6 Statistical properties of the sample

The studied sample of 139 ELGs (and 2 non-ELGs) was further divided into two subsamples of isolated and non-isolated galaxies, according to the clustering properties of its galaxies. The criteria for defining isolated galaxies are given in Popescu et al. ([1997]), based on the calculation of the nearest neighbour distances ( $D_{\rm NN}$). The $D_{\rm NN}$ represent real separations in the 3-dimensional space and are calculated (Popescu et al. [1997]) as the separation between each ELG and its nearest neighbour from a comparison catalogue (ZCAT; Huchra et al. [1992]; Huchra et al. [1995]), for galaxies with blue magnitudes brighter than 15.5. As an example, we illustrate the case of an isolated galaxy HS 1236+3937 (z=0.0184). It has its nearest neighbour ZCAT galaxy at a distance 8.68h^-1Mpc. At this redshift the average nearest neighbour distances between the non-isolated ELGs from the field environment are 0.75h^-1Mpc. Thus the galaxy HS 1236+3937 differs from its field counterparts by a factor of 12 in its clustering properties. Furthermore, the mean nearest neighbour distance for the isolated ELGs is 4.5h^-1Mpc, as compared to the mean $D_{\rm NN}=1\,{\rm h}^{-1}$Mpc for the non-isolated ELGs. Our sample of 139 ELGs contains 15 isolated ELGs (and 2 non-ELGs), which yields a higher fraction of isolated galaxies in the present study, than in our (earlier) photometric study of faint galaxies in voids (Vennik et al. [1996]).

Next we derive mean photometric characteristics of the studied sample and search for possible differences between the mean parameters of the isolated and non-isolated ELGs in Table 2 and in Figs. 5-7. Standard mean photometric parameters are evaluated in the B band. For the 16 ELGs which were observed only in the R band, their R-magnitudes were transformed into B assuming an average colour of B-R = 0.9 (which is valid for our sample, see below). Those transformed data are included in Figs. 5, 6. They do not show any systematic deviations. Figure 5 demonstrates some bias effects in our sample. While all our ELGs are distributed up to $z~\simeq~0.34$, the isolated ELGs are detected within $z~\leq~ 0.033$, only. The available redshift catalogs do not permit us to evaluate true isolation properties of the ELGs at greater distances (see Popescu et al. [1997] for a complete discussion). Hereafter we call the volume within $z~\leq~ 0.033$the local volume. The upper boundary of the absolute magnitude -  log z relation in Fig. 5 corresponds to the limiting apparent magnitude of about 19.5, which yields the formal completeness limit within the local volume of $M_{B} \leq -16.3$. However, the initial sample of ELGs was not selected as a typical continuum magnitude limited sample, rather as an emission-line flux limited sample, as described in Popescu et al. ([1997]).

Table 2 summarizes the mean parameters in B band and their errors of three subsamples of ELGs. The characteristics of the total sample of 124 non-isolated ELGs are given in the first row. The second and third rows of the Table 2 present the mean parameters of the non-isolated and isolated ELGs, within similar distance limits, i.e. within local volume.

Figures 6a-d show different correlations. In these figures the isolated resolved ELGs are presented by filled circles; non-isolated resolved ELGs - by open circles. The ELGs having almost stellar images are shown by crosses, in order to disentangle the possible seeing effects.

The analysis of the tabulated and plotted data leads to the following results:

The ELGs of our sample are distributed within a broad interval of distances (up to $z \simeq 0.34$), of absolute magnitudes ( $-22.5 \leq M_{ B} \leq -12.5$), and of radii ( $0.2 \leq r_{25} \leq 17.9$ kpc), and consist of almost all varieties of actively star-forming galaxies, as classified in SMB89, from the faintest Sargent-Searle dwarfs to the giant SBNs and include even a few AGNs.

Within the local volume where the isolation properties could have been evaluated, both the isolated and non-isolated ELGs show similar mean values of their total luminosities and radii, with those two parameters being distributed within $-19.6 \leq M_{ B} \leq -12.5$ and $0.2 \leq r_{25} \leq$ 6.3 kpc, respectively. According to their photometric and morphological characteristics the ELGs within local volume belong to the broad class of BCGs, with most of them being dwarfs (i.e. BCDs).

