© ESO, 2012

1. Introduction

The optical spectrum of BL Lacertae objects (BL Lacs) is dominated by a featureless synchrotron continuum, which makes the determination of their redshift by spectroscopy very challenging. Redshift determination of quasars and BL Lacs is usually based on emission lines originating in the narrow and broad line regions and/or on the absorption features of the host galaxy. However, in BL Lacs the emission lines are intrinsically weak and like the host galaxy absorption lines they often become so weak relative to the continuum that they fall below the noise level and become unrecognizable.

It is therefore not surprising that many BL Lac samples still have a high fraction of targets without spectroscopically determined redshifts. For instance, in one of the most studied samples, the 1 Jy sample (Stickel et al. 1991), 23 of 37 targets have a spectroscopically secured redshift with the remaining being unknown, lower limits from an absorption line system or uncertain due to a detection of a single line (Rector & Stocke 2001). Large (8 m and up) telescopes are often needed to secure the redshift (e.g. Sbarufatti et al. 2009), but even then there often remains a fraction of targets with unknown redshifts.

Sbarufatti et al. (2005) discussed in detail the method of estimating the redshift of BL Lacs through detection of their host galaxies. They showed that the absolute magnitude distribution of BL Lac host galaxies is roughly Gaussian with an average of and σ = 0.5 mag. This opens the possibility of estimating the redshift using the host galaxy as a “standard candle” if sufficiently deep optical or near-infrared images can be obtained. This method has been applied to several BL Lacs to estimate their redshifts (Nilsson et al. 2008; Meisner & Romani 2010). Given the width of the host galaxy luminosity distribution, the imaging redshifts are not as accurate as spectroscopic redshifts. The accuracy of the method is  ±0.05 in redshift, if high signal-to-noise data are available, but in practice the redshift error is usually larger due to noise in derived host galaxy properties (Sbarufatti et al. 2005).

In this paper we report the results of deep optical imaging of two γ-ray bright BL Lacs, RGB 0136+391 and PKS 0735+178, which were detected by the Fermi-LAT already during first three months of operations (Abdo et al. 2009). For γ-ray emitting BL Lac objects the redshift estimation is particularly important and there are large dedicated spectroscopy programs for determining redshift for Fermi-LAT detected active galactic nuclei. Still 56% of the LAT detected BL Lac objects have no redshift measurement (Ackermann et al. 2011), which makes population studies difficult. RGB 0136+391 is particularly interesting, because of its very hard γ-ray spectrum (spectral index 1.69) in the Fermi-LAT range and the very recent discovery of very-high-energy (VHE) γ-rays from the source (Mazin, in prep.). VHE γ-rays do not travel unaffected in the intergalactic space, but are absorbed by the extragalactic background light (EBL). The EBL leaves an imprint on the observed energy spectrum and if the redshift of the emitting object is known, the observed energy spectrum can be used to set constrains to density and evolution of EBL with redshift (e.g. Stecker et al. 1996; Aharonian et al. 2006; Mazin & Raue 2007; Meyer et al. 2012). Alternatively, if the redshift is not known the observed VHE γ-ray spectrum combined with the GeV range spectra can also be used to derive independent estimation for the redshift of the object (Prandini et al. 2010, 2011).

RGB 0136+391 was first identified as a BL Lac object by Laurent-Muehleisen et al. (1998) who saw a featureless optical spectrum. Following attempts (Wei et al. 1999; Bauer et al. 2000; Caccianiga et al. 2002; Piranomonte et al. 2007) have all failed to detect any spectral features and the redshift remains unknown. Piranomonte et al. (2007) derived a lower limit z > 0.193 based on the absence of host galaxy features in the spectrum. In the only imaging attempt so far, Nilsson et al. (2003) did not detect the host galaxy in a 1200 s R-band exposure.

