## 3. ResultsThe initial spectrum of III Zw 2 from 1998 May was presented in Falcke et al. (1999). It was highly inverted at centimeter wavelengths with a spectral index of between 4.8 and 10.5 GHz. The entire outburst spectrum from 1.4 GHz to 666 GHz could basically be fitted by only two homogeneous, synchrotron components which are optically thin at high frequencies and become self-absorbed below 43 GHz. This spectral turnover frequency stayed constant until 1998 November and hence we expected no strong structural change during this time (Fig. 1).
Our first three VLBA epochs were made while the spectral peak was constant at 43 GHz. The core itself is resolved at 43 GHz at all epochs. To represent the extent of the source, the non-zero closure phases at long baselines for the 43 GHz data were fit by two point-like components. A rough estimate of the formal statistical errors of the component separation was obtained by dividing the original beam size by the post-modelfit signal-to-noise ratio (e.g. Fomalont 1999, Sect. 2.3). The errors were of the order of for the first four epochs and for the fifth epoch. Additional to this very small statistical error, there should be a larger systematic error which is difficult to quantify. To minimize this systematic error we used very similar reduction procedures for each epoch. In accordance to the VLA data, the maps of the first three epochs show no structural change and the separation of the fitted components stayed constant within the statistical errors at 76as corresponding to 0.11 pc (see Table 1) for an angular size distance of 307.4 Mpc ( km/sec/Mpc, as used in this paper).
After 1998 November, the VLA observations showed a dramatic change in the spectrum. The spectral peak dropped quickly to 15 GHz within a few months (Fig. 1). Since the peak in this source is caused by synchrotron self-absorption (Falcke et al. 1999), the fast change in peak frequency implied a similarly strong morphological change, i.e. a rapid expansion. To roughly estimate the expected expansion speed, we applied a simple equipartition jet model with a dependence (e.g. Blandford & Königl 1979; Falcke & Biermann 1995). For an initial source size pc and a self-absorption frequency GHz in 1998 November, we calculate a source size of 0.32 pc for a self-absorption frequency of 15 GHz in 1999 March. Thus we predicted an apparent expansion speed of 1.9 c after the correction for cosmological time dilatation and asked for further VLBA-observations. Indeed the fifth epoch of VLBA observations showed a dramatic structural change compared to the earlier epochs (Fig. 2). A model of at least three point-like components is required now. It was not possible to fit the closure phases with a two-component model as in the earlier epochs or to get rid of this third component during self-calibration. The separation of the outer components for all five epochs is plotted in Fig. 3. While the separation at the first three epochs is consistent with an expansion speed of , the fifth epoch shows a rapid expansion. The apparent expansion speed between the outer components in the 4th and 5th epoch is .
This value is only a lower limit and increases to 2.66 c if one considers the time range from December until March during which most of the spectral evolution occurred. Applying the standard equation for superluminal motion, (e.g. Krolik 1999), to a value of constrains the maximal angle between the jet and the line of sight to , since . © European Southern Observatory (ESO) 2000 Online publication: June 5, 2000 |