Figures 6a and 6c show the distributions of the two different SB characteristics of our sample galaxies. Figure 6a presents the mean effective SB ( $<\mu_{\rm eff,c}^{B}>$), which is calculated as a mean SB within the half-light radius and is corrected for the inclination and for the galactic absorption as taken from the NED. This SB parameter includes both the light of the H II regions and that of the underlying parent galaxy. As could be expected, the quasi-stellar ELGs with almost unresolved images yield the highest SB (crosses in Fig. 6). The mean effective SB of the studied ELGs appears to have a lower bound at $\sim$24 B mag/ $\ifmmode\hbox{\rlap{$\sqcap$ }$\sqcup$ }\else{\unskip\nobreak\hfil \penalty50\h... ...\rlap{$\sqcap$ }$\sqcup$ } \parfillskip=0pt\finalhyphendemerits=0\endgraf}\fi''$. They are distributed over 4 magnitudes in SB. Remarkably, there are intrinsically luminous (giant) ELGs with low SB, too. Probably, those ELGs have smaller H II regions than those found in SBNs and could be classified as Giant Irregulars (GI). The isolated ELGs show marginally fainter mean SB ($22.3~\pm$ 0.7 B mag/ $\ifmmode\hbox{\rlap{$\sqcap$ }$\sqcup$ }\else{\unskip\nobreak\hfil \penalty50\h... ...\rlap{$\sqcap$ }$\sqcup$ } \parfillskip=0pt\finalhyphendemerits=0\endgraf}\fi''$) than the non-isolated (and non-stellar) ELGs ($21.7~\pm$ 1.1 B mag/ $\ifmmode\hbox{\rlap{$\sqcap$ }$\sqcup$ }\else{\unskip\nobreak\hfil \penalty50\h... ...\rlap{$\sqcap$ }$\sqcup$ } \parfillskip=0pt\finalhyphendemerits=0\endgraf}\fi''$), but the isolated ELGs avoid the brightest 2 mag in SB covered by the non-isolated ELGs.

Figure 6c shows the distribution of another SB parameter - the central SB of the exponential disk component $(\mu_{\rm0,c}^{\rm exp})$, with the same corrections applied as for the mean effective SB (see above). This parameter characterizes the underlying parent galaxy. Naturally, the SB profiles of the 33 ELGs, which are either quasi-stellar or show always convex or concave curvature over linear radius, could not be fitted by the exponential disk model. Further 14 ELGs with quasi-stellar images provide a poor fit with exponential model. Those ELGs are marked with crosses in Figs. 6c and 6d. Two features could be noted in the distribution of the disk central SB:
1) A conspicuous empty area in the upper right part of Fig. 6c, with the parametric limits: M _B > -16.5and  $\mu_{\rm0,c}^{\rm exp} < 21.6$. That means, all dwarf ELGs with M _B > -16.5 have dimmer disks than the typical Freeman's disk ( $\mu_{0}^{\rm exp}$ = 21.6). The disks of intrinsically brighter ELGs with M_B < -16.5 are scattered within 4 mag around and below of the Freeman's disk typical SB.
2) Isolated ELGs tend to have lower underlying disk central SB ( $<\mu^{\rm exp}_{0,\rm c}> = 22.8 \pm 0.8$), compared to the non-isolated ELGs of the same M _B (21.7 $\pm$ 1.0).

Figures 6b, d show that ELGs are confined into a narrow strip in log  r - M _B plane. The isolated ELGs tend to have larger linear extent than the non-isolated ELGs of the same luminosity. This tendency becomes more evident for the exponential disk model scale lengths. In Fig. 6d we perform a linear fit to the log disk-scale-length - M _B relation for both the isolated and the non-isolated ELGs within the local volume. The results show that the isolated ELGs possess underlying disks which are, on average, about a factor of 1.8 larger than the disks of the non-isolated ELGs. This difference appears statistically significant at the $1.75\sigma$ level.

Figure 7: The total (B-R) colour indices versus absoluteB-magnitude of non-isolated (circles) and isolated (filled circles) ELGs

Figure 8: The morphological classes of SMB89 versus absoluteB-magnitude

Both the isolated and non-isolated ELGs show blue mean colour of $<B_{\rm T}-R_{\rm T}> = 0.9\pm0.3$ (Fig. 7). We scrutinized the radial colour index (CI) profiles for the occurence of colour gradients in the range of high signal-to-noise ratio, which typically extends up to the 26 B mag/ $\ifmmode\hbox{\rlap{$\sqcap$ }$\sqcup$ }\else{\unskip\nobreak\hfil \penalty50\h... ...\rlap{$\sqcap$ }$\sqcup$ } \parfillskip=0pt\finalhyphendemerits=0\endgraf}\fi''$ isophote, but excluding the very nuclear region within a few arcseconds, where the colours can be affected by changing seeing conditions and/or the occurence of a single luminous (blue) H II region. Extranuclear colour changes are determined both by the stellar populations of the parent galaxy and by further H II knots, when distributed outside of circumnuclear region. We did not attempt to measure the exact values of colour gradients, but classified them (in Table 4, Col. 11a) as follows: definitely positive, i.e. with blue center and getting redder outwards (+), probably positive (+:), definitely negative (-), probably negative (-:) and no gradients (blank). Nearly half of the determined B-R colour index profiles show reliable radial colour changes.

