4 Optical data

4.1 Digitized Sky Survey (DSS)

The optical counterpart of the source is visible on the DSS and the HST observations of NGC 4698. We are certain of the identification from the relative offset between the nucleus and the optical counterpart. From the Automatic Plate Measurement (APM) catalog, the point source has R = 18.9 mag and B = 21.9 mag, indicating a very red color (B-R=3.0 mag). The original observation was made with the 48 inch Schmidt telescope at Palomar Observatory on 17 February 1950.

4.2 Hubble Space Telescope (HST)

We retrieved the calibrated F450W, F606W, and F814W WFPC2 images of NGC 4698 from the HST Archive to perform photometry of the optical counterpart of the source. The observations with the F450W and F814W filters were performed on 10 August 2001, while the observation with F606W was taken on 31 March 2000. In all the filters, the source is located on the WF detector, which has a scale of 0 $.\!\!^{\prime\prime}$1 per pixel. The total exposure time is 460 s ( $2\times 230$ s) in the F450W and F814W filters and 600 s (400 s and 200 s) in the F606W filter. We combined the separate images taken in each filter using the crrej task in IRAF to reject cosmic rays. Aperture photometry was done using the tasks available in the apphot package. The integrated magnitudes, determined using the aperture growth curve method, were transformed to the Vega magnitude system using the zeropoints given in the HST Data Handbook. We find $m_{\rm F450W} = 21.1\pm 0.1$ mag, $m_{\rm F606W} = 19.9\pm 0.1$ mag, and $m_{\rm F814W} = 18.6\pm 0.1$ mag.In the Johnson-Kron-Cousins system these values translate into $B=21.3\pm 0.1$ mag, $R=19.6\pm 0.1$ mag, and $I=18.5\pm 0.1$ mag.

Figure 3: XMM-Newton spectrum of XMMU J124825.9+083020. Data are from MOS1, MOS2, and PN.

4.3 Las Campanas (Magellan)

XMMU J124825.9+083020 was observed for 900 s ( $3\times 300$ s) on 6 April 2002 with the LDSS-2 spectrograph (Allington-Smith et al. 1994) on the 6.5 m Baade telescope at Las Campanas Observatory. We observed through a $1\hbox{$.\!\!^{\prime\prime}$ }25$ longslit in  $0\hbox{$.\!\!^{\prime\prime}$ }9$ seeing, with the slit oriented at the parallactic angle. We used the medium red grism, which gives a dispersion of 5.3 Å pixel^-1. The FWHM spectral resolution was 16 Å, and the spatial resolution was $0\hbox{$.\!\!^{\prime\prime}$ }38$ pixel^-1.

Basic data reduction was performed using the IRAF package. The individual spectroscopic frames were corrected for overscan, flat-fielded using domeflats and summed to obtain the final object frame. The spectral extraction was done by summing the counts within an aperture of 6 pixels ( $2\hbox{$.\!\!^{\prime\prime}$ }3$). Wavelength calibration was achieved using a polynomial fit to the lines in the sky spectrum.

The spectrum in the range 5000-7000 Å is approximately described by the form  $f_{\lambda}\approx \lambda^{4}$.

Figure 4: Normalized VLT spectrum with identifications of absorption lines marked. Telluric absorption lines are marked with the symbol $\oplus$. Shown for comparison is the spectrum of the BL Lac object 0548-322 (Barth et al. 2002), redshifted to z=0.43.

4.4 Very Large Telescope (VLT)

VLT observations were performed with VLT-UT3 (Melipal) plus the FORS1 spectrograph. FORS1 is equipped with a $2048\times2048$ pixel Tektronix CCD, which covers a $6\farcm8\times 6\farcm8$ field in the standard resolution imaging mode with a scale of  $0\hbox{$.\!\!^{\prime\prime}$ }2$ pixel^-1.

Two spectra, both with an exposure time of 570 s, were acquired on 21 April 2002, starting at 4:52:50 UT. The spectra were acquired using Grism #150I plus order separator GG435, which avoids overlapping of spectral orders over a given wavelength; this limited the spectral range to  4500-9000 Å. The slit width was  $1\hbox{$^{\prime\prime}$ }$ for both spectra, and this setup secured a final dispersion of 5.5 Å pixel^-1, corresponding to a FWHM resolution of 13 Å. Before the spectroscopic observation, a one minute R-band acquisition image was obtained on the same night starting at 04:42:52 UT under very good seeing conditions ($\sim$0 $\hbox{$.\!\!^{\prime\prime}$ }$6). The object appeared clearly elliptical and extended, with a bright core and a fuzzy halo. Moreover, an irregular spot located around 1 $\hbox{$^{\prime\prime}$ }$ north of the object core is apparent in the VLT image.

After correction for flat-field and bias, the spectra were background subtracted and optimally extracted (Horne 1986) using IRAF. He-Ne-Ar and Hg-Cd lamps were used for wavelength calibration. The wavelength calibration was checked against the position of night sky lines; the typical error was 0.5 Å. Finally, the two spectra were stacked together in order to increase the signal-to-noise ratio. We encountered problems during observation of the spectrophotometric standard, and hence our spectra are not flux calibrated.

Figure 4 shows the normalized VLT spectrum, with several absorption lines identified (see Laurent-Muehleisen et al. (1998) for a discussion on optical identification of BL Lacs). The detected lines include Ca II H&K $\lambda\lambda 3933, 3968$, the G band  $\lambda 4304$, H$\beta$  $\lambda 4861$, and Fe I  $\lambda 5270$: they all indicate $z \approx 0.43$. The line strengths are heavily diluted by the featureless continuum, but the spectrum shares close similarity to that of the BL Lac object 0548-322 (Barth et al. 2002). We shifted the spectrum of 0548-322, which has a redshift of 0.069, to z = 0.43.