4 The photometry errors

The internal errors of the instrumental (non-calibrated) magnitudes and those of the particular SB profile are dominated by the error of the adopted sky background value and includes, in addition, the random count error in aperture measurements. The smooth sky background level at the position of the (masked) galaxies was computed (within PIPS) by means of an iterative filtering process which produces a minimum surface between the surrounding background of every particular masked object. That means, the computed sky level depends on local SB fluctuations. We measured the sky background fluctuations on the filtered CCD frames by means of small apertures in the vicinity of the studied galaxies. For the 600 s B exposures (and 300 s R exposures) the sky background fluctuations are typically 0.2 percent. For the snap-shot 60 s exposures the corresponding fluctuations are 0.4 - 0.8 percent in B and 0.5 percent in R. Thus, the SB profiles are sky background limited at $\mu_{\rm lim} \leq$ 27.5 B mag/ $\ifmmode\hbox{\rlap{$\sqcap$ }$\sqcup$ }\else{\unskip\nobreak\hfil \penalty50\h... ...\rlap{$\sqcap$ }$\sqcup$ } \parfillskip=0pt\finalhyphendemerits=0\endgraf}\fi''$ for typical 60 s exposures and can be traced down to $\mu_{\rm lim} \sim$ 28.5 B mag/ $\ifmmode\hbox{\rlap{$\sqcap$ }$\sqcup$ }\else{\unskip\nobreak\hfil \penalty50\h... ...\rlap{$\sqcap$ }$\sqcup$ } \parfillskip=0pt\finalhyphendemerits=0\endgraf}\fi''$ on the deepest (900 s) frames. Beyond the given limits the error in the mean sky begins to exceed the signal from the object. Following Vader & Chaboyer ([1994]) we calculated the internal errors in intensity as

$$\Delta I = \sqrt{N_{\rm tot}+(\delta~n_{\rm sky}~A)^2},$$ (3)

where $N_{\rm tot}$ is the total number of counts, as measured on the adaptive filtered frame within aperture A, which is the area between successive isophotes (in pixels), $n_{\rm sky}$ is the mean sky counts per pixel, and $\delta$ is the fractional error in the mean sky value. This gives a signal-to-noise ratio of $(N_{\rm tot}-A~n_{\rm sky})/\Delta I$. The internal uncertainties are shown with error bars in Fig. 1.

Figure 2: The 1$\sigma$ deviations of the surface brightnesses of individual B profiles from their mean values (see text) as a function of the surface brightness

Next, we checked the internal consistency of the obtained SB profiles. For a number of ELGs we obtained several frames taken during different nights or different observing sessions. Most of the repeated observations were made in B passband. All these B frames were processed individually. Then, the mean SB profile was calculated for a particular galaxy and the residuals of every single profile from the mean profile were computed. The 1$\sigma$ dispersion of these residuals is plotted as a function of the corresponding B-surface brightness in Fig. 2. This consistency test involves 13 galaxies with 30 individual SB profiles mostly extracted from 60 s snap-shot exposures, and 6 profiles which were obtained from 900 s exposures. The mean error is nearly constant at the $\sigma~\sim~$ 0.075 mag/ $\ifmmode\hbox{\rlap{$\sqcap$ }$\sqcup$ }\else{\unskip\nobreak\hfil \penalty50\h... ...\rlap{$\sqcap$ }$\sqcup$ } \parfillskip=0pt\finalhyphendemerits=0\endgraf}\fi''$ level within 20.0 $\leq \mu \leq$26.0 B mag/ $\ifmmode\hbox{\rlap{$\sqcap$ }$\sqcup$ }\else{\unskip\nobreak\hfil \penalty50\h... ...\rlap{$\sqcap$ }$\sqcup$ } \parfillskip=0pt\finalhyphendemerits=0\endgraf}\fi''$, followed by a rapid increase at lower surface brightnesses. At higher SB ($\mu \leq$ 20.0), which is the range of typical central surface brightnesses of ELGs, the 1$\sigma$ dispersion increases again, as an effect of varying seeing conditions during the different observing sessions.