We searched for possible correlations of the colour gradients with different photometric and morphological characteristics, with the following results:
1) The colour gradients are not correlated with the integral colours.
2) The colour gradients are marginally correlated with the absolute B-magnitude in the sense that intrinsically fainter ELGs tend to show a larger fraction of positive gradients (all dwarf ELGs with M_B > -16.5 show positive gradients, if any).
3) The colour gradients are marginally correlated with the SB profile type: pure exponential disks (type 1) and disks with central flattening (type 3) show negative gradients, if any; definitely non-exponential profiles (type 5) show positive gradients, if any; among the multiple component profiles (type 2) the positive colour gradients clearly dominate; unresolved quasi-stellar profiles (type 4) typically do not reveal any reliable colour changes, as could be expected.
4) Among the isolated ELGs there is a slightly higher fraction of positive gradients, which is related to the higher fraction of non-exponential SB profiles among the isolated ELGs (see next paragraph).

The results of SB profile type classification are summarized in Table 3. Of our galaxies, $73\%$ have SB profiles with some disk component. Nearly half of those disks are superposed with extra light emitted by bulges or nuclei (i.e. by bright H II regions). About $25\%$ of our ELGs show unresolved images and their SB profiles could not have been disentangled from stellar light profiles. The isolated ELGs show slightly higher fraction of unresolved images and non-exponential light profiles, compared to non-isolated ELGs. But the difference is not statistically significant, because of the small number of isolated ELGs. Only for 8 of the 15 isolated ELGs an underlying disk component have been discerned in their SB profiles.

The three different morphology classifications presented in Tables 6, 7 will be discussed in detail in a separate paper. A snapshot statistics shows that about 83% of our ELGs belong to the various subtypes of BCGs, which include many subclasses of intrinsically faint ELGs, as discerned by SMB89 in their quantitative classification scheme. The remaining 17% ELGs are classified as giant SBN, with several of them possessing AGN features, or as interacting pairs (IP). Figure 8 shows the distribution of the absolute blue magnitude with the classification of SBM89. To establish this graph, we assigned numerical values to the classification scheme, namely 1 for SS, 2 for DHIIH, 3 for HIIH, 4 for DANS, 5 for SBN, and 6 for IP. Uncertain classification between two classes were assigned the average value, e.g. DHIIH/HIIH got a 2.5. Those galaxies with the rough classification BCD were also assigned to the value 2.5. As can be expected from the defination of the classification system, the absolute magnitude correlates with the classification. But the luminosity distribution within one class is rather broad, up to 4 mag peak-to-peak. This scatter reflects both, the intrinsic distribution of luminosities within each class (Salzer et al. [1989], Fig. 9a) and the uncertainties of our classification process.

TMT97 quantify the multiplicity of the H II regions and describe the shape of the outer isophotes. According to the TMT97-scheme 72% of the studied ELGs possess a single nucleus and 28% show the presence of double or multiple nuclei. The second parameter in the TMT97-scheme, describing the outer structure, yields similar frequencies: 71% of the ELGs are symmetric and regular objects, regardless of the multiplicity of their H II regions, and 29% show disturbed outer isophotes.

LTh85 classified the morphology of the BCDs in more qualitative terms based mostly on the regularity both of the center and of the outskirts of the galaxies. Considering here the appearance of the bright central region we assign 39% of the ELGs into class n (single SFR and regular isophotes), 52% of the ELGs into class i (several SFR and irregular isophotes), The remaining 9% of the ELGs are (non-dwarf) distant spiral galaxies of the types SBN, AGN, and a few interacting pairs.

 

 

Table 3: Comparison of the SB profile type frequencies of ELGs

Profile description (prof. type)	non-isolated		isolated
	Nr	%	Nr	%
Pure exponentials ( 1 d)	27	22	2	13
Multiple components ( 2 b/nd)	46	37	5	33
(disk+bulge/nucleus)
Central light depression ( 3 cnv)	17	14	1	7
Unresolved ( 4 st)	27	22	4	27
Non-exponentials ( 5 cnc)	7	5	3	20