5 Data and correlations

5.1 Data description

The B-R data derived from the Nieto's sample is presented as follows:

1.

Table 6 lists global characteristics of the studied galaxies and selected numerical, or coded, colour data. Note that reference isophotal colours in the Table were measured outside dust patterns, as well as feasible, and corrected for galactic extinction and K-effect as indicated above, the adopted corrections in B-R being given in the table.

2.

The set of Figs. A1 to A32 shows maps of the B-R isochromes and B isophotes in selected small fields, of generally adequate S/N ratio, near the centre of each object.

3.

Maps of B-R in FITS format are available in electronic form, upon request to the author. A few samples are available in the anonymous space of the Nice serveur: access by ftp ftp.obs-nice.fr; directory users/bigfic/anonymous/pub/michard.

5.2 Comparison with other surveys

Reference colours

Previous surveys by Franx et al. (1989) (FIH), Peletier et al. (1990) (PDIDC) and Goudfrooij et al. (1994) (Gea) contain reference colours measured at the isophotal contour of radius $r_{\rm e}/2$. We give in Table 6 a B-R isophotal colour C₂ measured at $r_{\rm e}/2.5$, more easily reached in our small field frames. Attempts to plot colour-colour graphs from these data gave somewhat surprising results, even using the two colours from the same survey. It has been supposed that some sets of data involve much larger errors than indicated by the authors. To get some insight into this question, we compared the various sets of reference colours to the mean colours inside $r_{\rm e}$ derived from aperture photometry by Poulain (1988) or Poulain & Nieto (1994), supplemented in a few cases by our own estimates from the interpolation of available aperture photometry. Obviously, the isophotal colours at $r_{\rm e}/2$ and the mean colours within $r_{\rm e}$ are not expected to be strictly equal, but cannot differ very much: the former might be somewhat bluer because the latter are possibly influenced by the central dust patterns. For these comparisons the observed mean colours were corrected using the combined corrections applied in each of the surveys.

The results of the described tests are as follows:

. $(B-R)_{\rm FIH}-(B-R)^{0}_{\rm e}$:

N = 17; Mean = 0.276; $\sigma$ = 0.008

. $(U-B)_{\rm FIH}-(U-B)^{0}_{\rm e}$:

N = 14; Mean = -0.030; $\sigma$ = 0.042

. $(B-R)_{\rm PDIDC}-(B-R)^{0}_{\rm e}$:

N = 24; Mean = 0.041; $\sigma$ = 0.050

. $(U-B)_{\rm PDIDC}-(U-B)^{0}_{\rm e}$:

N = 24; Mean = -0.009; $\sigma$ = 0.079

. $(B-V)_{\rm Gea}-(B-V)^{0}_{\rm e}$:

N = 46; Mean = -0.000; $\sigma$ = 0.036

. $(V-I)_{\rm Gea}-(V-I)^{0}_{\rm e}$:

N = 39; Mean = -0.017; $\sigma$ = 0.166

. $(B-R)_{\rm Us}-(B-R)^{0}_{\rm e}$:

N = 41; Mean = -0.006; $\sigma$ = 0.039.

The following suggestions may be made from the numerical results, supplemented by graphs of the compared quantities:

1.

The B-R colours of FIH are not in Cousins's system as stated in their paper. A constant correction is adequate to bring their data into this system. As regards their U-B colours, they generally agree well with the $(U-B)^{0}_{\rm e}$. In view of the small number of objects, 3 values too blue by 0.1 are enough to explain the mean and $\sigma$ above.

2.

The B-R colours of PDIDC are somewhat redder than indicated by the here used aperture photometry. This was also found in a direct comparison with our results. Their U-B colours contain a few very unlikely values: NGC 2768 is found much too red, NGC 4486 and 5638 much too blue (at more than $2\sigma$).

3.

The B-V colours of Gea are in good agreement with the $(B-V)^{0}_{\rm e}$ values. On the other hand their V-I data are plagued with a number of "impossible'' values, either too red (i.e. NGC 720) or too blue (i.e. NGC 3377, 4564, 5813) by more than $2\sigma$, that is more than 0.33.

