectrum of a galaxy. The flat synchrotron spectra of some irregular and amorphous galaxies may indicate a high escape probability of cosmic ray electrons in these galaxies. According to Klein et al. (1991) the efficiency of particle confinement is lower in less massive galaxies. In these, the relativistic electrons can escape very easily from their host galaxies and have no time to undergo loss processes which would steepen the particle spectrum. Additionally, in actively star-forming galaxies the occurrence of winds may produce a flat synchrotron spectrum.

Eleven galaxies show very steep synchrotron spectra. (). This may indicate that synchrotron and inverse Compton losses are strongly affecting the relativistic electron population of these galaxies, which may happen if these galaxies efficiently confine the cosmic-ray particles.

Comparison with the study of radio spectral indices by Gioia et al. (1982) shows that the influence of the thermal emission flattens the radio spectrum of most of the galaxies. Our analysis agrees with the investigation of the radio spectra of 13 galaxies by Klein (1988). The distribution of derived by Klein (1988) yielded a mean non-thermal spectral index of . This is in good correspondence with Fig. 2, and in excellent agreement with our mean value. However, the larger number of galaxies and the high quality of the data of our analysis have strongly increased the statistical significance, which can be seen if one compares the dispersion in in this work () with the dispersion derived by Klein (1988, . Our separation of thermal and non-thermal radio emission disagrees with the results for 32 galaxies obtained by Duric et al. (1988). Their distribution of the spectral indices is in principle similar to Fig. 2, but Duric et al. found variations of the thermal fraction at 5 GHz from 0 % up to 90 %. These variations were not only found among different galaxy types, but also for galaxies of the same type. Furthermore, the results of Duric et al. (1988) suggest a coupling between the thermal fraction and non-thermal spectral index, in the sense that galaxies with a high thermal amount of emission exhibit steep non-thermal spectra. There exists no physical reason for this kind of a correlation, and it may be a selection effect produced by bad data. If one plots from Table 1 versus the corresponding , no correlation is found (correlation coefficient 0.2). This shows that the statistical analysis of the data in our work is correct. In contrast to the findings of Duric et al. (1988), we can say that the constancy of and indicates that the physical conditions in the interstellar medium do not vary strongly from galaxy to galaxy and that the non-thermal spectral index seems to be the same among all normal galaxies with a mean value of .

© European Southern Observatory (ESO) 1997

Online publication: June 30, 1998
helpdesk.link@springer.de