PKS 0735+178 was classified and a BL Lac object by Carswell et al. (1974), who detected no emission lines, but a strong λ2798 Å MgII absorption system at z = 0.424 providing a lower limit for the redshift. Later spectroscopy (e.g. Stickel et al. 1993; Falomo et al. 1994; Rector & Stocke 2001) has confirmed the absorption system, but no emission features or host galaxy absorption lines have been detected and the redshift remains unknown. Optical imaging (Hutchings et al. 1988; Abraham et al. 1991; Stickel et al. 1993; Falomo & Ulrich 2000; Scarpa et al. 2000; Pursimo et al. 2002) has also failed to detect the host galaxy. Scarpa et al. (2000) give a lower limit R > 20.44 for the host galaxy based on a 440 s HST image through the F702W filter.

Despite of the strong (equivalent width W_obs  ~ 3.6 Å, Rector & Stocke 2001) MgII absorption line, the absorbing galaxy has remained hidden. Hutchings et al. (1988) reported a bridge of R-band emission from PKS 0735+178 towards the galaxy 7 arcsec NW with a SB of 26.0–25.2 mag/sq. arcsec. Later imaging by Stickel et al. (1993) did not see this bridge and the galaxy 7 arcsec NW was measured to have z = 0.645. There are two possible detections of the absorbing galaxy: Pursimo et al. (1999) reported a faint (R = 21.9) elongated structure 3 arcsec NE of PKS 0735+178 and Falomo & Ulrich (2000) saw faint emission 3.5 arcsec E of 0735+178 suggesting it could be the intervening z = 0.424 system, although they couldn’t rule out the possibility that it is just a background target.

Fig. 1 Summed image of RGB 0136+391 with the stars discussed in the text labeled. Field size is 5.7 × 5.8 arcmin. North us up and east is to the left.

The brightness of the BL Lac nucleus relative to the host galaxy often makes it very difficult to determine the host galaxy parameters (magnitude and effective radius) accurately or to detect the host galaxy in the first place. Given that the nucleus is often highly variable, it is advantageous to wait for the moment when it is in a low state. We are currently performing optical monitoring of  ~60 blazars, including RGB 0136+391 and PKS 0735+178 (see¹ and e.g. Aleksić et al. 2011, for results of this monitoring). We are running a parallel program at the Nordic Optical Telescope (NOT) to perform deep I-band imaging of a sample of gamma-ray emitting blazars or potential gamma-ray emitters whenever our monitoring shows that the optical brightness has fallen below a pre-defined threshold. Instead of observing the whole sample at once we wait for the optimal moment. To improve the chances of detecting the host galaxy the threshold level is chosen to represent a very low optical level of each target considering the historical light curve.

In December 2008 the R-band brightness of RGB 0136+391 was observed to be close to the lowest level since the beginning of our monitoring (fall 2002) and in January 2011 the brightness of PKS 0735+178 reached R = 16.7, a very low level considering the long-term historical light curve (Ciprini et al. 2007). Observations were promptly triggered and performed at the NOT and the results are reported in this paper.

Throughout this paper we use the cosmology H₀ = 70 km s^-1 Mpc^-1, Ω_M = 0.3 and Ω_Λ = 0.7.

2. Observations and data reduction

RGB 0136+391 was observed at the Nordic Optical Telescope (NOT) through an I-band filter with almost uniform transmission between 725 and 825 nm on December 7th, 2008. We used the ALFOSC camera with a gain of 0.726 e⁻/ADU and readout noise of 4.2 e⁻. The field of view of the 2048² E2V chip was  ~6.5 × 6.5 arcmin (0.189 arcsec/pixel). Altogether 131 images with individual exposure times between 25–100 s were obtained resulting in a total exposure time of 3830 s. In order to make a fringe correction image the telescope was moved between each exposure. The images were bias-subtracted and flat-fielded with twilight flats and a fringe correction image was constructed from the exposures and subtracted. The images were then registered using 15 unsaturated stars in the field and coadded. The summed image shown in Fig. 1 has a full width at half maximum (FWHM) of 0.61 arcsec. Due to the presence of a quarter Moon in the sky the background brightness was 19.2 mag/sq. arcsec,  ~0.8 mag brighter than average dark sky on La Palma². The field was calibrated using I-band observations made with the 35 cm KVA (Kungliga Vetenskapsakademien) telescope on La Palma in photometric conditions on Jan 11th, 2012. The standard star field PG 0231+055 (Landolt 1992) was observed at the same airmass as the RGB 0136+391 field and the I-band magnitudes of stars 1–6 were derived from these observations. The same stars were then used to determine the zero point of our summed image. The error of the derived zero point is 0.02 mag, excluding color effects (see below).

