Next Contents Previous

3.4.5. X-ray variability

The X-ray flux of 3C 273 varies considerably. The first report of this variability is by [Marshall et al. 1981] who detected an 80% flux increase in a 40000s Ariel V observation. This is the only report of very strong and short flux variation in 3C 273. Since Ariel V was not an imaging instrument, there is a possibility of source confusion. This result is therefore in strong need of confirmation by an imaging instrument.

The flux of the soft excess has been found [Courvoisier et al. 1987] to be variable in time and to vary independently of the medium energy component. [Courvoisier et al. 1990] reported a variability of 4% in about 17 hours in the medium energy component observed by EXOSAT. [Leach McHardy & Papadakis 1995] analysed 14 ROSAT observations. They confirm that the soft and medium energy X-ray components vary independently and reported their fastest variations to be by about 20% in 2 days in the soft energy band (0.1-2.4 keV) observed by ROSAT.

The 2-10keV flux observed by EXOSAT and GINGA indicate variations from 0.60 . 10-10 to 1.68 . 10-10ergs cm-2 s-1 for the medium energy component. This was obtained with 13 observations spanning about 5 years. The resulting light curve is hopelessly undersampled, preventing any description of the flux variations.

This data set also provided evidence that the spectral slope in the same energy domain varies by small but significant amounts. The spectral slope is not correlated with the flux but is anti-correlated with the logarithm of the 2-10keV count rate divided by the UV photon rate [Walter & Courvoisier 1992]. This result is interpreted in this paper as being the signature that the medium energy X-ray component is due to a thermal Comptonisation process of the UV photons by an electron population of about 1MeV covering a few percent of the UV source with an optical depth of 10-20%. Work in progress shows that when using the BATSE data as displayed in Fig. 2 one finds a weaker anti-correlation between the spectral slope. This work should, however,take into account the delay between the UV and the X-ray light curves. This delay (discussed below) indicates that the UV flux Comptonised in the hot regions may not be that observed simultaneously, but rather the UV flux observed about 2 years earlier. This may not have had much effect in the early phases of the monitoring because the UV flux was relatively quiet compared with later epochs.

Next Contents Previous