1 Introduction

As part of our investigation of steep-spectrum ( $\alpha >0.5$, $S\propto\nu^{-\alpha}$), low-frequency-variable (LFV; $\nu <$ 1 GHz) sources, we have made a series of images with sub-arcsecond resolutions (Mantovani et al. 1992) of a sample of sources. These sources were selected from the papers of Cotton (1976), McAdam (1980), Spangler & Cotton (1981), Fanti et al. (1983) and Altschuler et al. (1984). The aim was to detect the high-brightness components required by the refractive scintillation model for low-frequency variability (Rickett 1986). Most of the sources in the sample showed compact features (deconvolved sizes < 0.15- 0.3 arcsec) both in MERLIN, 408 MHz and VLA, A-array, 5 GHz images (Mantovani et al. 1992). Further observations with VLBI show these features to contain components which are bright and compact enough to explain the variability at low frequency by propagation effects in the interstellar medium; see, for example, 3C 99, Mantovani et al. (1990a). Sources such as these do not generally show any variability at high frequency (Padrielli et al. 1987).

However, there are sources like 0621+400 (3C 159), which are variable at low frequencies and not at frequencies >2GHz, where the compact components are too weak to account for the observed variability. In these cases, it is possible that the variability is caused by instrumental effects. The source 3C 159 has been monitored for about 10 years at 408 MHz. This source has a steep radio spectrum and an extended double radio structure -- a combination which is very unusual for a variable radio source. Browne et al. (1985) have suggested that the variations, which seem to show an annual cycle, may not be intrinsic but could arise from the combined effects of strong source linear polarization and ionospheric Faraday rotation.

Ionospheric Faraday rotation can easily reach 7-8 rad m^-2 (Sakurai & Spangler 1994). With the plausible estimate of 5 rad m^-2as an expectable difference in the ionospheric RM, one finds that the position angle difference at 408 MHz is 2.7 radians; enough to produce the effect being discussed.

MERLIN observations at 408 MHz (Cerchiara et al. 1994) show that 3C 159 is highly polarized ($\sim$10$\%$). The plane of polarization of the source emission could be rotated by changes in ionospheric Faraday rotation relative to the linearly polarized E-W arm of the Northern Cross Bologna telescope used for the monitoring program. A source with a linear polarization >6$\%$ could exhibit apparent variations of roughly the observed size if the ionospheric Faraday rotation changed by $\sim$90$^\circ$ between observations. The 3C 159 observations were made at transit during the day in summer and the night in winter and so they were accompanied by annual changes in the ionospheric electron content.

The purported variability measured in the two extended radio sources 0235-197 and 1203+043 may have originated in a similar manner to that in 3C 159. They were monitored at 408 MHz with a similar instrument, the Molonglo Cross. With a peak-to-peak fractional variability of $\sim$10%, they were classified as "probably variable'' by McAdam (1980).

Consequently, although it was expected that most of the sources belonging to our sample of steep-spectrum, low-frequency-variable sources would be core-dominated sources, it is likely that the sample has been contaminated by lobe-dominated, strongly-linearly-polarized, extended sources.

In order to test if ionospheric Faraday rotation is the cause of the apparent variability of 0235-197 and 1203+043 we have investigated the linear polarizations of these sources at 320MHz with the VLA in the "A'' configuration and with the already available 5GHz VLA C-array data. Both sources were also observed in the X (8.4 GHz) and U (15 GHz) bands. These observations allowed high resolution images of the "hot spot'' regions to be made. The images were combined with available, high-resolution, C band observations to produce rotation measures (RMs) for the outer parts of the sources.

 

 

Table 1: VLA observing dates

Source	Band	Array	Duration	Observing
	MHz		minutes	Date
0235-197	320	A	87	22 Dec. 1992
	4885	A	30	30 May 1986
	4885	C	15	27 May 1989
	8440	A	49	10 Sep. 1990
	14940	A	51	10 Sep. 1990
1203+043	320	A	67	09 Dec. 1992
	4885	A	39	30 May 1986
	8440	A	28	10 Sep. 1990
	14940	A	32	10 Sep. 1990