PKS 0735+178 was observed with the same instrumental setup at the NOT on January 18th, 2011. Altogether 55 images with exposure times of 150–300 s were obtained. The images were processed in a way identical to RGB 0136+391. The final coadded image shown in Fig. 2 has a total exposure time of 10 250 s and a FWHM of 0.60 arcsec. The observations were influenced by almost full (~90%) Moon 21 degrees away from PKS 0735+178. As a result the sky background is relatively high (17.7 mag/sq. arcsec) and there is scattered light present in the raw images resulting in an uneven background. The fringe removal procedure mostly eliminates the scattered light but leaves a small (0.1% peak-to-peak amplitude) undulation of the background at large spatial scales, barely visible in Fig. 2. Therefore, the background of PKS 0735+178 and its point spread function (PSF) stars was subtracted by fitting a two-dimensional linear slope to the sky pixels surrounding the targets, effectively removing any remaining tilt in the background. Calibration of the field was done through star 5, for which Ciprini et al. (2007) give I = 15.12 ± 0.06. This star was unsaturated in several exposures, which were used to derive I-band magnitudes for stars 1−3, which were then used to determine the zero point of the coadded image.

Since the transmission curve of the filter used here closely matches the I-band transmission (Bessell 1990), the color effects are expected to be small. According to the NOT zero point monitoring pages³ the (V − I) color term is –0.03. Two of the stars used for calibration, star 4 in RGB 0136+391 and star 5 in PKS 0735+178, have V − I ≈ 0.8 (Nilsson et al. 2007; Ciprini et al. 2007, present work). Assuming the rest of the stars used for calibration have similar colors and that the host galaxy V − I = 2.6 (elliptical galaxy color at z = 0.5, Fukugita et al. 1995) we derive the error of host galaxy magnitudes due to filter mismatch to be  ≈ 0.05 mag.

Fig. 2 Summed image of PKS 0735+178 with the stars discussed in the text labeled. Field size is 5.2 × 5.4 arcmin. North us up and east is to the left.

3. Analysis

We searched for evidence of the host galaxy by fitting two-dimensional models to the observed image. The detections are further tested by simulations, which are also used to derive error bars for the derived parameters.

The fitting procedure is the same as employed in our previous studies (e.g. Nilsson et al. 1999, 2008). In short, the two-dimensional model consists of two components, the unresolved core and a host galaxy whose surface brightness profile follows the de Vaucouleurs law (β = 1/n = 0.25, where n is the profile index from Sersic 1968). The model is described by five parameters: location x-y, assumed to be the same for the core and the host galaxy, core magnitude I_core, host galaxy magnitude I_host and host galaxy effective (half light) radius r_eff. Since the host galaxies were known from previous imaging to be very faint, the ellipticity and position angle were kept at constant value =0. We used a Levenberg-Marquardt loop to find the set of parameters which minimized the χ² between the data and the model. The models were convolved with an empirical PSF derived from stars in the vicinity of the target. In the case of RGB 0136+391 we used stars 1–4 and in the case of PKS 0735+178 stars 1–3. The fitting proceeded in two phases. First we fitted a model consisting of an unresolved component only, i.e. the model had three free parameters (x, y and I_core). After this the position was fixed and the full core + host galaxy model was fit with I_core, I_host and r_eff as free parameters.

