In order to determine the radio spectrum for each source of the sample, we
relied on a number of radio source catalogues, especially
on: 151-MHz 6CII, III, and IV
(Hales et al. 1993; and references therein, 365-MHz Texas (Douglas et al. 1980), 408-MHz B2 and B3 (Colla et al. 1973 and Ficarra et al. 1985, respectively), 1.4-GHz Arecibo observations in the sky strip (Owen et al. 1983) and 83GB (White & Becker 1992), 4.8-GHz GB6 (Gregory et al. 1996) and MGII and MGIV (Langston et al. 1990 and Griffith et al. 1991, respectively). The flux densities were adjusted to the common scale of Baars et al. (1977). Errors of the flux density were taken directly from the individual papers or calculated according to the formulae given there. A fitting the data points with a polynomial function of the type , where , , was performed.
Three cases are considered: (1) a straight line, where c=d=0 (s-spectrum),
(2) a curved fit, where c=0 or d=0 (c- or c+ spectrum), and (3) a composite
fit, where (+c- spectrum if b>0 and d<0, or
-c+ spectrum if b<0 and d>0). The reduced -test was used to decide whether the data could be fitted with either of the above functions. The spectra which could not be adequately fitted to the data (typical for variable sources) are considered to be complex (cpx-spectrum). In some sources the best fitted spectrum consists of two straight lines with a break frequency (sb-spectrum).
The fitted spectrum was then used to determine a spectral index as a derivative (i.e. a slope) of the above type function at the frequency of 1.4 GHz, . Thus, using a functional form of the spectrum, one has two advantages in respect to the popular two-point spectral indices: (i) the slope is not affected by the errors of individual flux density measurements, and (ii) it can be easily transformed into another frequency, e.g. emitted one. In Sect. 6, statistics of the spectra and distributions of are given and discussed.
The radio morphology is determined mostly on the basis of VLA maps published in Papers II, III, IV, and V. Because the strongest sample sources (mostly 3C sources with S1.4>2 Jy), as well as strong compact sources which had been already observed with the VLA by Perley (1982), were not reobserved during the GB/GB2 project, their morphology was specified from other publications. Extended double sources are classified either as edge-brightened (FRII) or edge-darkened (FRI)(classification of Fanaroff & Riley 1974), although morphological type of some of the sources should be determined as an intermediate type.
The other sources have either compact structure dominated by a flat-spectrum unresolved radio core, frequently with weak one-sided or two-sided extended emission detected with the VLA, or compact steep-spectrum (CSS) structure. In some cases a distinction between these two types is not easy. A compact source with well fitted spectrum of "s" or "c-" type is classified as CSS regardless of a frequency of the maximum of its fitted spectrum, while a compact core-dominated source with detected or undetected "1s'' or "2s'' emission, having more complex spectrum of "cpx", "+c-" or "-c+" type is considered here as a separate morphological category. It was shown by Machalski & Inoue (1990) that the GB/GB2 sources with deconvolved angular size arcsec differ significantly in the fringe visibility function, suggesting a variety of structures on the angular scale of about 0.1-0.5 arcsec, and have radio spectra which can be only conventionally classified as flat or steep. The distribution of for the CSS sources (Fig. 1 (click here)) illustrate the problem.
Figure 1: Spectral-index distributions of sources of different morphological type. is a slope of the fitted spectrum at 1.4 GHz (cf. Sect. 3.1). Their mean values and standard deviations are indicated
The morphological types and structural parameters of the sources added to the sample have been taken from available maps of the NVSS and FIRST surveys, or from our own, unpublished, VLA maps. The observed structures were crudely deconvolved into elliptical Gaussian components; relevant data, i.e. their map coordinates, integrated 1.4 GHz flux density, half-intensity major and minor diameters, major-axis position angle (degrees east of north), as well as the source's "largest angular size" LAS and "overall position angle" OPA, are given in the Appendix (Table 6.4 (click here)).
Due to the large time base of the 1.4-GHz observations (Sect. 2.2), from 8 to 13 years, and a sufficient number of independent flux measurements, it was possible to determine the sample sources which vary significantly at this frequency. The 1.4-GHz variability in the GB/GB2 sample was the subject of separate analyses (Ryś & Machalski 1990; Machalski & Magdziarz 1993a). After supplementing older data in the sample with those already available from the NVSS and FIRST surveys, the time base over which the 1.4-GHz variability has been observed extends to 18-24 years. Compactness of these sources and the lack of nearby confusing neighbours (except of 0804+499B, whose flux densities have been carefully corrected for confusion) provide that their flux densities measured with single-dish telescopes and the VLA are comparable.
The variable sources in the sample are listed in Table 2 (click here). For each source, the variability parameters as defined by Machalski & Magdziarz (1993a), i.e. the number of independent observations, n, mean weighted 1.4-GHz flux density and its error (calculated with Eqs. (1)), chi-squared statistics, "apparent fluctuation" of flux density around its mean value, Y(n) (the Y(n) statistics has the physical meaning of the quantity , but is not numerically identical with it), and time lag between the first and last observations, T, are given in consecutive columns of Table 2 (click here). Note that all of these variables but one (1213+350) are of complex (cpx) spectrum with . It is worth to emphasize that most of these sources are known to be variable at frequencies higher than 1.4 GHz; those which were or are systematically monitored for variability are marked in the last column of Table 2 (click here).