Although, within an area of about diameter around the COMPTEL position the East and West sources are the only ones showing significant variations, their location at the 2.3 and 2.6 COMPTEL confidence contour, respectively, and their separation from the IPN triangulation annulus virtually excludes their association with GRB 940301. We, however, point out that the predictions of the Paczyski and Rhoads (1993) model are that GRB 940301 is followed by a radio flare with a peak flux of 1.4 mJy at 92 cm, 250 days after the event, for a source at a distance of 0.5 Gpc. The time scale is not unlike the variations we have seen. Even though the sources are probably not associated, their variability may be representative of the variations that searches for radio counterparts to GRBs will have to be sensitive to.
Remarkable aspects of both sources, but especially the East source, are their large flux density variation and their inverted low-frequency spectra. From the lack of significant variations in the East source during a period of 3 months in the spring and summer of 1994 it does appear, however, that the flux increase in the East source must have started before the time of the GRB. The high 21 cm flux density in November 1993 suggests that an 'outburst' may have occurred somewhere in 1993. Since outbursts are generally delayed at longer wavelengths the start of the 92 cm outburst could have occurred in the beginning of 1994. This would be consistent with the strongly inverted spectral index of the source at that time. The steep spectrum in 1994 is also consistent with most of the emission being due to a 'new' source with little, if any, contamination by an underlying flatter spectrum component, a hypothesis supported by the low 1992 WENSS flux density. The flattening of the spectrum of the East source in January 1996 suggests that the source was approaching its maximum flux density by then.
The West source first increased and then declined by almost the
same factor. Also for this source the data in the spring and summer of
1994 do not reveal significant variation. The West variable source
also had a flat to inverted spectrum in 1994. In January 1996 this was
almost unchanged. The West source therefore showed an outburst of
about 20 mJy at 92 cm sometime between 1992 and 1996. There is no
evidence for variations of this order at 21 cm.
The fact that fairly flat spectrum radio sources were detected at the location of the two variable sources well before the time of GRB 940301 suggests that the variable radio sources are located in galaxies with Active Galactic Nuclei (AGN) activity and may well originate in these AGN. Studies of Low Frequency Variability (LFV) in complete samples of extragalactic radio sources (e.g., 318 MHz: Dennison et al. 1981; 408 MHz: Fanti et al. 1983; Padrielli et al. 1987) have shown that many compact, flat spectrum, extragalactic sources vary in intensity at meter wavelengths. Typically these variations are of the order 10-50 % on timescales of 1 year. Few sources, however, are reported to have varied by factors of up to 2 or 3 (408 MHz: Hunstead 1972; 318 MHz: Condon et al. 1979; 318 MHz: Dennison et al. 1981). Good data only exist for bright (several Jy) sources. It is generally believed that these low-frequency variations in AGN are due to interstellar refractive scintillation. The spectral behaviour of the East variable during its flux rise is, however, more suggestive of an intrinsic variation. Such variations could be due to a violent event (e.g., a shock) in the core/jet of the AGN. Within the context of the AGN-variability interpretation it does remain unusual, however, that the variations are so large at 92 cm while only modest (East) or small (West) variations have been seen at 21 cm (although we recognize that our time coverage at 21 cm is rather scanty).
© European Southern Observatory (ESO) 1997
Online publication: June 30, 1998