If the host galaxy is below detection limit these fits usually converge towards r_eff → 0 or r_eff → ∞. In case we found r_eff > 0 we tested the significance of the detection by performing fits to simulated data. We created a set of  ~100 simulated images of the target using the fitted core and host galaxy values. The β value for the surface brightness profile of the simulated galaxy was drawn from a Gaussian distribution with average equal to 0.25 and σ = 0.025. The fitting, however, was made with fixed β = 0.25. Each model was convolved with a PSF consisting of a Moffat β = 2.5 profile with a FWHM equal to the observed one, an ellipticity randomly drawn from an uniform distribution between 0.0 and 0.06 and a position angle drawn from an uniform distribution between 0.0 and 180 degrees. To these images we added Gaussian noise representing photon and readout noise, scaled to the same level as in the observed data and a random offset drawn from a Gaussian distribution with σ = σ_sky/30, where σ_sky is the sky rms scatter, representing the error made in sky level estimation.

For each simulation we also constructed a second PSF, which slightly deviated from the PSF used in convolving the model. The ellipticity of the second PSF was drawn from an uniform distribution with mean equal to the ellipticity of the first PSF and maximum deviation of 0.02. Similarly, the position angle deviated maximum 10 degrees from the position angle of the first PSF. To this second PSF we also added random photon and readout noise plus an offset the simulate the sky determination error. The second PSF was then used in the simulated fits as the assumed PSF. Using a different PSF for the model creation and for the fitting enabled us to roughly simulate the PSF variation seen in the images and to produce residuals which qualitatively resemble the residuals seen in actual images, although the latter tend to be a bit more complex than what these simple simulations can produce. The limits to the distributions above were selected to match the residuals quantitatively, i.e. the peak-to-peak residuals in the simulations were on average equal to the real residuals. Similarly, the average χ² in the simulations was close to the χ² of the fit to the real data.

After the simulations we computed the standard deviation of the host galaxy magnitudes σ_host and the range of effective radii which includes 67% of the simulated values. We didn’t use standard deviation for the effective radius because the distribution was asymmetric with a longer tail towards high values. To consider a host galaxy detected we required σ_host < 0.3 mag and that the median effective radius of the simulations is >2σ_reff, where σ_reff is the lower 67% error bar. The more relaxed criterion for r_eff was used because r_eff is generally more difficult to constrain and requiring r_eff > 3σ_reff would exclude host galaxies whose surface brightness profile clearly exceeds any reasonable PSF variability.

4. Results

4.1. RGB 0136+391

Figure 3 shows the surface brightness (SB) profile of RGB 0136+391 together with the profiles of four stars in the FOV. The profiles of the stars have been scaled to the same total flux as RGB 0136+391. The SB profile of RGB 0136+391 is totally consistent with a point source, with no excess in the outer part of the profile to indicate the presence of a host galaxy. Examination of the two-dimensional residuals (Fig. 3, insert) yields the same conclusion: no obvious signs of the host galaxy above PSF residuals can be seen. The model fits confirm this conclusion: the final values returned by the fits are I_core = 15.20 ± 0.01, I_host = 19.46 ± 0.32 and arcsec, where the error bars come from the simulations described above. This target thus remains unresolved.

Fig. 3 I-band surface brightness profile of RGB 0136+391 (thick solid line) and four stars in the field. The stars are labeled in Fig. 1. Insert: residuals after subtracting the PSF from RGB 0136+391. Field size is 14  ×  14 arcsec. The error bars include contribution from the empirical scatter in each bin and the error in sky background determination.

We then derived a lower limit for the redshift of RGB 0136+391 using the following procedure: if the distribution of BL Lac host galaxy magnitudes is as shown by Sbarufatti et al. (2005), then 95% of BL Lac host galaxies are brighter than M_R = −22.0. We thus created 100 simulated BL Lac objects at z = 0.25, 0.30, ..., 0.45 with core magnitude I_core = 15.20 and I_host and r_eff corresponding to and R_eff = 10 kpc. We first computed the absolute I-band magnitude M_I using R − I = 0.70 (Fukugita et al. 1995) and the apparent I-band magnitude m_I from (1)where DM is the distance modulus, K_I is the I-band K-correction from Fukugita et al. (1995), A_I = 0.15 is the galactic extinction (Schlegel et al. 1998) and E(z) = 0.84 ∗ z is the evolution correction, computed using the PEGASE code (Fioc & Rocca-Volmerange 1997) by assuming galaxy formation 11 Gyr ago and passive evolution thereafter.