In order to evaluate the true magnitude calibration errors we compared our total magnitude measurements with the available published data, as quoted in NASA/IPAC Extragalactic Database (NED). There are very few reliable photometric data available in the literature for the studied ELGs. For two galaxies, namely HS 1311+3628 (UGC 8303) and HS 1312+3508 (UGC 8323), the $B_{\rm T}$ magnitudes are given in RC3 (de Vaucouleurs et al. [1991]). For four further galaxies, (HS 1331+3906, HS 1336+3114, HS 1400+3927 and HS 1402+3657) total magnitudes have been listed by Garnier et al. ([1996]), with typical uncertainties of about 0.5 mag. For the HS 1400+3927 our total brightness is about 1 mag fainter than the one quoted by Garnier et al. ([1996]). The reason of this rather large difference is unclear. The remaining five common galaxies show reasonably small magnitude residuals with a mean value of $<B_{\rm T,our} - B_{\rm T,cat}>$ = +0.17 mag, and an 1$\sigma$ dispersion of 0.21 mag.

Available B-magnitudes of the studied ELGs were collected by Popescu et al. ([1996], Table 4, ($B_{\rm L}$, Col. 7)) from various literature sources. The mean accuracy of these heterogeneous data was estimated to be of $\sim$ 0.5 mag. In Fig. 3 we compare these magnitudes with our data (open circles on the plot), for 22 ELGs in common with our sample. The dispersion of individual values is large (1$\sigma$ = 0.78 mag). However the mean value of the magnitude residuals, $<B_{\rm T,our} - B_{\rm L}>$ = +0.26 $\pm$ 0.16, is in agreement with the accuracy of the literature data, $B_{\rm L}$, and no systematic trend is evident.

Figure 3: Comparison of our measured B-magnitudes with the data obtained from literature: our $B_{\rm T}$ magnitudes versus $B_{\rm T}$ magnitudes obtained from RC3, or as quoted in Garnier et al. ([1996]) - filled circles ($\bullet$); our B25 magnitudes versus m24.5, as measured by Odewahn & Alderding ([1995]) on the 103a-O photographic plates - crosses ( +); our $B_{\rm T}$ magnitudes versus heterogeneous magnitudes as collected by Popescu et al. ([1996]) from different literature sources ($B_{\rm L}$) - open circles ($\circ$)

Further we compare our isophotal magnitudes (B₂₅) with the photographic m_24.5 magnitudes measured by Odewahn & Alderding ([1995]) on the blue 103a-O plates (Fig. 3, crosses). Our B₂₅ magnitudes are systematically more luminous: $<B_{25} - m_{\rm 24.5,O}>$ = -0.42 $\pm$ 0.06, as could be expected because of the limiting isophote differences. However, the dispersion of individual data points is reasonably small (1$\sigma$ = 0.25), and no systematic trend is evident within the given magnitude interval. As a result of these comparisons we estimate the accuracy of our isophotal and total B-magnitudes to be not worse than $\pm$0.2 mag.

For the R-magnitudes there are no comparison data available in the literature, except the photographic $m_{\rm 23.5,E}$ magnitudes measured by Odewahn & Alderding ([1995]) on the red 103a-E plates. In Fig. 4 our total colour indices ( $B_{\rm T}-R_{\rm T}$) are plotted versus $(m_{\rm 24.5,O}-m_{\rm 23.5,E}$). Since the mean colour of ELGs is $<B-R> \simeq$ 0.9, the photographic $m_{\rm 24.5,O}$ and $m_{\rm 23.5,E}$ magnitudes of common galaxies should refer to nearly the same aperture. Rejecting one single discordant photographic colour value of $m_{\rm 24.5,O}-m_{\rm 23.5,E}$ = 2.34 for HS 1325+3255, the remaining 10 common galaxies show that our measured colours are systematically bluer by about 0.13 mag than the photographic colours (1$\sigma$dispersion of 0.21 mag). The given systematics in the colour indices is probably mainly caused by differences in the CCD and photographic magnitude systems.

Figure 4: Comparison of our measured total (B-R) colour indices with photographic colours from Odewahn & Alderding ([1995])