2 Sample selection

2.1 General remarks

The selection of two separate subsamples of highly-inclined disk galaxies - consisting of a large number of interacting/merging candidates in the mass range $M_{\rm sat}/M_{\rm disk} \approx 0.1$, and of relatively isolated disk galaxies - is a crucial point for the comparison of disk parameters. Since presumably not all of the merging processes of this order of magnitude - compared to large merging events - are able to change the structure of affected galaxy disks completely, two facts must be considered in order to classify both subsamples:

First, only merging events in a more progressed phase will have any appreciable effect on the disk structure. We therefore did not consider candidates in an early stage of interaction. Instead, the sample contains some highly-inclined disk galaxies that show no separate satellite merging with the disk component, but strong evidence for accretion in the recent past (indicated by both a disturbed disk structure and characteristic effects such as warping and tidal tails). Most of these objects are located in galaxy groups with a few members, e.g. NGC 3628, NGC 4634, or NGC 4762. Due to their perturbed structure a morphological classification is difficult.

Second, a control sample of non-interacting edge-on disk galaxies must include all morphological types in the range $0 \le T \le 9$ (Table 1). The galaxies should be isolated without indication of interaction and accretion. In particular, it is important to include the latest type spirals like Sd - so called "superthin'' galaxies - since they belong to a class of disk galaxies having the smallest known axis ratios (between 1:9 and 1:20). These galaxies are characterized by velocity curves of modest gradient (Goad & Roberts [1981]; Griv & Peter [1996a]-[c]), indicating that the "superthins'' represent kinematically almost unheated galaxy disks.

Beyond these "limitations'' and the specific criteria described in the next section it seems difficult to give a generalized definition of the selection criteria. As a consequence, it is unavoidable that the sample of non-interacting galaxies is slightly "polluted'' by some galaxies that were affected by interactions/minor mergers during their past but show no indication of these effects today. This effect would, however, only influence the non-interacting galaxies and lead to an underestimation of the actual differences between both samples.

2.2 The sample of interacting/merging galaxies

In this study, the notation "interacting/merging'' refers closely to the classification scheme introduced by Arp & Madore ([1987]) and will therefore be used as a synonym for all galaxies which fulfil the following criteria:
Galaxies with interacting companions, interacting doubles, galaxies with peculiar disks, galaxies with tails, loops of material or debris, irregular or disturbed galaxies, chains and groups of galaxies. A more detailed description of these classifications is given in Arp & Madore ([1987]).

With these limitations in mind, we selected interacting/merging candidates with an inclination $i \ge 85 \hbox{$^\circ$ }$ (derived from a first visual inspection). During this pre-selection we did not impose any morphological restrictions except that their type should be not much earlier than $T \approx 0$. The candidates were chosen from optical prints in "A Catalogue of Southern Peculiar Galaxies and Associations'' (Arp & Madore [1987]), "Atlas of Peculiar Galaxies'' (Arp [1966]), and the NASA "Atlas of Galaxies'' (Sandage & Bedke [1988]). We also selected some systems from the catalogue "Satellites of Spiral Galaxies'' (Zaritsky et al. [1993], [1997]).

Another selection criterion for the minor mergers and those galaxies in the subsample with very close companions was the mass ratio between the main bodies. For all the candidates which could be separated into two individuals (i.e. for $\approx 20$ galaxies of the interacting/merging sample) this ratio was checked by an estimation of their total fluxes within a certain aperture or ellipse that contains all intensities down to the sky brightness. The resulting mean ratio is $M_{\rm sat}/M_{\rm disk} \approx 0.08~ \pm~ 0.035$. Typical examples (with errors of $\approx 0.005$) are NGC 1531/32 ( $\approx 0.05$), NGC 128 ( $\approx 0.045$) or NGC 1888 ( $\approx 0.07$). For interacting galaxies located a less dense group or for those with a remote companion this mass ratio can be larger.

2.3 The sample of non-interacting galaxies

The selection criteria for disk inclination and morphological types were the same as described for the interacting/merging galaxies. Our principal sources for the subsample of highly-inclined, non-interacting galaxies were the ESO-Uppsala catalogue (Lauberts & Valentijn [1989]), the RC3- (de Vaucouleurs et al. [1991]) and UGC (Nilson [1973]) catalogues, the "Carnegie Atlas of Galaxies'' (Sandage & Bedke [1994]), and "The Hubble Atlas of Galaxies'' (Sandage [1961]). To check their isolation, larger fields $(\approx 10' - 15'$, depending on the distance of individual objects) were inspected visually using the Digitized Sky Survey.

In order to benefit from a better spatial resolution of some closer objects there was only a lower limit in apparent angular disk diameter of $\ge 2 \hbox{$^\prime$ }$ for both galaxy samples. Finally, the sample of non-interacting galaxies was filled up by 11 edge-on galaxies of the Barteldrees & Dettmar ([1994]) data set.

2.4 Distribution of morphological types

Several studies of the properties of edge-on galaxies argued that some of the disk parameters of spiral galaxies, e.g. the ratio of disk scale length to scale height, might be correlated with the morphological type (de Grijs & van der Kruit [1996]; de Grijs [1997]; Schwarzkopf & Dettmar [1998]). Therefore, it was also important to ensure that both subsamples investigated in this study are not affected by selection effects. This applies, in particular, to the distribution of morphologigal types and to the redshifts/distances of selected objects. Due to their relation to the absolute properties of galactic disks the latter will be discussed in detail in Paper II.

The distribution of morphological types (according to NED) is shown in Fig. 1a for the complete galaxy sample, and in Fig. 1b for both subsamples of non-interacting and interacting/merging galaxies, respectively. The statistical test of Kolmogorov & Smirnov (Darling [1957]; Sachs [1992]) - hereafter KS - quantifies the similarity of both subsamples with a result of 0.04. That is significantly lower than the value necessary for the 20%-limit (0.2), which is the strongest of the KS-criteria. Considering also the unavoidable errors ( $\Delta T \approx \pm 1$) introduced by an automated type classification (Corwin et al. [1985]; Lauberts & Valentijn [1989]) both distributions are statistically indistinguishable. The gross of galaxies covers the range between $3 \le T \le 7$, with a strong peak at T=3. This peak can also be observed in the frequency distribution of galaxy types in the available catalogues. The effect is caused by the selection criteria of the classification programs used in the catalogues (Sect. 6 in Lauberts & Valentijn 1989).