At each redshift we then fitted the 100 model images using the same procedure as in the error simulations and computed the standard deviation of I_host and r_eff. The result of these simulations was that at z = 0.45 (m_I = 19.52) the host galaxy became unresolved using our criteria. Thus at z ≤ 0.40 we would have detected the host galaxy with  >95% probability and thus assign a conservative lower limit z > 0.40 for RGB 0136+391.

4.2. PKS 0735+178

Figure 4 shows the SB profile of PKS 0735+178 together with stars 1–4. The stellar profiles have been again scaled to the same total flux as PKS 0735+178. Contrary to RGB 0136+391, there is a clear excess over the stellar profiles. Model fits yield I_core = 16.19 ± 0.01, I_host = 18.64 ± 0.07 and r_eff = 1.8 ± 0.4 arcsec meeting our criteria for successful detection. Adding the error from zero point determination and filter mismatch the error of I_host becomes 0.11 mag. Given the clear excess and the fit results we conclude that we have detected the host galaxy or a foreground galaxy (see Sect. 5).

Fig. 4 Surface brightness decomposition of PKS 0735+178. In addition to PKS 0735+178 and model surface brightnesses, the profiles of stars 1–4 are shown. Insert: the residuals after subtracting the core + host galaxy model. Field size is 14.2  ×  14.2 arcsec.

The error bars from the simulations do not take into account possible systematic errors, which are difficult to predict and/or quantify. We have tested the sensitivity of our result with respect to our particular choice of software by performing the fits with the GALFIT software (Peng et al. 2002). We used the same image, PSF and sky background as before with the only difference being the software. The result from this fit was I_host = 18.57 and r_eff = 2.0 arcsec, i.e. the host galaxy magnitude differs by 7% and the effective radius by 11% from our result. The two programs thus provide very consistent results and we do not expect the systematic errors due to software choice to dominate over statistical noise.

Assuming that the detected galaxy is the host galaxy we estimated the redshift of PKS 0735+178 using the results of Sbarufatti et al. (2005). Using Eq. (1) with A_I = 0.068 (Schlegel et al. 1998) and M_I = −22.8 we iteratively determined the the redshift consistent with the observed I-band magnitude to be z = 0.45 ± 0.06, where the error bar includes the contribution from host magnitude error and the uncertainty of the method itself.

5. Discussion of PKS 0735+178

Scarpa et al. (2000) derived an upper limit m_R > 20.44 for the host galaxy, which together with our result implies R − I > 1.7. This is redder than the expected R − I = 1.2 for ellipticals at z = 0.5 (Fukugita et al. 1995), possibly an indication of systematic differences between the two studies, or of dust reddening in the host galaxy. If significant dust absorption is present in PKS 0735+178, our redshift estimate is too high.

Fig. 5 47.4  ×  47.4 arcsec field around PKS 0735+178 after subtracting the scaled PSF from PKS 0735+178 and star 4. The image has been smoothed with a Gaussian kernel with σ = 1.0 pix. Galaxies G1 and G2 and the rectangular test feature R are discussed in the text. H is the host galaxy of PKS 0735+178. The dotted ellipse and circle mark the location of the features discussed by Pursimo et al. (1999) and Falomo & Ulrich (2000), respectively.

Using the I-band data obtained here and previous R-band data we now discuss the identity of the MgII absorber. Galaxy G1 in Fig. 5 can be directly ruled out due to the redshift z = 0.645 measured by Stickel et al. (1993). Pursimo et al. (1999) and Falomo & Ulrich (2000) detected weak R-band emission 30 NE and 35 E of PKS 0735+178, respectively, and discussed the possibility of these features being the MgII absorber. Neither of these emission features are visible in our PSF-subtracted I-band image (Fig. 5). Since our images were obtained in bright moonlight and thus not optimal for detecting faint diffuse emission, we tested our sensitivity with the following procedure.