4.

Our B-R at $r_{\rm e}/2.5$ are in good agreement with the $(B-R)^{0}_{\rm e}$, as expected from our calibration sources.

Colour gradients

Although the present material is far from ideal, as emphasized before, to measure the small colour gradients in E-type galaxies, we have looked at the correlations between our "outer gradient'' G12 (see Table 6) and the results from other surveys. We have also compared the gradients for the two available colours in the PDIDC and Gea surveys. This was done by calculating the impartial correlation between centred variables, eventually rejecting extremely divergent values. The results are as follows:

1.

If $x=-\Delta(B-V)_{\rm Gea}/\Delta\log{r}$ and $y=-\Delta(B-I)_{\rm Gea}/\Delta\log{r}$we find for 42 objects (2 rejected) y=1.95x+0.009 with a coefficient of correlation $\rho=0.75$. Thus the gradients in two colours from these authors are mutually consistent.

2.

If $x=-\Delta(B-R)_{\rm PDIDC}/\Delta\log{r}$ and $y=-\Delta(U-R)_{\rm PDIDc}/\Delta\log{r}$we find for 37 objects (2 rejected) y=2.20x+0.036 with a coefficient of correlation $\rho=0.51$. The correlation between the gradients of PDIDC in the two colours appear "weaker than might be expected'' to quote these authors. Note that the two B-R and U-R gradients of FIH are very poorly correlated (with only 14 data points).

3.

If $x=-\Delta(B-V)_{\rm Gea}/\Delta\log{r}$ and $y=-\Delta(B-R)_{\rm Us} /\Delta\log{r}$we find for 21 objects y=1.40x-0.019 with a coefficient of correlation $\rho=0.60$.Our B-R gradients are reasonably consistent with the B-V gradients of Gea.

4.

If $x=-\Delta(B-I)_{\rm Gea}/\Delta\log{r}$ and $y=-\Delta(B-R)_{\rm Us} /\Delta\log{r}$we find for 20 objects y=0.73x-0.027 with a coefficient of correlation $\rho=0.82$.Our B-R gradients are again reasonably consistent with the B-I gradients of Gea.

5.

If $x=-\Delta(B-R)_{\rm PDIDC}/\Delta\log{r}$ and $y=-\Delta(B-R)_{\rm Us} /\Delta\log{r}$we find for 19 objects y=0.92x+0.013 with a coefficient of correlation $\rho=0.41$.Our B-R gradients do not correlate as well with the gradients of PDIDC as with those of Gea.

Dust patterns

The mappings of large dust patterns by various authors are generally in reasonable ageement. This is the case when comparing our data with the V-I map of NGC 1052 by Sparks et al. (1985), or the maps of Goudfrooij et al. (1994b), for NGC 2974, 3377, 4125, 4374, 4660.

Discrepancies arise, when one of the intercompared surveys is of much lesser resolution, or perhaps affected by the effects here described as "differential seeing''. For instance, Goudfrooij et al. find a blue inner disk in NGC 3610 and 4473, in contradiction with our data. An interesting case is NGC 4494: for this object V-I maps have been obtained from HST frames by Forbes et al. (1995), and by Carollo et al. (1997). These show a dust ring of subarcsec radius, obviously located in the inner disk, and much stronger on the W side of the major axis. We have overlooked this feature, although it produces in our B-R map (Fig. A23) a clear E-W colour asymmetry (for this same object, Goudfrooij et al. note a minor axis dust lane). NGC 4494 is an ideal example in favour of our speculation above, that a central accumulation of dust might be responsible for the isolated red nuclei found in a number of objects of our shp class. From a glance at the V-I radial profiles and V-I images of the above quoted authors, it seems that several other galaxies of the "power-law'' class have both nuclear dust and a sharp red peak.

Dust features may well be missed in the present survey due the poor S/N ratio of part of our frames (quoted P in Table 1). For instance a faint dust pattern is found near the center of NGC 5846 by Goudfrooij & Trinchieri (1998), which was not detected from our low exposure frames.