Firstly, since Falomo & Ulrich (2000) do not give a magnitude for the feature they detected, we cannot estimate if it was detectable in our image. We therefore concentrate on the elongated feature reported 30 NE of PKS 0735+178 by Pursimo et al. (1999). To roughly simulate this feature we created a rectangular test object marked with R in Fig. 5. Note that we did not create this feature at the correct location to better compare with the actual image. The feature has a uniform I-band surface brightness of 24.3 mag/sq. arcsec and its dimensions and integrated magnitude 21.9 correspond to those of the elongated featured reported by Pursimo et al. (1999). The test feature is just visible over the background noise and thus we should have detected the Pursimo et al. (1999) feature in our image if its R − I color  >0. Since the range of R − I colors of E to Im galaxies at z = 0.424 is expected to be form 1.1 to 0.4, respectively (Fukugita et al. 1995), we conclude that we should have detected the Pursimo et al. (1999) feature, if its colors are within the range of “normal” galaxies. Given the non-detection in our I-band image the Pursimo et al. (1999) feature is unlike to be the absorber. It could be a line emitting cloud whose emission wavelength falls into the R-band but outside the I-band. Of the lines bluewards of the R-band, λ5007 [OIII] is the most likely candidate. This line would place the cloud at z = 0.13–0.44 when 50% transmission limits of the R-band filter are considered.

Since we do no detect the feature discussed by Pursimo et al. (1999) and the detection probability of the Falomo & Ulrich (2000) feature cannot be estimated we turn our attention to galaxy G2. This galaxy lies at projected distance of 82 (46 kpc) from PKS 0735+178. Given the rest frame equivalent width of the MgII line in PKS 0735+178, W₀ = 3.6 Å/(1 + z_abs) = 2.5 Å, we ask if this galaxy could be the absorber. For comparison we use Zibetti et al. (2007), who studied 2844 MgII absorbers with W₀ > 0.8 Å and 0.37 < z < 1 using the SDSS DR4. We first note that W₀ = 2.5 Å implies very strong absorption: 93% of the systems studied by Zibetti et al. (2007) have W₀ < 2.5 Å. If we concentrate on the relevant section of the parameter space in Zibetti et al. (2007), 0.37 < z < 0.55 and W₀ > 1.58 Å, we note that MgII absorbers can be found up to projected distances of  ~100 kpc from the background source. In this section of the parameter space the probability of finding an MgII absorber at projected distance of 46 kpc is  ~25% (see Fig. 9 in Zibetti et al. 2007). Given that the probability of finding an Mg II absorber at a given distance decreases with W₀ and the W₀ of the MgII line in PKS 0735+178 is at the high end of the distribution, the probability of G2 being the absorber is probably considerably lower. The I-band magnitude of G2 is 20.4 ± 0.1 which gives R − I = 0.5 when combined with the R-band magnitude in Pursimo et al. (1999). This color is consistent with a late type (Scd) galaxy at z = 0.424. The absolute magnitude of G2 at z = 0.424 would be . In the light of these results G2 cannot be completely ruled out as the absorbing galaxy, although finding such a system is relatively unlikely.

As the next option we consider the possibility that the detected host galaxy is actually a foreground object and not the real host galaxy. If this is the case, then the impact parameter must be very close to zero since the detected galaxy appears to be well centered on the BL Lac nucleus. To check for possible offsets between the core and the host galaxy we performed the host galaxy fit with 7 free parameters, i.e. in addition to I_core, I_host and r_eff also the x-y positions of the core and the host galaxy were allowed to change freely. The core-host galaxy offset was 0.67 pixels (0.13 arcsec) in this case, indicating that the core is well centered on the host galaxy. Such a system is capable of producing gravitational lensing phenomena, but Rector & Stocke (2003) did not detect any signs of gravitational macrolensing (Einstein rings or multiple images) in their 8.46 and 14.94 GHz maps with  ~0.2 arcsec resolution. Furthermore, the result of Zibetti et al. (2007) indicate that strong absorbers are associated with galaxies with intense star forming, which should be visible in the optical spectrum as narrow emission lines. These findings do not disprove the hypothesis that the detected galaxy is an intervening system with a small impact parameter, but they make it less attractive with respect to the alternative, that we have detected the actual host galaxy.

Lastly we note that the derived redshift 0.45 ± 0.06 is within errors equal to the absorption redshift 0.424. It could thus be possible that the absorption occurs in the host galaxy of PKS 0735+178 itself. This would imply the PKS 0735+178 host galaxy to be a gas-rich system since strong Mg II absorption is often associated with damped Lyman-alpha systems (DLAs) with high HI gas content (Péroux et al. 2004). Further observations are clearly needed in order to determine which, if any, of the above scenarios is the correct one. A spectrum of G2 is needed in order to see if its redshift is compatible with being the absorbing galaxy. An ultraviolet spectrum of PKS 0735+178 detecting the Lyman-alpha line would also shed new light on the nature of the absorber by measuring of the HI column density.

6. Conclusions

We have presented deep I-band imaging of two BL Lacertae objects, RGB 0136+391 and PKS 0735+178, obtained during a deep minimum in the optical light curve. Despite of a deep exposure obtained under excellent seeing conditions (06) the host galaxy of RGB 0136+391 remains unresolved. We derive z > 0.40 for RGB 0136+391 using simulated host galaxies at different redshifts with M_R = −22.0 and R_eff = 10 kpc. This makes RGB 0136+391 one of the most distant VHE γ-ray sources detected up to date.

The surface brightness profile of PKS 0735+178 reveals an excess over a point source, which, if fitted by a de Vaucouleurs profile, corresponds to a galaxy with I = 18.64 ± 0.11 and r_eff = 1.8 ± 0.4 arcsec. Under the assumption that BL Lac host galaxies can be used as standard candles (Sbarufatti et al. 2005) we derive z = 0.45 ± 0.06 for PKS 0735+178.

We also discuss the identity of the galaxy responsible for the strong MgII absorption line in PKS 0735+178 at z_abs = 0.424 using previous R-band imaging and the I-band data in this paper. We do not detect the two faint objects suggested earlier to be responsible for the absorption or any other obvious candidates in the vicinity of PKS 0735+178. We identify a galaxy 82 of PKS 0735+178 (46 kpc projected distance) as a candidate for the absorber, although the probability of strong absorption (W₀ = 2.5 Å) at this projected distance is very low. Further observations are needed in order to reveal the identity of the absorber towards PKS 0735+178.

Acknowledgments

The data presented here were obtained with ALFOSC, which is provided by the Instituto de Astrofisica de Andalucia (IAA) under a joint agreement with the University of Copenhagen and NOTSA. This research has made use of NASA’s Astrophysics Data System Bibliographic Services. This research has made use of NASA’s Astrophysics Data System Bibliographic Services. This research has made use of the NASA/IPAC Extragalactic Database (NED) which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration.

References

All Figures

	Fig. 1 Summed image of RGB 0136+391 with the stars discussed in the text labeled. Field size is 5.7 × 5.8 arcmin. North us up and east is to the left.
In the text

	Fig. 2 Summed image of PKS 0735+178 with the stars discussed in the text labeled. Field size is 5.2 × 5.4 arcmin. North us up and east is to the left.
In the text

	Fig. 3 I-band surface brightness profile of RGB 0136+391 (thick solid line) and four stars in the field. The stars are labeled in Fig. 1. Insert: residuals after subtracting the PSF from RGB 0136+391. Field size is 14  ×  14 arcsec. The error bars include contribution from the empirical scatter in each bin and the error in sky background determination.
In the text

	Fig. 4 Surface brightness decomposition of PKS 0735+178. In addition to PKS 0735+178 and model surface brightnesses, the profiles of stars 1–4 are shown. Insert: the residuals after subtracting the core + host galaxy model. Field size is 14.2  ×  14.2 arcsec.
In the text

Fig. 5 47.4  ×  47.4 arcsec field around PKS 0735+178 after subtracting the scaled PSF from PKS 0735+178 and star 4. The image has been smoothed with a Gaussian kernel with σ = 1.0 pix. Galaxies G1 and G2 and the rectangular test feature R are discussed in the text. H is the host galaxy of PKS 0735+178. The dotted ellipse and circle mark the location of the features discussed by Pursimo et al. (1999) and Falomo & Ulrich (2000), respectively.

In the text